HgCapital Trust (LSE: HGT) is Britain's second-largest listed tech company. So, with the UK desperately short of exposure to the technology sector, it might be thought that managers would be keen to invest in the £2 billion fund. But far from it. Very few UK specialist funds have any exposure at all and the shares trade on a 26% discount to net asset value (NAV).

The lack of exposure is because HgCapital Trust is an investment trust. Professional managers don't like investing in a fund managed by someone else, even if that is a private-equity fund holding businesses they couldn't invest in themselves. The discount is more recent – in late 2024, it was on a small premium – and reflects wider fears about the impact of AI on software companies.

HgCapital Trust’s NAVis vulnerable

HgCapital Trust's NAV is an independently calculated valuation derived from those of comparable listed companies. If, as is usual, there is no directly comparable company, the net is thrown wider. This means that the NAV is vulnerable to sentiment towards companies that compete with its holdings, and also towards firms that may superficially appear similar, but in reality focus on a very different area of the market. So when the software shares in general fell sharply due to fears about AI, investors could see how that would affect the NAV.

HgCapital Trust has spent the last year trying to show that even if AI is a threat to the broader sector, it is enhancing rather than threatening the outlook for its investments. This argument is supported by the numbers. HgCapital Trust's interim report showed revenue and net cash-flow growth of 16% and 19% respectively for the portfolio over 12 months, of which 11% and 17% was organic. However, the enterprise value to earnings before interest, tax, depreciation and amortisation (EV/Ebitda) was reduced to 22.9 from 24 times, so the NAV was 4.9% lower over the period.

This valuation had been over 26 times at the end of 2024. Arguably that – and the sector average it was based on – was too high with or without the long-term threat of AI. However, at HgCapital Trust's current share price of 393p, the look-through valuation is 17 times EV/Ebitda, falling to 14 times on the current year's figures.

Visma's postponed listing is bad news for HgCapital Trust's portfolio

The bad news is that the listing of Visma – its largest holding at 12% of the portfolio – has been postponed. The good news is a steady stream of realisations at significant premiums and a series of interesting new investments. In 2026, HgCapital Trust has made three full exits, one partial exit and received realisation proceeds from four re-financings. The full exits were at an average uplift to book value of 31%, slightly above the ten-year average, so there is no sign of inflated valuations. Three new and two increased investments were announced.

HgCapital Trust has investment commitments amounting to a hefty 85% of NAV, points out Iain Scouller of brokers Canaccord, although it has the ability to opt out of any if it so chooses. It had unused debt facilities of £300 million at mid-year, but there is also debt in the investee companies and in the Hg-managed funds through which HgCapital Trust invests. So while the balance sheet looks sturdy, there is debt below the surface. This is not necessarily worrying: HGT should aim to take advantage of opportunities when valuations are depressed.

As for risks around AI, HgCapital Trust says its portfolio represents one of the largest software environments actively deploying AI at scale, driving transformation via AI. What's more, “investors have begun to distinguish more clearly between businesses positioned to benefit from AI and those more exposed to disruption from it”, says Jim Strang, HgCapital Trust's chair.

Far from being a victim of AI, HgCapital Trust sees itself as a beneficiary, making its shares look highly attractive. Hence the managers' commitment to increase their 6% shareholding to 15%. Their lead should be followed.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Despair in Asia over the Democrats' sweep of the US Congress turned into cautious bullishness as Wall Street woke up on Wednesday. Investors in East Asia, a region at the centre of the AI chip supply chain, interpreted the vote as a clear rejection of AI. Japan's Nikkei index touched a five-month low, with shares in SoftBank falling nearly 9% in intra-day trading, say Peter Landers and Gunjan Benerji in The Wall Street Journal. South Korean and Taiwanese indexes also sold off.

An NBC exit poll showed that 56% of US voters regarded data centres as a “crucial or major factor” in how they voted. Texas and Georgia, which have both elected Democratic governors, together account for “about a third of all AI data-centre spending and projects”, says Tapas Strickland of Moomoo ANZ. “The fear is that the pace of the AI infrastructure roll-out could slow, potentially quite markedly in Texas,” where Democrats want a moratorium on new data-centre development.

Still, more red tape may not burst the AI bubble. “State-level restrictions could have implications for specific” companies and data centres, but ultimately 'data-centre projects will probably shift to friendlier states”, Gil Fortgang of T Rowe Price tells the Financial Times. The Democrats may slow data-centre development, but won't want to “kill the goose that lays the golden eggs”, says Alicia Garcia-Herrero of Natixis.

Democrats’ victory leaves glass half full for markets and investors

Markets bounced back on Wednesday as analysts encouraged a glass half-full view. First, Republicans did not contest the legitimacy of the US election results (no longer a given in the US). Second, a divided US government – with one party in the White House and another in Congress – is historically good for investors. “Control of Congress really only has implications for the economy if it means significant legislation can be passed,” says Daniel Morris of BNP Paribas Asset Management. We're actually in the opposite scenario – gridlock is likely and hence no new regulation.

This is “the best possible outcome for markets”, says Nannette Hechler-Fayd'herbe of Lombard Odier. History suggests that, on average, the US equity market performs best in the third year of a presidential term, says Russ Mould of AJ Bell. Double-digit returns have been the norm since World War II, with only one presidential third year bringing losses.

But there is nervousness about how Donald Trump will react to a setback. Gold rallied 3% to $4,040 (£2,999) an ounce, suggesting investors are fearful. The vote will have “big implications” for US and global politics, but economically the direct implications won't be huge, says Christopher Hodge of Natixis. Corporate earnings and the economic cycle are more decisive factors for investment returns.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

“Our investment philosophy is very simple,” says Stephen Yiu, manager of the Blue Whale Growth Fund. “It’s basically Follow the Money.”

“The way that we understand how the world economy operates is that there are three sources of spending: consumer spending, enterprise [business] spending and government spending. You want to invest into companies that are going to receive incremental spending from one of these three sources.”

That thinking determines where Blue Whale’s highly concentrated and unconstrained portfolio looks like at any point. At present, Yiu is heavily invested in tech stocks and has been for several years – but he is very clear that he is not primarily a tech investor.

He’s simply looking for the type of companies that have the tailwinds to support them. That changes a lot over time, and investors who don’t change in response can get caught out.

Speaking on the MoneyWeek Talks podcast, which is available on all podcast platforms and YouTube, he says: “If I go back in time when we first started in 2017, the leadership - the companies that end up receiving money – are very different,” he says. Consumer spending was a lot more robust in the 10 years up to 2022, helped by factors such as low interest rates, low inflation, the benefits of globalisation and very little geopolitical uncertainty.

“If you look at companies that are exposed to consumer spending, they are in a much more difficult place now.”

Selling software, buying hardware

Since then, technology – more recently artificial intelligence – has been the beneficiary of higher spending. However, even within the tech sector, Blue Whale’s holdings have evolved as the winners and losers changed.

“A very important part of any investor playbook is that you need to sell companies that could have their earnings power disrupted,” says Yiu. So around two years ago, he moved out of the software stocks that had been key parts of his portfolio in the early years for the fund

These companies had been extremely attractive – they had switched away from selling one-time software licences to a subscription model, giving them recurring revenue. Costs of switching away from them were often high, meaning that many users were locked in.

However, while many investors assumed all technology companies were likely to be AI winners, Blue Whale recognised that “AI is actually a big risk to the moats of software companies” because it would make it easier to create rival products – even to the extent of helping smaller firms create their own systems.

So today, while the fund is still heavily weighted towards the tech sector, its actual exposure is principally in hardware stocks – beneficiaries of the huge demand for chips and other components that power the AI boom.

Structural changes

This rotation into stocks that are benefitting from physical shortages is evident in other additions to the fund, some of which may seem surprising for a fund that is associated with high-quality growth stocks.

Blue Whale now holds an oil and gas producer and more recently a refiner, anticipating sustained high oil and fuel prices driven by conflicts and geopolitical tensions – an example of what Yiu calls a shift in the investment regime.

Similarly, the fund has invested in a European defence firm, on the basis that the amount of government spending flowing into the sector has shifted structurally.

“Before 2022, nobody should ever be interested in European defence companies,” he says. Defence budgets were “very much linked to European nominal GDP and European GDP has been very disappointing for many years”.

Now, capital is pouring into the sector, while the fundamentals look better: faster growth, stronger balance sheets and probably a less demanding relationship with their government clients. “Then you look at the return on investment profile, the margins, the shareholder returns, the free cash flow conversion, it’s all completely changed.”

Bottom-up, not top-down

While Blue Whale pays close attention to top-down factors such as geopolitics that shape a company’s prospects, Yiu stresses that they are fundamentally bottom-up investors, focused on understanding what everything means for a company’s earning power.

That’s a crucial part of the firm’s advantage over other investors, he says. After all, Blue Whale invests in very large companies, where everybody has access to the same primary information. Access to management may be limited and unlikely to confer any advantage.

“You have to go back to the basics, which is ‘Do you understand the business or the ecosystem better than the market?’”

About the podcast

MoneyWeek Talks is a podcast that helps you unlock the secrets to financial success. Editors Kalpana Fitzpatrick, Andrew Van Sickle and Cris Sholto Heaton are joined by influential guests – from CEOs and entrepreneurs to economists and fund managers – to share their top tips on managing money, investing wisely and building wealth.

Follow the MoneyWeek Talks podcast and get ready to make it, keep it and spend it with confidence. You can also watch the episodes on our YouTube channel.

Concerns over the scale of capital expenditure on artificial intelligence by the industry’s biggest tech companies have seen trillions of dollars wiped off their combined market valuation.

Between its peak on 6 October and the end of the month, the Nasdaq 100 dropped by more than 9%. Estimates put the market value of these losses at around $5 trillion.

“For three years, tech giants only had to announce investment in artificial intelligence (AI) to be rewarded by the stock market,” Grégoire Kounowski, head of advisory at wealth manager Norman K. observed. “That is no longer the case. The third-quarter earnings season marks a turning point: investors now want to know when this spending will pay off, and how it will be financed.”

AI’s biggest names – especially hardware providers like Nvidia (NASDAQ:NVDA) – have seen surging profits and their market capitalisations boom as spending on AI infrastructure has soared.

But investors are becoming increasingly sceptical over the sustainability of these increased AI profits, and appear ready to penalise businesses that can’t demonstrate clear returns on their AI investments.

Has AI cannibalised Google Search?

The rout began when Google’s parent company Alphabet (NASDAQ:GOOGL) reported its results for Q3 of 2026 on 28 October. Diluted earnings per share (EPS) fell 26.1% year-on-year to $2.12 – well short of analyst forecasts of $3.02.

Although Google Search volumes reached record levels, revenue growth in the segment slowed to 12% from 17% in the previous quarter – suggesting that Google’s own AI overviews may be hurting its ability to monetise its core product.

Alphabet’s shares tumbled by 12.6% the following day and lost a further 1.4% on 30 October.

“Alphabet saw $463 billion wiped off its market value [in a single day],” said Dan Coatsworth, head of markets at wealth manager AJ Bell. “Alphabet’s results could go down in history as a major turning point in the AI story. They showed another quarter of cash burn and that AI summaries on search were affecting advertising revenue. The bear case many had dismissed is now becoming a reality.”

Alphabet was joined by Intel (NASDAQ:INTC) which fell by 28.6% on 30 October, following an alarmingly bleak forecast for Q4 accompanying its Q3 results the previous evening.

“The gigantic declines we’ve seen this week in shares such as Alphabet and Intel suggest that investor sentiment has been turned on its head,” said Coatsworth. “Current price action implies tech investors want to jump off the treadmill rather than slow it down.”

How do investors feel about AI?

The longevity of the AI stock market boom will largely depend on how the market sees the potential of the technology.

Research from the Association of Investment Companies (AIC), an industry body representing the UK’s investment trusts, indicates that investors are more negative than positive at present, despite recognising some of the technology’s long-term potential..

Based on a survey of 400 private investors and 200 intermediaries, the AIC found that misinformation (74%), job losses (73%) and loss of human skills (72%) are the top three concerns investors have about AI. Loss of privacy and increased pressure on water resources were also of concern to 51% and 45% of respondents, respectively.

“Given the apocalyptic warnings from AI companies themselves, it’s no surprise that investors are concerned about potential negative impacts of AI,” said Nick Britton, research director at the AIC. “We’ve also seen waves of AI-related job losses reported in the news, while actual or potential benefits of AI remain more distant and hypothetical for most people.

“As long as the AI investment boom continues and investors are making money, such fears may be overlooked,” Britton continued. “But if there is a market correction then AI companies may have to work harder to persuade investors that they can deliver growth while addressing their wider concerns.”

However, 68% said improved access to healthcare and medical breakthroughs was positive.

Will the AI boom recover, and should you invest?

This isn’t the first time that investor sentiment has led to AI selloffs over the last four years, and the market has recovered following every previous bout of pessimism.

Key to whether or not stocks recover will be upcoming earnings releases from the likes of Apple (NASDAQ:AAPL), Palantir (NASDAQ:PLTR) and Advanced Micro Devices (NASDAQ:AMD).

“What investors will want to see to avoid sell offs is a 'reasonable' amount of capital expenditure and growth beating expectations,” said Susannah Streeter, chief investment strategist at wealth manager Wealth Club. “That’s the sweet spot, growth without runaway spending.”

If you think the AI boom could recover then you could consider investing in an AI-focused exchange-traded fund (ETF), or an investment trust that focuses on technology such as the Allianz Technology Trust (LON:ATT) or Polar Capital Technology (LON:PCT).

Equinix (Nasdaq: EQIX) is a real-estate investment trust that invests in data centres in order to rent them out to technology companies. It has amassed an impressive portfolio of 282 AI-ready data centres in more than 70 countries and has partnership agreements in place with a large number of tech giants, including Nvidia, Oracle, Amazon Web Services, Dell and Google.

AI is clearly a transformative technology. However, rather than opt for companies working on AI, many investors have adopted the indirect approach of buying firms facilitating the boom, notably those operating data centres. Data centres are used to store vast amounts of information collected by technology companies, which need the data to train their AI models.

At present, there is a huge demand from Big Tech, which is trying to build out capacity as quickly as possible. However, just because an industry is doing well doesn't mean that all the shares in it are worthwhile investments.

Equinix's shares seem to have lost momentum

All this sounds positive, but there are a few notable problems to bear in mind. All the tech companies have been expanding data-centre capacity on the assumption that the demand for them will continue to grow at a high rate. This supposition may be overoptimistic, especially given the mixed impact of AI on real-world productivity that many firms have experienced. Several major studies have highlighted a high failure rate of AI projects.

There has also been growing interest in AI models developed by technology firms in Asia that potentially require much less computing power and therefore need fewer data centres. Furthermore, a backlash against data centres – which are unsightly, noisy and wasteful as they consume a great deal of water and energy – has affected some projects that Equinix has been involved in (such as one in South Africa). The upshot is that demand could slow down, or even reverse.

However, even if appetite for data centres continues to grow, Equinix's extremely high valuation looks likely to hamper the stock's progress. The firm is expected to grow revenue at under 10% a year over the next few years, a solid but not spectacular pace, and it achieves a return on capital employed (ROCE) of only 6%. Yet the stock trades at 50 times estimated 2027 earnings, and nearly ten times current sales. While other data-centre firms also have high valuations, some of them are more modestly valued. Iron Mountain, for instance, is a similar business selling for a more modest (but still high) 40 times 2027 earnings.

From a technical perspective, moreover, Equinix's shares seem to have lost momentum. They have drifted downwards over the past six months and are trading below both their 50- and 200-day moving averages. I would therefore suggest shorting Equinix at the current price of $953 at £3 per $1. I would cover them if they hit $1,283, which gives you a total downside of £990.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

The latest battery pack from Chinese company CATL promises to put range anxiety to bed for good, with the potential for 1,500km on a single charge. When they finally come to top up, electric car drivers can add 520km of range in just five minutes. Chinese car maker BYD has introduced a vehicle-mounted drone platform. In theory, the average Chinese electric car driver could drive from Beijing to Shanghai on a single charge, receiving food deliveries on the way.

This is just one way Chinese manufacturers are reshaping industries and a clear sign of China’s extraordinary innovation impulse. China is leading the way on patents, currently submitting around 4x the level of patents as the US. It is a global leader in areas such as robotics, clean energy technology and electric vehicles (EVs), and is increasingly competitive with the US on artificial intelligence (AI).

Yet many conventional economic indicators for China look depressing. Retail sales, for example, have barely grown over the last 12 months. The housing market, long a drag on household wealth, continues to go backwards. Deflation is a spectre looming over the economy. There is huge over-supply in certain industries thanks to inefficient government subsidies and difficulties in key export markets, such as the US.

Potential investors in China must deal with these competing realities. The Chinese stock market has real breadth and encompasses the good and bad of its economy. On the one hand are innovation powerhouses, on the other are sclerotic state-owned enterprises run for the good of the country rather than the good of shareholders.

Shifting sources of growth

For Will Mackintosh-Whyte, fund manager of asset manager Rathbones’ multi-asset portfolios, this is the sign of an economy in transition.

“We believe that the growth model in China is changing,” he says. “Property and infrastructure are no longer doing the heavy lifting they once did. Increasingly advanced manufacturing – semiconductors, automation, robotics, machine tools, EVs, batteries – and increasingly sophisticated capital goods, technology and AI… are the new growth drivers.

“China's manufacturing competitiveness remains an important source of support to the economy, and is likely to be a bigger driver of growth than the consumer,” he added. Domestic consumption, by contrast, “while strategically important, is currently weaker, weighed down by the continuing softness in property impacting consumer confidence”.

This dual-speed economy is evident in funds investing in the region. While returns from the Chinese stock market have, as a whole, been pretty miserable over the last 12 months, some funds have delivered excellent returns. The gap between the top and bottom performing funds in the IA China sector is 64% over three years. Top-performing funds such as the T. Rowe Price China Evolution Equity Fund or JPM Greater China Fund have risen more than 50% over that period, but the weakest funds are under water.

Targeting innovation

China has seen the same boom in AI infrastructure spending as the US. It is now becoming increasingly competitive with the US, albeit with a different model. Its AI is open-source, meaning it can be used and adapted by anyone. This makes it different from US AI, which is owned by individual companies. China doesn’t have the sophisticated ‘frontier’ models for AI, such as those from Anthropic, but these aren’t necessary for many of the hoped-for productivity gains from the new technology.

Mackintosh-Whyte, for example, is gaining exposure to AI growth through cloud provider Alibaba (NYSE:BABA).

“It gives exposure to Chinese cloud, enterprise AI and model adoption, which has its own chips and around 10,000 enterprise large language model (LLM) clients,” he says.

He also has exposure through non-US listed companies with significant Chinese businesses. “ASML (AMSTERDAM:ASML) and Applied Materials (NASDAQ: AMAT) are names we own who produce products which are part of the semiconductor supply chain where China is an important end market. We have copper exposure – directly and via owning Freeport-McMoRan (NYSE:FCX) – where the Chinese investment in the electricity grid is driving significant demand for the metal.”

Dale Nicholls, portfolio manager of Fidelity China Special Situations (LON:FCSS), says it is still early days on AI investment and there will be winners and losers, but initial signs for many of the large platforms – TenCent (HK:700), Alibaba and Bytedance - is positive.

“Encouragingly, we are seeing them become more disciplined elsewhere, reducing losses in highly competitive areas such as food delivery while directing more capital towards AI infrastructure, cloud and applications that could strengthen their long-term competitive position,” said Nicholls.

Nevertheless, as in the US there are some worries over the potential for an AI bubble. Valuations in AI companies in China may not be as stretched as their US peers, but they have been subject to similar excitement. With that in mind, Nicholls is seeking less obvious routes into the Chinese innovation trade.

He expects the next phase of adoption to extend beyond digital applications into physical or ‘edge AI’.

“Robotics, autonomous driving, smart devices and wearables should become increasingly important as intelligence is embedded directly into products and services,” he says. He holds robotaxi group Pony.ai (HK:2026) in the portfolio.

Looking beyond AI, Nicholls also holds battery pioneer CATL (HK:3750). While the electric car market looks saturated and most companies are unprofitable, CATL should be a major beneficiary of the electrification trend. It is also moving beyond cars, leveraging its capabilities in energy storage systems to capture growing demand from the AI infrastructure buildout.

“As AI deployment increases electricity requirements, we see opportunities across the wider power ecosystem,” said Nicholls.

Roderick Snell, manager of the Baillie Gifford Pacific Fund, is investing in some direct AI beneficiaries, such as Mech-Mind (HK:9615), a company that develops 3D machine-vision systems – effectively the eyes of robots. However, he is also investing in energy, resources and construction companies that will benefit from demand for power infrastructure and critical materials, both from AI and from the electrification of China’s economy.

How can you invest in China?

A low cost route to access China’s stock market would be through an index-tracking fund or exchange-traded fund (ETF). The HSBC MSCI China UCITS ETF (LON:HCHA), for example, would give investors a decent weighting in Alibaba and TenCent. The problem is that it also gives access to a lot of ‘old economy’ Chinese companies. It has just 11% of assets invested in technology as of 30 September, with higher weightings in consumer and financial companies. Like the broader Chinese stock market, it has fallen over the past 12 months.

Paul O’Neill, chief investment officer at Bentley Reid outlines a possible alternative: “We own the Amundi Core MSCI China A Swap ETF (LON:C027). It tracks the MSCI China A index, which covers large and liquid A-shares listed in Shanghai and Shenzhen, priced in RMB.” The ETF’s top holdings include CATL and chipmaker Cambricon (SHANGHAI:688256). The ETF mixes battery and AI hardware leaders with banks, insurers and utilities.

“The top ten are only about 18.5% of the fund, and it is diversified across roughly 400 stocks,” said O’Neill.

Active funds may provide more targeted exposure to China’s new innovation economy. The Baillie Gifford Pacific fund, for example, has 48% of assets invested in technology, with only negligible amounts in areas such as real estate. The Allianz China A-Shares Equity Fund focuses on ‘created in China’ industrial and technological advances.

The innovation story in China is every bit as powerful as that of the US. There are industries where China is leading the world and it looks increasingly competitive in areas such as AI. Investors have plenty of options to target this part of China’s economic growth story, while ignoring its sluggish, legacy industries.

AI and technology stocks perked up early this week following the publication of the IPO prospectus for Anthropic. Investors took heart from news that the Claude chatbot maker pulled in more than $16 billion in revenue in the first half, says Ipek Ozkardeskaya of Swissquote.

But cracks have begun to emerge in parts of the AI landscape. The cost of insuring debt issued by CoreWeave, a “neocloud” data-centre business, recently rose to allow for “a more than 60% chance of default within the next five years”, according to Bloomberg. CoreWeave is a highly leveraged owner-operator of potentially fast-depreciating assets, says Victor Balfour of Rothschild & Co. It's little surprise the shares have plunged 67% since their peak in June 2025.

That doesn't necessarily spell doom for AI more generally. The technology “hyperscalers” are much more robust – Microsoft shares have risen 38% since the start of July. These companies are issuing a lot of debt to pay for their AI investments, but their strong operating cash flows and balance sheets are “reassuring”.

Stijn Van Nieuwerburgh of Columbia University estimates that spending on AI data centres, power systems, networking infrastructure and specialised chips will hit $10.3 trillion in 2025-2032. On an annual basis, investment will run at 3.63% of US GDP. For comparison, the late 1990s dotcom boom saw capital spending equivalent to 1.1% of annual GDP; the 1870-1890 US railroad mania was previously the historical leader at 2.24% of GDP.

As the credit boom goes on, Van Nieuwerburgh warns financing is moving into an opaque world of “joint ventures, private credit, securitisation, special-purpose vehicles, lease commitments and loan guarantees”. That makes it much harder to keep track of where dangerous credit risks are accumulating.

AI profits bubble?

The Nasdaq 100 trades on a forward price-to-earnings ratio a little over 20 times, a premium rating but not obviously bubbly. That's because the “bubble is in profits, not valuations”, says Edward Chancellor on Breakingviews. Big American blue-chips have reported breathtaking earnings growth; S&P 500 earnings per share rose 26% in the 12 months to July, far above the long-term average of 7%. Morgan Stanley predicts the 42 most AI-exposed firms will grow profits by 67% this year.

But the AI profit boom is something of an accounting “mirage”. When a hyperscaler such as Microsoft buys a chip from Nvidia, the latter is able to book an immediate sale, but the former acquires an “asset” that depreciates slowly over a number of years. Unless the chips deliver the promised earnings bonanza, an earnings hangover will eventually catch up with the hyperscalers – “by the end of the decade… annual depreciation charges will range between $500 billion and $1 trillion”.

Earnings bubbles have occurred before, notably in US housebuilders at the 2006 peak of the property bubble. Should US profit growth revert to the historical mean and losses emerge at data-centre operators and AI frontier labs, then the damage could “metastasise through the financial system”.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

JCDecaux (Paris: DEC) has become an unexpected winner of the global AI boom. The company, which owns thousands of billboards and other advertising assets worldwide, isn't the most obvious AI play. But thanks to its vast real-world estate, it's reaping the benefits of a shift in spending by advertisers. As the market has woken up to this potential, the shares have charged higher, rising more than 70% this year alone.

Why JCDecaux's advertising strategy is a success

JCDecaux operates in the so-called out-of-home (OOH) advertising market. As I mentioned in a recent feature, this is one of the few traditional advertising mediums that have seen demand from advertisers rise over the past decade. According to WPP, one of the world's leading advertising companies, OOH advertising revenue is expected to grow 5% to $57.5 billion globally in 2026, then grow another 5.3% in 2027. By 2031, the group has pencilled in a total market size of $73.7 billion. WPP calls this growth a “structural anomaly”: while other traditional advertising mediums, such as radio, TV and print, are shrinking as a percentage of global spending, OOH is holding its own.

In the $720 billion-a-year global advertising market, digital platforms get the largest share of spending. Platforms such as Facebook and Google account for half the market. These are designed to get users to click through to websites and buy products or services. Alongside these are AI platforms, such as ChatGPT and Gemini, that have started embedding advertisements into responses, a market expected to exceed $100 billion by 2031, up from only $5 billion today.

The thing is, no one really knows whether people are seeing the advertisements displayed on these platforms. Fraud is rife and more than half of all online traffic is now just AI bots. Estimates vary, but on average, figures suggest that as much as $1 of every $3 advertisers spend online is lost to fraud or non-human engagement. In some cases, that rises to $1.5 of every $3.

Advertising has always had some level of such “leakage”, but this level is unprecedented. OOH fills in this gap nicely. Advertisers can pick one location, say Bank Underground station in the City of London, and know they're in one place, with tens of thousands of people passing through every day. It's straightforward to target a particular audience in a given place with digital advertising, but there's never a guarantee that that person or digital device is where it says it is. It's hard to move a station, airport, or shopping centre, where the most lucrative OOH billboards tend to be located. Stadiums are another top play.

That's not to say OOH is winning at digital's expense. It's part of a toolkit. OOH is mainly used for visibility and brand building and helps drive awareness of the brand, increasing the likelihood you'll click a similar advertisement the next time you see it online. As analysts at Berenberg note, OOH enables the “critical need for brand differentiation as AI rapidly commoditises content”.

JCDecaux is a family success story

JCDecaux effectively created the market for outdoor advertising. In 1964, Jean-Claude Decaux invented the concept of “street furniture”, with bus shelters financed by advertising in Lyon (France). Two years later, it started rolling out the same concept across Europe. Since then, the business has grown alongside the advertising industry. It has worked with local governments, advertising agencies, clients, and world-leading designers, such as Norman Foster, to create street furniture and OOH advertisement placements that fit the environment, using space that would otherwise go unused.

Today, the founding Decaux family still owns a majority of the business (66% via JCDecaux Holding SAS) and it holds a 12% share of the global OOH market, making it the largest single player. France is still its largest single market with 17% of revenue generated in the country, followed by the UK (11%) and US (8%). The rest of Europe accounts for 30% of revenue, Asia-Pacific for 21% and the rest of the world 14%. Street furniture is the group's biggest revenue segment, making up just over 50% of sales and 66% of earnings before interest, tax, depreciation and amortisation (Ebitda).

Should you invest in JCDecaux?

Operating OOH assets is not particularly capital- or resource-intensive. The company has reported near 20% Ebitda margins for the past few years and Berenberg has pencilled in margin growth to 22.2% by 2028. While the group has some debt, it's manageable at 1.1x Ebitda. What's more, with free cash flow of €336 million in 2025, rising to €354 million by 2028, analysts believe the company is on track to wipe out its debt within the next three years.

However, that's unlikely as the group is hunting for deals in multiple markets, in the range of around €100 million – or around 50% of annual free cash flow after dividends. Even without these deals, Berenberg believes the group can grow its top line in line with the market over the coming years, with adjusted Ebitda rising faster – 6%-9% thanks to margin growth.

Even after running 70% higher this year, the shares trade at a relatively undemanding valuation of 17.1 times forward earnings and a free cash-flow yield of 6.6%. That does not appear to give any account for its growth or possible bolt-on deals. On Berenberg's 2028 estimates, JCDecaux is selling at 14.4 times earnings and a free cash-flow yield of 7.2%. The shares also yield 3%.

JCDecaux stands out for its rock-solid balance sheet, free cash flow and presence in a market that is full of AI-related risk, and seems to be holding its own against the digital onslaught.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Our ETF portfolio has become a little over-exposed to tech, so it's time to introduce a new fund.

“Unprecedented” is a word that gets thrown around too easily, but make no mistake: the AI infrastructure boom that many investors are betting on would be unprecedented in scale – if it materialises.

Total spending based on the announced data-centre pipeline could amount to $10 trillion in the US between 2025 and 2032, according to a paper by Stijn Van Nieuwerburgh of Columbia Business School, which would be 3.6% of GDP. The previous record was the 2.2% of GDP spent on railways between 1870-1890.

This doesn't mean that it's impossible, but if expectations are not met, makers of chips and other hardware – who are earning huge profits from an incredibly tight market – may not enjoy as many years of limitless demand as forecasts currently suggest. And the recent jitters in data-centre and memory stocks are a sign that the cost of disappointment could be severe.

The most unexpected downside could be in emerging markets; almost 45% of the MSCI Emerging Markets is now in tech, mostly in hardware suppliers. At present, we have 10% of our MoneyWeek ETF portfolio in the iShares MSCI Emerging Markets ETF (LSE: EMIM). We also have 5% in WisdomTree True Emerging Markets ETF (LSE: WEMP) to tilt to traditional emerging markets (there's no China, Korea or Taiwan in that fund).

I have often said that if we start to worry more about the sustainability of the AI cycle, we need to keep an eye on our ETF portfolio's tech exposure via EMIM. Increasingly, we feel the urge to act on that, so we are going to cut the 10% in EMIM to 5%. This brings our headline emerging-markets exposure to 10%, but the real effect is to bring down the total tech exposure that is hidden in our overall emerging-markets position.

We can still have higher tech exposure if we want, but do so in a way that becomes explicit – eg, hold something like iShares MSCI Global Semiconductors (LSE: SEMI). Or we can dial it down altogether. Either are valid – the key is that we can see more clearly what we have chosen to do.

When we look at how different sectors are performing, it is clear the market's top anti-AI or risk-off trade is healthcare. You can see in the chart how it has moved broadly inversely to the tech-influenced markets indemerging-ex over the past year.

A change to the MoneyWeek ETF portfolio

This makes sense. Other defensives such as consumer staples are battling a number of problems, including changing tastes and weaker pricing power. Of course, healthcare is not risk-free either. Medical costs are a political issue. The patent cliff never goes away. Concentration is not low: Eli Lilly is 11% of the sector index. It is not cheap: on a forecast price/earnings ratio of 17.5, the MSCI World Health Care is at the same valuation as the overall MSCI World. However, demographics bode well for demand and earnings should be steady. It is attractive in its own way, as well as diversifying us from AI. So we will add 5% in Xtrackers MSCI World Health Care ETF (LSE: XWHS) to our portfolio.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

US insurance company Lemonade (NYSE: LMND) was founded just over a decade ago. It uses AI at every stage of the insurance process, from selling policies to dealing with claims, though it still employs human staff, arguing that this helps cut administrative costs.

While it is true that AI is undoubtedly going to change the world, not every AI-related share is a compelling investment. To make a success of AI, a company must pass several tests.

First, it has to come up with a way of doing something more efficiently. Next, it must find a way to make money. It will need to fend off competition, both from incumbents and new entrants into the sector. Finally, even firms that meet all these criteria can become poor investments if the shares rise so far that they become overvalued.

Lemonade is still losing money

Lemonade says it spends approximately 5% of the money it earns from premiums on handling claims, as opposed to the 9% figure typical for the industry as a whole in the US.

Lemonade also believes that AI can significantly accelerate operations, with over half of claims paid out instantly and most policies sold in under 90 seconds. Lemonade is undergoing a period of rapid growth, with sales increasing from $94 million in 2020 to $738 million in 2025. Revenue is expected to keep expanding at around 30% a year.

But the company continues to lose money, with estimates suggesting that – at the very least – it will be in the red both this year and in 2027 (though management believes that the firm may start to achieve positive earnings before interest, taxes, depreciation and amortisation – Ebitda – in 2027).

The competition is catching up to Lemonade

Another difficulty is that while Lemonade may be ahead of the technology curve today, other companies are catching up by pouring large sums into AI. For example, rival Allstate announced this summer that it is launching ALLIE (Allstate's large language intelligent ecosystem), while other major competitors such as State Farm plan to use AI too.

And while Lemonade argues that its status as an AI-driven company will always give it an advantage over companies built around legacy systems, it also has to compete with other AI-driven companies, notably Root.

Lemonade could begin to turn a consistent profit and fend off competition from both new and old rivals. But investors still have to acknowledge that its valuation is relatively expensive.

It trades at 3.7 times sales, more than four times the 0.83 times sales that Allstate, a profitable and fast-growing company, sells for. Moreover, Root, which operates a similar tech-driven model but manages to turn a profit, only trades at 0.48 times sales. Furthermore, Lemonade's recent share-price performance suggests that investors are starting to sour on it. The stock has fallen by more than 50% from its 52-week high and has lagged the wider market over one, three, six and 12 months.

It is also below both its 50- and 200- day moving average. I would therefore go short at the current price of $46.46 at $45 a share. Put the stop-loss at $67.46, which gives you a total downside of £945.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Sheikh Tahnoon is the younger brother of the UAE’s ruler, Sheikh Mohamed bin Zayed (MBZ). He combines his role as the emirates' top money man and dealmaker – he runs the UAE's largest sovereign wealth fund and the First Abu Dhabi Bank – with a vital position as head of national security. He is perhaps the individual who has done most to further his country’s relations with the US.

Twenty years ago, the United Arab Emirates was “at the centre of a political firestorm” when a bid to buy up six major US seaports, through the Dubai-owned logistics company DP World, triggered uproar, says the Financial Times. Alleging the oil-rich Gulf state was a financing hub for al-Qaeda, some in Congress warned of a dire threat to US security if the UAE was allowed to run strategic ports. DP World was forced to back out of the sale. It also proved “a watershed moment for the UAE – and its relationship with America”.

The ruling Al Nahyan dynasty, implacably opposed to Islamic extremism, decided it was time to fix this “fundamental misunderstanding”. And they did. Two decades on, the UAE is one of America's most capable allies and business partners in the Middle East.

Sheikh Tahnoon, who is rarely seen in public without his trademark dark glasses, cuts an enigmatic figure, says The Guardian. His influence is global. He presides over an economic empire worth roughly $1.5 trillion – equal to the entire output of Indonesia or Turkey – with an extraordinary breadth of investments. In the UK alone, he oversees healthcare providers and property giants and manages investments in public utilities such as Thames Water. This summer, the sheikh's investment conglomerate IHC bought a controlling stake in Richard Caring's Ivy hospitality empire.

Sheikh Tahnoon's race to global dominance with G42

But the business that has done most to cement relations with the Trump administration and Silicon Valley is Sheikh Tahnoon's AI and cybersecurity company G42, which has now successfully positioned itself at the heart of the global AI establishment. Microsoft has invested $1.5 billion and its vice-chair, Brad Smith, now has a seat on the G42 board. G42 is either a direct investor in or a major partner of Anthropic and OpenAI.

Donald Trump has “clutched Tahnoon and G42 close to his bosom”, and last year reversed a decision to restrict the export of AI chips to the UAE, paving the way for the construction of the biggest data centre in the world in a campus outside Abu Dhabi, says The Wall Street Journal. Dubbed “Stargate UAE”, the completed facility will have a 5GW power capacity, up to five times the size of today's largest hyperscale data centres. The US chip deal was announced two weeks after one of Tahnoon's investment vehicles acquired $2 billion worth of cryptocurrency issued by the Trump family's company, World Liberty.

Sheikh Tahnoon's father, Sheikh Zayed, who established the ruling dynasty following independence from Britain in 1971, is believed to have fathered 30 children, but it was the offspring of his third and favourite wife, Fatima, who emerged as the nation's powerbrokers. Tahnoon and his five brothers are known in the UAE as the bani Fatima, or “sons of Fatima”, says The Guardian, and grew up steeped in her “worldview that placed their family at the centre of the universe”.

Whether playing chess, video games or throwing Brazilian jiu-jitsu moves, Tahnoon, 57, is “obsessed with winning”. But his activities as the UAE's security chief reveal a darker streak, says The Guardian. “The UAE has a long history of enforced disappearances, secret detention sites and vicious reprisals against the family members of exiled dissidents.” Indeed, G42 is the “culmination of Tahnoon's dual identity as a financier and spymaster”. Much of its expertise derives from DarkMatter, a now-dissolved company renowned for its hacking and surveillance activities.

It may trouble some that the UAE is emerging as such a strong global force in AI and that Tahnoon is at the heart of it. Very few people outside the UAE even know his name. “He's quiet,” one former US national security official told The Guardian, “and he's always everywhere.” If he gets his way, his global reach has only just begun.

Artificial Intelligence (AI) has received bad press in recent weeks. Major AI models from OpenAI and Anthropic have been caught hacking other companies and even the Australian government, ‘going rogue’ and leaving their testing environments in the process.

Meanwhile, debates around the environmental impact of AI technology are heating up, with protests centred on data centres and their water usage.

Some are calling for a pause in the technology’s development. But concerns around AI safety are unlikely to hinder AI spending, says Jessica Inskip, director of investor research at stockbrokers.com, in the latest episode of MoneyWeek Talks, which you can also watch on YouTube.

Speaking to MoneyWeek's digital editor-in-chief Kalpana Fitzpatrick, she said: “I don’t think they’ll slow down spending, because they could redirect that spending elsewhere. The debates might even fuel the spending.”

Inskip says that debates around how energy-hungry AI is are fuelling attempts by firms to reduce power usage. She said: “In the US, our infrastructure is incredibly outdated in regards to the grid, and so a lot of the capex spending is going towards updating that infrastructure so it can support [AI].”

Water usage has also been a significant flashpoint over the environmental impacts of AI, with critics pointing towards the huge cooling demands AI data centres have. But firms are thinking of innovative ways to approach this issue too.

“We see that even with the water usage [debate]. There's new [data centres] that bring in water one time and then cool themselves. Now we're exploring data centers in space. Google has a project where they're taking their TPUs into space. [Others] are looking at even the ocean where it's very very cold because of the cooling that's needed.”

Inskip believes the concern around AI safety will in fact accelerate spending, not least because the technology is driving returns for its investors.

The beneficiaries of the AI boom are not just investors. Inskip claims that despite many locals being opposed to the building of data centres, they can actually be a boon for local economies.

“The environmental concern is very evident, and I completely agree with that. I actually visited an area where some data centres were and I remember seeing signs for rallies against these centres.

“But on the flip side, [the town] where this data centre was, the economy all of a sudden started booming. It’s like we had the industrial revolution or the railroads – it starts creating jobs,” she claimed. “You’re building these data centres and you need electricians and construction workers. So there is certainly good and bad with it.”

How investors can benefit from the AI boom

AI firms are now at the heart of many investors’ portfolios, with AI-linked stocks representing around half the S&P 500, a stock market index comprising 500 of the largest firms in the United States.

Many people will hold tech giants like Nvidia or Google’s parent company Alphabet, but Inskip feels that IBM (NASDAQ:IMB) may have slipped under the radar for investors.

AI firms can be broadly put into two groups: the ‘picks and shovels’, meaning the firms that are creating artificial intelligence, and the firms that are driving adoption.

For corporate adoption, IBM is an interesting stock pick, according to Inskip. She said: “On the enterprise side in the US we have IBM. A lot of financial institutions are very embedded with their data, and if you're a financial system or a healthcare system and you’re incorporating AI, you’re highly regulated so you’re going to want guardrails.”

These clients would not want AI agents to gain access to unauthorised and confidential data, nor would they want their confidential data to train any other models and potentially leak. This is the market IBM is targeting.

“So IBM is already embedded with their mainframe systems. They have a consulting arm which means they’re in the room when executives are making decisions from shareholder pressure over what is your AI strategy, and IBM is already there.

“What I love about IBM is they don’t have their own [AI] model. They partner with all of them.”

For more on the opportunities and dangers from the AI boom and why Inskip is bullish on Nvidia, watch MoneyWeek’s full podcast with Jessica Inskip on YouTube, or wherever you get your podcasts.

About the podcast

MoneyWeek Talks is a podcast that helps you unlock the secrets to financial success. Editors Kalpana Fitzpatrick, Andrew Van Sickle and Cris Sholto Heaton are joined by influential guests – from CEOs and entrepreneurs to economists and fund managers – to share their top tips on managing money, investing wisely and building wealth.

Subscribe to the MoneyWeek Talks podcast and get ready to make it, keep it and spend it with confidence. You can also watch the episodes on our YouTube channel.

One of London's most visible pieces of advertising has greeted railway passengers arriving at London Bridge station for the best part of 100 years. Emblazoned on a Grade II-listed building next to the track on Park Street is the slogan, “Take Courage”, the motto of the Courage Brewery.

This is considered one of the largest and most memorable so-called “ghost adverts” in London, although it's unclear when it was first painted. The Anchor Brewery originally built the property in 1820 and it remained a central location for brewing and operations until 1981, when it was acquired by the local authority. These ghost adverts can be seen all over London and remind us that advertising has been a core part of the UK economy for hundreds, if not thousands, of years.

A brief history of advertising

No ghost adverts in London are more than 300 years old (most of the city has since been rebuilt), but there are older examples elsewhere. Some of the earliest date from ancient Egypt and Mesopotamia, where traders and market vendors used clay tablets and papyrus posters. Archaeologists have also found examples of adverts chiselled out of stone dating back 3,000 years in ancient Babylon. Similar examples appear in ancient Greece, Rome, China and across the Middle East.

In ancient Rome, owners would commission a designer and manufacturer to produce a signboard that served as the business's primary street advertisement. This, it could be argued, was the beginning of what we now know as the advertising industry.

Advertising took another step forward in the 16th century with the advent of newspapers and magazines. The first weekly gazettes appeared in Venice in the early 16th century and in Britain the first weekly publications appeared in the 1620s. From the very beginning, newspapers, magazines and pamphlets carried advertising that helped foot the bill for printing and distribution. However, it wasn't until the mid-1800s that a confluence of factors accelerated the growth of the advertising industry into what we recognise today.

Early print advertisements were primarily in books, mainly due to each printer's desire to cross-sell. Quack medicines also commanded a lot of page space. That began to change in the 1850s and 1860s, when advances in mass production lowered manufacturing costs and an increasingly affluent middle class emerged around the world for the first time in modern history. This new wealthy class had discretionary income and sought out a variety of new produce.

Advances in technology, health and the growing demands of the affluent middle class produced a windfall for companies that could capitalise on these trends. Forward-thinking business owners accelerated growth with pioneering marketing campaigns. Thomas J. Barratt, the chairman of Pears Soap, was one of the first executives to build a brand around an advertising campaign. Barratt has been called one of the fathers of modern advertising in London thanks to the campaigns he created in the first few years of the 20th century. Barratt sought to position Pears as one of the country's highest-quality soap brands. To do so, he created an advertising campaign to drive consumer purchases. One of his most memorable slogans was, “Good morning. Have you used Pears' soap?” He also ran a series of adverts featuring well-groomed middle-class children, linking the product to domestic comfort, high-society aspirations and daily cleanliness. Barratt's approach focused on aspiration and a strong, exclusive brand image, backed by a robust supply chain to meet demand.

Companies were developing similar approaches across the pond. In the last decade of the 1800s, companies such as Procter & Gamble and Quaker Oats drove sales across the United States through national advertising campaigns. Tobacco producers were particularly prevalent in all markets.

Through the first few decades of the 1900s, the first global advertising agencies grew out of local offices. Advertisers began refining strategies across different media, such as print, radio and out-of-home billboards. In the 1920s, psychologists turned their attention to advertising, developing concepts of behaviourism and the consumer's basic emotions, such as love, hate, and fear, refining the strategies Barrett pioneered in London ten years earlier. Exploiting the insights of behavioural psychology became the calling card of one particular agency in Chicago: Lord and Thomas. Founded in 1873 and now known as FCB, it is the third-oldest advertising agency in the US still operating today. Albert Lasker bought the firm in 1912 and devised a copywriting technique that appealed directly to consumer psychology.

One of his most famous campaigns was for Lucky Strike cigarettes. The company wanted to encourage more women to smoke its cigarettes and so Lasker developed a series of ads encouraging women to smoke cigarettes rather than eating high-calorie snacks, using phrases such as “reach for a Lucky instead of a sweet”. The strategy helped Lucky Strike increase market share by more than 200% in its first year.

How the advertising industry became the money machine

Over the past 100 years, advertising spending has tracked global GDP growth. While spending is higher in some countries than others (the UK has the highest relative spend among major economies as a percentage of GDP), the global average has risen from roughly 0.3% of global GDP in the 2000s to around 0.8% today. According to WPP, that figure could hit 1% of global GDP by the end of the decade.

Despite this growth, the industry has often faced criticism for wasteful spending and poor returns on investment. One of the best-known criticisms of the industry is encapsulated in the quote: “I know that half the money I spend on advertising is wasted. My only problem is that I don't know which half” – a saying usually attributed to Henry Ford or the founder of Unilever and later the first Viscount Leverhulme. However, there's no evidence of either of these fathers of industry making such a comment. Ford was in fact highly committed to advertising, stating: “A man who stops advertising to save money is like a man who stops a clock to save time.”

The advertising industry has changed significantly since its early days, but the basic principles the early pioneers developed still apply. Advertising should be designed to capture attention and inform as directly as possible. Today, the industry can be divided into two parts. On the one hand, there are the businesses that sell space to advertisers. This market is dominated by the big four: Alphabet, owner of Google and YouTube; Meta, the owner of Facebook; ByteDance, the Chinese owner of TikTok; and Amazon. Together, these account for around 60% of global advertising revenue throughput, up from around 50% a few years ago. These are all digital-native or digital-dominant platforms that have leveraged their global exposure and creator content to build massive advertising operations. Outside these top four, the rest of the top 25 global advertising sellers include more traditional firms such as Fox, Walmart and JCDecaux.

The second part of the industry is made up of the agencies that design, plan and coordinate advertising across platforms. The biggest fish in this pond in the UK is WPP. Originally called Wire and Plastic Products, the company originally manufactured wire shopping baskets before it became an advertising holding company in the 1980s under the stewardship of Martin Sorrell. Today the group operates a sort of one-stop shop for companies and organisations that want to communicate their message to the outside world.

This process of coordinating campaigns and advertising spending is becoming increasingly challenging. We've long moved on from a world where advertisers only had to worry about painting signboards. Today, advertisers have a plethora of media to consider and the fastest-growing market is, unsurprisingly, generative search (AI). WPP Media's mid-year market forecast notes that advertising revenue on generative search platforms such as ChatGPT is expected to reach $5.1 billion globally in 2026, representing roughly 0.4% of total advertising revenue.

However, the media agency forecasts the market will grow at a compound annual growth rate of nearly 100% over the five years to 2031. This puts it on track to become the fastest advertising segment to reach $100 billion in annualised revenue in recent years. It took 22 years for traditional search (the adverts you might see when you search on Google) to reach this benchmark. It took 14 years for revenue on social media platforms such as Facebook to reach $100 billion, and spending on streaming TV platforms such as YouTube and Netflix has yet to exceed $60 billion annually, despite rapid consumption growth. The expansion of advertising on generative AI platforms is expected to drive much of the industry's growth in the coming years.

Billboards have not gone out of fashion

One medium seemingly benefiting from the growth of AI-based search is the so-called out-of-home advertising market. Out-of-home is one of the oldest forms of advertising. It generally refers to posters or signs located outside a user's property or premises, such as the ghost signs dotted across London. While spending on other forms of traditional media such as TV, print, and radio continues to decline, out-of-home spending is holding its own. According to WPP, that's because the medium is virtually guaranteed to deliver its message directly to humans, unlike digital platforms.

Indeed, earlier this year, we passed the point where more than half of the internet is now made up of machine-to-machine interactions (think AI agents filling in forms, or bots “liking” AI-generated videos), meaning there's an increasing chance human eyeballs will never see the ads that have been paid for. With out-of-home, advertisers can deliver their message to large numbers in physical environments simultaneously, the sort of scale and visibility that digital channels such as social-media platforms increasingly struggle to match.

One of the most prominent operators in this space in the UK is Global Media. The private company manages more than 253,000 outdoor advertising sites across the UK, as well as radio stations. It is one of the most important partners for the London Underground and operates billboards across some of the UK's most important transport hubs and across Europe. In the company's most recent fiscal year, out-of-home advertising revenue rose from £379.9 million to £425.9 million, and adjusted profits for the outdoor business rose 25% to roughly £155 million.

Alongside Global, JCDecaux (Paris: DEC) leads the world in this space. The Paris-based company accounts for around 12% of the global out-of-home advertising market, twice that of its closest competitor. Most of its assets are in Europe (30%), 8% in the US, 21% in the Asia-Pacific and 14% in the rest of the world. More than half of its revenue comes from out-of-home placements on the street, with 13% coming from billboard sales. In addition, 36% of sales come from adverts placed on public transport or around stations.

Analysts at Berenberg expect the company to grow 5.5% this year, faster than the wider market, thanks to growth in Asia and the US, which is underrepresented in the portfolio. While the company earns a healthy Ebitda margin of 21.4%, roughly 60% to 70% of group costs are fixed, meaning the capital structure has significant operational gearing. Still, the business is managed very conservatively, with near-zero debt and a 66% institutional shareholder in the form of the Decaux family. The shares trade at a forward p/e ratio of 15 with a free cash-flow yield of 7.3%.

The best advertising agencies to invest in

Then there are the advertising agencies. This market is really dominated by five major players: Publicis (Paris: PUB), WPP (LSE: WPP), Havas (Amsterdam: HAVAS), Omnicom (NYSE: OMC) and Dentsu (Tokyo: 4324).

Publicis is widely described as the best operator in the sector. Over the past five years, the company has won key contracts from competitors and poached top talent. It has focused on expanding its data and consumer-identity capabilities to gain an edge in the digital advertising market and capitalise on AI-driven market growth. Following a major restructuring and repositioning, growth is accelerating and is expected to rise from around 2.2% in 2026 to nearly 7% by 2028, according to Berenberg. The company is also generating strong cash flow. It is set to end 2026 with a net cash position of roughly €1.7 billion and the stock trades at a free cash-flow yield of 10.3% and a forward P/E of 11.

Once the leader of the group, WPP is now the laggard – and by a wide margin. Since the company lost its visionary CEO Martin Sorrell in 2018, it has really struggled to find its feet. Multiple rounds of job losses, re-organisations and rebranding have left the business gasping for air. Revenue this year is expected to come in at about £9.5 billion, down from £11.4 billion in 2024. However, following the arrival of new CEO Cindy Rose, green shoots have started to emerge. A year into her tenure and the former CEO of Microsoft UK has managed to stem the bleeding. According to numbers gathered by COMvergence, WPP Media has won around $3 billion in new business so far in 2026, versus $2.8 billion in losses last year. Account retention is now running at 43%, up from 16% in 2025. Still, the company has a lot of work to do to prove it is back in business. Analysts forecast revenue declines until 2028. In the meantime, the company is expected to cut jobs further to improve cash generation. It carries debt of around £2.5 billion, excluding leases, against Ebitda of £1.5 billion. A lot of bad news is baked into the valuation, with the shares trading at a forward p/e multiple of 4.9 and a dividend yield of 5.9%. The stock is trading at a free cash-flow yield of 5.2%.

Investing beyond the top tech players in advertising

The big tech players might dominate the ranks of the biggest advertising sellers, but they are no longer the pure-plays they once were. Alphabet, Meta and Amazon are funnelling hundreds of billions of dollars of advertising revenue back into the ground to expand their AI operations. This is turning businesses once touted for high returns on capital and asset-light models into capital-intensive infrastructure plays. Still, there's no denying they remain at the top of the pyramid for advertising spending. Alphabet, Meta and Amazon generated a combined $160.8 billion in advertising revenue in the second quarter of 2026, with Alphabet leading the way. Overall ad revenue rose 14.5% thanks primarily to the group's dominance in AI-powered search. Revenue from search rose 17%. YouTube advertising revenue rose 12.6% year on year as it continued to grab market share from TV budgets. The platform's annual advertising revenue has surpassed the combined spending of traditional legacy giants such as Disney, NBCUniversal, Paramount and Warner Bros. Discovery.

Beyond the top tech players, it may be worth a look at Pinterest (NYSE: PINS). This company has risen to become one of the top 25 advertising platforms in the world over the past decade thanks to its rich pool of ever-growing user content. Part social-media platform, part scrapbook, the platform is designed to help users find ideas, such as for home decor and fashion, and drive them to stores. More than half of the platform's users say they use it to shop. In a world where brands are focused on user-generated content to drive sales, this is a huge edge. The platform has logged 11 consecutive quarters of user growth (there are 640 million in total) and revenue expanded 18% in the second quarter. Pinterest has had a tough time as a public business, with the shares down 65% in the past five years. However, it's on track to generate around $1bn of free cash flow this year, against a market cap of $10.5 billion. It has net cash on the balance sheet and management has spent $2 billion buying back stock. It could be worth a look.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Utilities and infrastructure used to be a sector typically associated with producing steady inflation-linked income, but not now.

That’s according to Jean-Hugues de Lamaze, manager of the Ecofin Global Utilities and Infrastructure Trust, who said there are strong growth prospects and a high demand for capital expenditure (Capex) in listed infrastructure.

Speaking to Cris Sholton Heaton on the MoneyWeek Talks podcast, which is now available on all podcast platforms including YouTube, de Lamaze said ageing infrastructure put in place post-World War Two was offering investors plenty of opportunities.

“Infrastructure…energy networks, pipelines, but I also mean our bridges, roads, airports, which to a large extent were built in the 1950s, ‘60s, ‘70s,” he said.

“Since then, the percentage of capital expenditure…has been declining to a level of say 2.5%, 3% today. It used to be twice that a few decades ago.

“So we end up in a situation where we have a lot of infrastructure which is getting [sic] obsolete and the…opportunity to replace ageing infrastructure is…a big driving theme.”

De Lamaze said a fresh need to power artificial intelligence (AI) datacentres, electric vehicles (EVs), but also a wider electrification of the energy industry, at a time when major countries are struggling to produce power, was another reason why there are growth prospects in the sector.

“We’re lacking power generation resources…OECD economies, [including] the UK, have phased out coal to a large extent.

“In the UK, it’s been astonishing. Coal used to account for 50% of electricity consumed up until 15 years ago. Now, it’s zero.

“We haven’t invested in nuclear [power]. So we are in a situation where we are relying increasingly on intermittent power generation resources, [such as] wind and solar.”

He added: “This is an opportunity…where you have scarcity of power generation resources and demand for electricity, which is starting to increase and that’s, we believe, a long-term trend, at least for the next decade or two decades.”

Should you invest in water and waste management firms?

De Lamaze also put forward the case for investing in firms offering water and waste management services.

“Water faces a major scarcity phenomenon in urban areas worldwide,” he said. “For companies like Veolia…the French non-regulated water operator, it represents a major global market opportunity.”

Even toll roads can provide investors with opportunities.

“I don’t know whether you’ve travelled through France over the summer. You may have noticed that [if] you take a motorway, last year, you were paying €5, and this year you’re paying €5.30,” de Lamaze said.

“It’s a nice inflation pass-through revenue.”

For more, watch the full episode of MoneyWeek Talks with Jean-Hughes de Lamaze in conversation with MoneyWeek’s Cris Sholto Heaton on YouTube or listen on any podcast platform.

About the podcast

MoneyWeek Talks is a podcast that helps you unlock the secrets to financial success.

Editors Kalpana Fitzpatrick, Andrew Van Sickle and Cris Sholto Heaton are joined by influential guests – from CEOs and entrepreneurs to economists and fund managers – to share their top tips on managing money, investing wisely and building wealth.

Subscribe to the MoneyWeek Talks podcast and get ready to make it, keep it and spend it with confidence. You can also watch the episodes on our YouTube channel.

The video games industry has seen more change in recent times than almost any other. Over the past 15 years, gaming technology has “moved very quickly, the behaviours and culture have grown and expanded exponentially, and there has been a constant stream of changes that has made the industry exciting”, says Greg Weller, head of gaming partnerships at Generation Media. Some of the changes have been positive. The industry has become “an established, mainstream constituent of the entertainment industry”, with an estimated 3.6 billion people around the world now playing games in some form, says Gavin Smith, a senior commercial banker at Arbuthnot Latham.

However, “rising development costs, greater regulatory scrutiny and the concentration of player attention around a handful of major franchises could end up limiting that growth”, says Smith. Artificial intelligence, too, clearly has “significant transformative power”, though it's too early to say whether this will be good for the sector.

The industry's reputation for being “recession-resistant” has already been tested, with companies cutting around 45,000 jobs since 2022, as Adam Smart, global director of products for gaming at AppsFlyer, points out. Still, the opportunities outweigh the risks, making it a great time to invest. Consultant BCG estimates the market will grow by about 6% a year, reaching a value of $350 billion by 2030. Other estimates put the growth rate even higher.

Browser-based video games are the future

The big growth has come from games that you can play on your mobile phone or through your web browser as they are “really easy for virtually anyone to play” without having to splash out on expensive gaming hardware, says Matthew Dolgin, a senior equity analyst at Morningstar. Many of them also have a social element or are integrated into social media, which is bringing more and more people into gaming, including many of those who wouldn't otherwise have ever considered playing video games.

At the same time, the balance of power between mobile companies and the app stores has shifted. Until recently, developers just accepted that 30% or more of their revenue would go to Google Play or the Apple Store, says Stein Janssen, chief operating officer at browser-based games website Poki. But this has been increasingly challenged in the courts and in legislation. Apple has faced an investigation from the European Commission as well as lawsuits. Janssen expects this pressure to lead to a reduction in the cut that Google and Apple are able to take from sales of mobile games.

Indeed, many mobile games companies are starting to bypass Google and Apple completely by “starting their own stores for people to download games or buy in-game items”. Others are switching from mobile games funded by payments (either up front or in-app) to ones that are free, but rely on advertising revenue. Browser-based games are the future, says Janssen, as they can be played immediately, rather than waiting for a download.

Shifts in the big-budget video games subsector

Mobile and browser gaming may be the fastest-growing part of the industry, but the big budget games (or the AAA games as they are sometimes known) are still doing well. Revenue for this subsector will grow by a still respectable 4.7% a year for the next four to five years, according to BCG. Whenever “there are truly engaging games on the market new people start playing, and every year we see new generations of gamers log on”, says Andrew Bowell, CEO of immersive entertainment studio Iconic Interactive. Throw in the older generations who are already at home with games and the industry “should continue to grow”.

At the same time, outside expanding areas such as Asia, much of the growth is less about attracting new players and more about how revenue is collected – or in other words, about getting existing players to spend more, says Noam Korbl, CFO at PropFirms. Large parts of the industry have “moved from selling a boxed product once to charging for continued access, cosmetics, season passes and subscriptions”. Recurring spending from an existing player base is “far more predictable than hoping a single release performs well in its launch quarter, and investors tend to pay more for predictability than for creativity”.

Another big trend affecting AAA gaming is what Smart calls “platform convergence”, where the “old lines between console, PC and mobile blur as cross-platform play and cloud gaming let the same title reach players everywhere”. This means that studios and publishers now receive “diversified revenue streams”. This doesn't completely insulate them from the financial consequences of a flop, but it does mean that a shock in one segment, system or region “doesn't necessarily sink the whole industry”.

Video games conquer films and TV

Modern games have moved away from being just single products to being “franchises made up of a wide-ranging bundle of intellectual properties, with distinctive characters and even music”, all of which “lend themselves to broader application than just computer games”, says Aminder Khatkar, a partner at Brandsmiths. Such intellectual property (IP) can (and has) been exploited for lots of different things, including experiential events, but the most obvious application is in film and TV. There is a “definite convergence”, says Khatkar, between gaming and TV and movies.

The conversion of characters and franchises that have their roots in gaming into films and TV shows represents “one of the biggest opportunities across the media industry”, says Smith. Recent adaptations such as drama series The Last of Us and Fallout have shown that “gaming IP can attract substantial audiences beyond gaming itself”. Successful gaming franchises have “established fan bases, global reach, and richly developed worlds that sustain audience engagement across a range of formats”. In some cases, gaming IP is becoming more valuable than traditional film or television enterprises. Nintendo, for example, is expanding franchises such as Pokémon, Zelda and Super Mario into films, merchandise and theme parks.

The number of games being adapted into TV shows or films is increasing, says Stefan Seidel, a professor of information systems at the University of Cologne. Well over 200 adaptations have been commissioned since 2019, according to market research firm Ampere Analysis. And when an adaptation succeeds, “it lifts the games that already exist”. After the Fallout television series, for example, “the years-old Fallout 4 video game climbed back into the top five of the US sales chart, and daily players of the older games stayed far above pre-series levels for months”.

Interestingly, the circular effect is bigger for TV adaptations than films. The typical TV show increases the number of people playing a particular title by more than 200%, compared with 48% for films, according to Ampere's research. Still, even the boost from film is substantial and far greater than the increase in numbers that comes from updates and new downloadable content. The games industry is starting to become a much bigger and lucrative version of the toy industry, says Heather Delaney of Gallium Ventures, where TV shows based on the toys have long boosted sales.

Will virtual reality live on?

Delaney is a bit cooler on virtual reality (VR), which many previously saw as the wave of the future. Indeed, Facebook changed its name to Meta in October 2021 due to its belief that the future lay in what it called a “Metaverse” of people communicating (and playing) through virtual-reality headsets. Recently even Meta has been pivoting away from both the Metaverse and VR in general, closing three of its VR studios and laying off 10% of staff in the area, in favour of “adaptive reality” glasses that merge digital content with the physical environment. VR turned out to have too many limitations when it comes to gaming, not least the feeling of isolation while playing.

Meta's “gradual retreat” from VR “probably tells us something about where the wider subsector is heading”, says Smart. Enthusiasm for VR came at a time when people “were stuck at home” during the Covid pandemic and looking for escapism. Still, VR is unlikely to entirely disappear as it has attracted a “passionate core audience” and when done well can provide “one of the most exciting experiences in gaming”. The launch of new hardware, such as Valve's Steam Frame, may attract a new audience to VR.

Others are more optimistic. Meta may have cut back its investment on VR, but it has not completely abandoned it and it is still trying to push the technology, albeit in a way that is less high-profile, says Khatkar. Indeed, Meta's partial retreat shows the sector no longer needs to be “artificially propped up” by big tech firms but is strong enough to be left to individual companies producing software that can meet the demand, says Matt Celia of Light Sail VR. More than 20 million Meta Quest headsets have now been sold, with one in four teenagers in the US owning a VR headset, and a new product upgrade is likely in the near future. More than a million people use the headsets every day, “which is relatively high for an emerging technology”. Several independent studios and apps have started to make money from VR games.

AI won't kill the video games industry

One of the biggest questions hanging over the industry is the impact of AI. Some of the fears are clearly justified. It's hard to deny that the demand for processing power and chips created by AI “has pushed up the cost of consoles and computer equipment”, says Sean Kealy, VP of equity research at Panmure Liberum. But fears that AI will allow anyone to easily create games at zero cost, making games companies redundant, are also exaggerated – at least for the foreseeable future. AI “is not capable of producing a video game in and of itself, by itself, straight away”.

The release of footage generated by Google's cutting-edge AI world-building tool Project Genie, which caused the share price of many developers to fall when it was released in February, demonstrates the limitations of modern AI. “Video generation struggles to maintain coherent frames over more than a few minutes, with the entire world behind you different from the one that you walked through just seconds previously,” says Kealy. He also points out that there are open questions around copyright, not just in terms of the use of copyrighted content in AI, but also in terms of copyrighting AI-generated content.

There's a long way to go before the human element in games creation can be bypassed completely, agrees Seidel. The more likely outcome is that AI will be used in something like the same way as the industry has over the past few decades used “procedural generation” – where game elements such as the appearance of monsters and treasure are randomly created. After a lot of trial and error, games companies found this worked best when it was accompanied by designers “who kept evaluating and adjusting what the tools produced, and who continued to design the parts of the world that mattered most by hand”.

At the same time, AI could help the industry in two main ways. Firstly, it will help keep costs under control. With the typical cost of making a game having “risen over time from $50 million to $500 million”, anything that helps the industry “take a leaner approach to game development” will be good for developers, says Bowell. There could be particularly big time-saving efficiency gains when it comes to creating characters, environments and texturing. The use of large language models will also make the interactions between gamers and computer-controlled characters (NPCs) more “non-scripted and dynamic, which in turn will make games more interesting and replayable”, says Massimiliano Calamai, games director at Smallthing Studios.

Over time, the positive and negative aspects of the AI revolution will make “distinctive intellectual property and strong distribution even more valuable”, says Marc Fernandez, the chief strategy officer at Neurologyca, which tries to produce AI that better understands context. The big winners will be studios with “valuable IP, engaged communities, and the ability to turn adaptive, personalised worlds into long-term player engagement”.

We look at some of the most promising investments to profit from all these trends below.

The best gaming investments to buy now

Take-Two Interactive (Nasdaq: TTWO) owns Rockstar Studios, the company behind the successful Grand Theft Auto franchise. Gamers are eagerly anticipating GTA VI, the latest instalment in the franchise. It is “an example of exceptional intellectual property that can help sell software, move hardware and command the culture”, says Greg Weller of Generation Media. The company also owns games studio 2K, which has several successful franchises, and mobile developer Zygna, which allows it to also benefit from the boom in mobile gaming. Take-Two has a strong record, with revenue nearly doubling between 2001 and 2006. The stock trades at a reasonable 21 times expected 2028 earnings.

If Take-Two is a growth story, then Ubisoft (Paris: UBI) is about value. The company has faced many challenges and has struggled with sales and profitability, says Matthew Dolgin of Morningstar. But with rival Electronic Arts now a private company, Ubisoft is the best option for those who want to invest in a traditional games company with multiple large franchises, which include the Assassin's Creed and Far Cry series. Ubisoft looks cheap on multiple valuation metrics, trading at less than half the estimated value of its net assets.

CD Projekt Red (Warsaw: CDR) is an example of just how volatile the fortunes of games companies can be. It has struggled since the release of a hotly anticipated game resulted in mixed reviews. Its sales and share price are now well below pandemic peaks. Development delays have also been a problem. However, the company still makes money from licensing the brand rights to its hit series of Witcher games and is preparing several big releases in the next few years, including Witcher 4 and Cyberpunk 2077 II, which should substantially boost revenues. The stock trades at 25 times estimated 2027 earnings.

Sony (Tokyo: 6758) is not a pure play as it only makes about a third of its sales from games and related services, with music and entertainment systems also being major sources of revenue. The importance of gaming to the firm is only set to rise, however, following its decision to partially spin off its financial services business. It sells games hardware, most notably the PlayStation (which includes a VR headset), as well as its own software. Some of its game franchises, most notably the post-apocalyptic drama The Last of Us, have also become successful TV series. The stock trades at 16 times expected 2028 earnings.

Nintendo (Tokyo: 7974) is a games company with a long pedigree. It still produces a regular stream of new titles and hardware (most recently the handheld Switch 2) and it has also been finding new sources of revenue. Nintendo has been working harder to make money from its major franchises outside gaming. The Super Mario Galaxy Movie, for example, has already made more than $1 billion at the box office and a major new film based on The Legend of Zelda series is due out next spring. The stock trades at 21.5 times projected 2028 earnings.

One smaller UK-listed company worth looking at is Everplay (Aim: EVPL). Everplay has three businesses, including German developer Astragon and Storytoys, which produces educational apps for children between the ages of two and eight using licensed IP. The big business is Team 17, which publishes independent games such as Worms and Wardogs. The latter recently sold a million copies on the first day of its release. Everplay has an impressive record of monetising the IP of the developers that it works for, says Sean Kealy of Panmure Liberum. Revenues have more than doubled between 2020 and 2025 and the stock trades at only 10.3 times projected 2027 earnings.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

“We really do earnestly believe AI could kill all humans!”, says Anthropic researcher Evan Hubinger on X. He thinks there is more than a 10% chance we could all be dead “within the next decade”. Talk of AI's apocalyptic potential is in the air. Senior figures at AI labs are said to be terrified by the capabilities of recent models, which could be used to create bioweapons or elude human control and go rogue. The panic hit a new level at the weekend when a group of AI CEOs, including Dario Amodei – Hubinger's boss at Anthropic – publicly backed calls for a slowdown in AI development.

Anthropic: the AI company at the centre of a media storm

Scary stuff. Perhaps Silicon Valley really has been seized by a collective spasm of conscience about the consequences of building AI. Or maybe this is a PR campaign so slick and devious that it would make Alastair Campbell weep. For one thing, the timing is highly suspicious. Anthropic, the AI startup at the centre of the current media storm, is preparing to launch the largest initial public offering (IPO) in history in a matter of weeks.

A basic tenet of critical thinking is to pay attention to vested interests. Dario Amodei is not a neutral commentator. He is trying to secure an IPO valuation of up to $2 trillion. Yet much of the tech media, keen for a dramatic story, uncritically treats the self-serving pronouncements of AI executives as objective statements of fact.

The more conspiratorially minded pointed out that the tweet by Jacob Coxon, the Anthropic researcher whose resignation triggered the latest news cycle, was viewed 140 million times despite the fact that his account had almost no prior activity. While Coxon's concerns are probably genuine, the way his message was picked up and boosted by multiple influential figures appears less than organic.

On the face of it, it's not obvious how dire warnings about existential risk would be beneficial to the AI industry. Big tobacco spent years suppressing information about the dangers of its products. So why are AI executives so keen to talk about how their technology could be used by terrorists or lead to the extinction of swathes of white-collar work?

Is Anthropic doom trolling?

The answer, as Cal Newport, a computer science professor at Georgetown University, argues, is the pervasive use of a marketing technique he calls “doom trolling” (a spin on phone addicts' “doom scrolling”). Fear sells. Outrageous claims about the dangers of large language models (LLMs – the currently favoured AI technology) go viral, generating vast amounts of free media coverage for the company that originated them. Talk of existential risk makes AI products appear hugely powerful and desirable.

This buzz helps to distract from the less exciting reality. Yes, LLMs can do impressive things in highly structured domains such as coding and translation, where clear failure conditions help limit their tendency to go off the rails. In other areas (including journalism), their catastrophic tendency to make up information greatly circumscribes their usefulness.

In short, the LLM is a new software category, but investors are not going to pay trillions of dollars for a newer version of Microsoft Excel. Instead, these tools must be imbued with a dark, apocalyptic glamour. Such doom-mongering is longstanding industry practice. As Parmy Olson notes on Bloomberg, in 2019 OpenAI said it would hold back its GPT-2 model from general release on the grounds that it was too dangerous – this for an LLM that struggled to answer primary-school-level reasoning tasks. Terrifying indeed.

This year the AI doom campaign has been turned up to max. Barely a week goes by without claims (all originating from within the AI companies themselves) that a bot has gone on a rogue hacking spree, CEOs, including Anthropic's Dario Amodei, have backed a slowdown in the development of AI or that a new model can't be released because it creates serious cybersecurity risks (it is then released shortly afterwards anyway). The effect has been to generate precisely the sort of frenzied atmosphere that one would want to surround a trio of high-stakes AI-linked IPOs: SpaceX in June, Anthropic scheduled for October, and OpenAI sometime next year.

Who benefits from AI doom trolling?

The AI industry's calls for regulation carry the whiff of “regulatory capture”. Government red tape is more burdensome for upstarts than it is for big established players. New safety regulations could help AI leaders throttle the competition. There is persistent suspicion that Anthropic's Amodei would like to see regulations that effectively excludes his main competitor – cheaper, open-source, often Chinese AI – from major Western countries. If you can't beat them, ban them.

It is also possible that calls for a slowdown represent an attempt to put a brave face on the fact that the AI arms race is becoming too expensive. OpenAI is on course to spend $45 billion this year alone on training and inference (the cost of running AI), but the performance of new models is running into diminishing returns. That isn't an ideal backdrop in which to successfully list a growth company. Instead of breaking the bad news to investors and tanking the valuation, why not piously tell the media you are choosing to slow down development because of your abundant love for humankind?

Don't fall for the hype

What does all this mean for ordinary investors? For starters, don't fall for the tech hype machine. Steer clear of this year's big flashy IPOs, a crowded trade if ever there were one. Secondly, diversify widely. Continue to look for the sort of underexplored investment themes that we cover in depth. And finally, keep your head. There is no knowing how long the current AI fever will last, nor exactly how wide the damage will be when it breaks, but you can at least regain some tranquillity by tuning out the endless talk of doom. Perhaps AI really will kill us all in some hypothetical future. But for now, it is the AI-marketing hype that represents a clear and present danger to our collective mental health.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

The Chinese economy is facing headwinds. There is weakness in consumer spending, its property market is still in a slump, and there’s weakness in business investment.

Speaking on the latest episode of MoneyWeek Talks, Dale Nicholls, manager of Fidelity’s China special situations fund says while there may be negative things to say about the country’s economy, there are also bright spots to look out for,

“In terms of the domestic business, things are relatively muted. But, as always there’s pockets of strength in certain areas.”

He pointed to some constituents of his fund which are producing good numbers, like high-end mall operators and some restaurant chains, and added that while there is a “relatively weak consumption market”, the firms that have the right business model are taking market share.

Artificial Intelligence (AI) is also an exciting area for the region. “The companies that are involved particularly with anything AI-related, there’s somewhat of a tech boom. Business is strong for the companies that are involved with that,” he said.

AI is a major growth area for the global economy, and the controversies around it are mirrored in China too. For example, the concerns around the extent of capital expenditure and the return on investment.

Nicholls said: “The big technology companies – the Alibabas, the Tencents of the world – the market’s reacted badly to them increasing capex, but this feels to me a little bit overdone.”

The investors who are scorning increased capex are not asking the right questions, Nicholls claimed, namely whether the firms have a reputation for bringing returns on their investment.

“A lot of these companies have pretty good track records of generating good returns, and the feedback we get from them on what the return they’re getting on their spend is pretty good,” he said.

Nicholls was confident that parts of the AI market in China still provide good value. “Some of those big tech names I would put in that category. They’ve been sold off but if you think about the businesses as the sum of its parts – things like cloud – you’re actually seeing accelerating growth now, so it’s definitely seeing things pick up.”

Is China winning the AI race?

Although the models produced by US-based firms are currently the global leaders in AI, Chinese models are giving them a run for their money.

“[China] has some of the most competitive LLMs (large language models) out there, but the market doesn’t seem to be giving them a lot of value for that. Particularly if you look at the standalone LLM companies listed in China, they’ve done quite well.”

Chinese models have already disrupted the Western AI market multiple times. The release of DeepSeek’s R1 model in January 2025 brought with it a lot of panic in the West as investors reacted to Chinese AI challenging Western models.

A similar panic was caused when Kimi K3, another Chinese AI model, caused panic was released in July.

“Kimi K3 is interesting because it’s quite different in terms of size relative to others, but the performance is right up there with global frontier models. It’s much bigger, and they’re pricing it that way as well. So not quite the levels of US models in terms of their pricing, but obviously much higher than the other open weight models that the Chinese have been offering.

“So I think it’s another indication of the innovation that’s happening on the ground.”

For more, watch the full episode of MoneyWeek Talks with Dale Nicholls in conversation with MoneyWeek’s Cris Sholto Heaton on YouTube – or listen on any podcast platform.

About the podcast

MoneyWeek Talks is a podcast that helps you unlock the secrets to financial success. Editors Kalpana Fitzpatrick, Andrew Van Sickle and Cris Sholto Heaton are joined by influential guests – from CEOs and entrepreneurs to economists and fund managers – to share their top tips on managing money, investing wisely and building wealth.

Subscribe to the MoneyWeek Talks podcast and get ready to make it, keep it and spend it with confidence. You can also watch the episodes on our YouTube channel.

What is a data centre?

A data centre is a big, highly secure building full of computers and the other physical infrastructure that the digital world is built on. They've been around for decades (since the 1940s, reckons IBM, in the form of “server rooms”), but the internet, cloud computing and now the widespread adoption of AI have all radically increased the need for giant – and now “hyperscale” – centres. Nowadays, much of what we do online every day – from Google searches to banking to streaming Netflix – depends on computing power and connectivity housed in data centres. There's no single authoritative figure for the number of such centres in the world, because there's no universally accepted definition of how big an installation has to be to count as one. But in terms of large data centres, commercially operated by companies such as cloud providers, there are around 11,000–12,000 worldwide, with the majority in the US.

What's inside a data centre?

A typical data centre will contain servers (powerful computers that process, store and distribute data); networking equipment (to connect the servers to the internet); sophisticated cooling systems (otherwise computers get very hot, and then stop working); and power infrastructure (such as substations, back-up generators and batteries). In recent decades, data centres have got much bigger, with so-called “hyperscalers” – the likes of Microsoft, Google and Meta – building vast campuses covering many acres. A data centre with 150 racks holding 25 servers each equates to 3,750 servers; but that's no longer considered large. According to the International Data Corporation, a true “hyperscale data centre” is one that contains at least 5,000 servers and occupies at least 10,000 square feet of physical space and uses more than 100MW of energy. And many are now vastly larger than that. A hyperscale data centre can hold thousands of racks and hundreds of thousands of servers – all built on vast sites covering hundreds of acres.

Why are data centres controversial?

Much media coverage of the backlash (or “techlash”) against data centres focuses on distrust of Big Tech and AI, and fear of job losses. More than half of Americans say they're more concerned than excited about the growing use of AI in daily life, according to a recent report from the Pew Research Centre. But the dominant sources of opposition are more mundane, says The Economist. Polls suggest that what worries people about the centres is their perceived local impact. In a recent Fox News poll, 75% of respondents cited concerns such as energy use, utility and water bills, construction disruption and traffic as the main reason they opposed a data centre in their area. Just 11% pointed to AI itself.

The politics around building data centres

More than 500 US counties and municipalities have passed data-centre bans or moratoriums, while dozens of proposed projects have been abandoned. Research group Data Centre Watch calculates that grassroots groups blocked or delayed at least 75 data centre projects worth about $130 billion in the first three months of 2026 alone. That's close to the $156 billion worth of facilities that were disrupted during all of 2025. What's striking, too, is how swiftly politicians have shifted their position. The Republican governor of Texas, Greg Abbott, once touted the state as the “epicentre of AI development” – it boasts the most of any state except Virginia. Last week, though, now up for re-election, he said data centres had “dug their own grave” by overreaching. Across the US, the issue has emerged as a surprise key battleground in the US midterm elections, with some unexpected opponents (Republicans) and supporters (union-backed Democrats) of development. With typical grace and tact, president Donald Trump opined that: “The only reason that communities throughout the USA should not want data centres is if they want to end up being backwards and poor. If they want to be successful and rich, with far lower taxes and jobs all over the place, let data reign.” China is delighted with the US backlash, he claimed. “If we kill the golden goose, you will only have yourselves to blame.”

Is Donald Trump right about data centres?

Perhaps. “Voters and politicians alike are making a big mistake” in blocking data-centre growth, says The Economist. One common myth is that data centres guzzle endless water. In fact, a mid-sized centre uses about as much as two golf courses and much less if it recycles its water, as many now do. Concerns over energy use are more credible, but there's little evidence that they raise energy prices for households; rather, by increasing demand they spread the fixed costs of infrastructure among more kilowatt-hours sold. And in any event, the “answer to competing demands on resources – whether water, electricity or something else – is to price them according to their scarcity, thereby encouraging more provision and higher living standards over time. It is not to ration inputs according to the arbitrary judgment of scolds.”

What about here in the UK?

Britain has a claim to being the home of the first data centre – in the large shed at Bletchley Park where wartime cryptographers built Colossus, the world's first programmable electronic computer. Today, the UK has 450 “large data centres”, according to government figures, with more than half of them in or close to London. In September 2024, the government designated data centres as critical national infrastructure and in January 2025 announced five “AI Growth Zones”, with £28.2 billion of planned investment. Our biggest data centre to date, says James Price on CapX, is a 148MW site outside Cardiff, but progress is slowing, with the likes of the UAE and Japan now steaming ahead. This, of course, is hardly a surprise. “In Britain, we can force £100 million to be spent on a bat tunnel; imagine what our planning system will do with a gigawatt campus.”

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Law Debenture (LSE:LWDB) is a unique investment trust. It is part equity income portfolio and part professional services company, all in one high-performing £1.7 billion wrapper.

It has been a stalwart of the MoneyWeek Investment Trust portfolio since September 2015, with the second-best performance of all the trusts selected. Over the past ten years the shares have produced a total return of 263%, versus 130% for the FTSE All-Share index. It has also outperformed its benchmark by a wide margin over one, three and five years.

This performance is by far and away the best of any trust in the UK equity income sector. The secret of its success is down to the independent professional services (IPS) business.

The success of Law Debenture’s IPS

The IPS business provides a selection of relatively mundane, but essential services focused around three groups: pension services, which provides pension trust services across the UK; corporate trust, a provider of trust and escrow services for transactions such as bonds and mergers and acquisitions (M&A); and corporate services, which provides company secretarial and entity management services.

Inflation-linked organic growth has been complemented by acquisitions – most recently the purchase of company secretarial unit Konexo in 2024 – have all helped IPS grow both top and bottom lines.

These businesses grew net revenue by 6% in the first half of 2026 – the ninth consecutive year of mid-to high-single-digit growth.

These businesses are relatively specialist, so they are difficult for customers to do in-house. They are also low-margin and only profitable at scale, giving incumbent providers a significant edge.

There are only a handful of peers: the four public comparables include London-listed JTC, which is in the process of going private.

Dividend support

IPS accounts for only 15% of Law Debenture's net asset value (NAV), but over the years, its profits have helped support around a third of the trust's dividend distributions to investors.

That gives James Henderson and Laura Foll, the managers of its equity portfolio, a huge advantage for their strategy.

While other UK equity income trusts need to prioritise owning dividend-paying shares to support their distributions, they can take a more flexible approach.

For example, in February 2020, Law Debenture built a position in Rolls-Royce, which then had to suspend its dividend during the pandemic. Other income trusts might have been forced to sell, but Henderson and Foll held on. The position has since generated a return of more than 1,100%.

Other examples include Marks & Spencer and Babcock. Both of these have been held for capital gains, despite their poor dividend credentials.

A strong first half

The portfolio – which accounts for 85% of NAV – is built around growth, income and value. There is also a higher percentage of smaller and medium-sized companies than you might usually find in a UK equity income trust, which reflects the solid foundation provided by IPS.

For example, in the first half, positive contributors in the investment portfolio included companies viewed as beneficiaries of AI, such as Ceres Power and Infineon Technologies. Henderson and Foll also added positions in LSEG and Relx, two previous winners that had sold off due to concerns about the threat of AI.

Strong performers during the period included Beazley, Senior, International Personal Finance, Schroders and Tate & Lyle – all of which received takeover offers, which is “further evidence of the valuation opportunity available in UK equities”, say the managers.

The trust's own shares have also been trading well. For the six months to the end of June, they produced a total return of 16.2%, as they moved from a 2.5% discount to a 1.7% premium on top of the underlying NAV return.

The quarterly dividend was increased by 6% to 8.875p. Based on the company's current target for the year, the shares currently yield around 2.9%.

Henderson is set to retire in June next year, with Foll taking over the portfolio as lead manager. Given her ten years' experience at Law Debenture – as well as 12 on the Global Equity Income team at Janus Henderson – it should be business as usual when she assumes sole command.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

“What should I do if there's a AI crash?” a friend asked me recently. It is a very sensible question – we don't know there will be an AI crash, but having a clear plan to follow when you start to worry is better than waiting and panicking.

However, it's also a very difficult question, because the AI theme is such a huge part of the market: tech is over 35% of the MSCI World index (once you allow for firms such as Amazon and Alphabet assigned to non-tech sectors), while the trillions of AI capital expenditure is also buoying other sectors.

My first suggestion is to look at what wealth preservation trusts such as Capital Gearing (LSE: CGT), Personal Assets Trusts (LSE:PNL)and Ruffer Investment Company (LSE: RICA) hold.

These have diversified portfolios intended to cushion a market downturn, while still achieving growth. You could put some of your portfolio directly into these trusts, or you could look at how they allocate to cash, bonds, gold and other assets such as infrastructure as a template.

Even if you are a growth investor who is comfortable with high volatility to earn higher long-term returns, portfolios like these give you ideas for temporarily reducing risk that may be better than holding cash.

If you prefer open-ended funds, Orbis Global Balanced stands out for an active approach with more of a bottom-up value philosophy than most multi-asset funds.

Hedge against an AI crash with value stocks

If you want to stay entirely in stocks yet still dial down risk, you need to consider what kind of stocks are not caught up in the AI boom and may sell off less or rebound more quickly.

Think about this top down – by region (eg, UK and Europe) or sector (eg, pharmaceuticals and financials). Or you could look for value-focused stockpickers who favour other sectors.

That said, keep in mind that a European industrial that makes power equipment held in a value portfolio may still be a play on data-centre construction.

So it is difficult to anticipate how widely any pain from an AI crash may spread.

The most value-focused global trust is AVI Global (LSE:AGT), while most UK trusts have a value bias.

Among open-ended funds, Ranmore Global Equity has consistent returns from a portfolio that is very different to a typical global fund.

Some niche funds to consider

A third option is to look at very niche strategies whose medium-term returns should hopefully be unrelated to the AI-heavy global index.

Majedie Investments (LSE:MAJE) is now centred around such investments. It's an interesting holding in its own right, while looking at its strategy may help shape your own.

There are many specialist investment trusts and funds such as BH Macro (LSE:BHMG), BioPharma Credit (LSE:BPCR), BlackRock Frontiers (LSE:BRFI), Nippon Active Value Fund (LSE:NAVF) and Rockwood Strategic (LSE:RKW) or Polar Capital Global Insurance.

However, picking such funds is an approach for experienced investors who clearly understand what they are buying.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

The Nutshell Growth Fund invests in quality stocks from exceptional businesses, but only when their valuations offer an attractive prospective return. Our concentrated portfolio of around 30 global companies is selected for two characteristics that do not always come together: exceptional financial quality and a reasonable price. By quality stocks, we mean businesses with a strong record of revenue and profit growth, resilient margins, attractive returns on capital and substantial cash generation. Quality alone, however, is not enough. A wonderful company can still be a poor investment when too much future success is reflected in its share price.

Three quality stocks for your portfolio

Adobe (Nasdaq: ADBE) provides software tools used to create and manage digital content, including Photoshop, Illustrator, Acrobat and Premiere Pro. Its products are vital to the daily workflows of designers, marketers and large companies, creating high switching costs and strong customer retention. Its subscription model provides predictable recurring revenue, high margins and substantial free cash flow. Adobe can reinvest this cash into product development while continuing to return capital to shareholders.

The concern is whether generative AI strengthens Adobe's product suite or allows cheaper competitors to erode its position. We believe Adobe's established customer relationships, proprietary content and ability to integrate AI directly into widely used products give it significant advantages. Importantly, the market is no longer placing a premium valuation on those strengths. Adobe's earnings multiple has fallen as investors have focused on the competitive threat from AI. We believe much of that risk is now reflected in the price. Adobe does not need to return to its former valuation: continued moderate growth, resilient margins and strong cash generation should produce an attractive prospective return. Management have backed their confident outlook by announcing a significant share buyback programme earlier this year – further support for the idea that the share-price weakness is overdone.

Auto Trader (LSE: AUTO) operates the UK's largest digital automotive marketplace, connecting car buyers with thousands of vehicle retailers. Its scale creates a powerful network effect: buyers visit because it offers the broadest choice of vehicles, while retailers advertise because that is where the buyers are. This makes its market position extremely difficult to replicate. Auto Trader also benefits from a capital-light business model, high margins and strong cash conversion. It does not own the vehicles listed on its platform; instead, retailers pay for advertising, data and digital services. The shares have weakened due to concerns about relationships with dealers and the impact of AI on online search. We believe these concerns underestimate the value of Auto Trader's brand, audience, inventory access and proprietary market data. Its reduced valuation offers investors the opportunity to own a highly profitable and cash-generative franchise at a reasonable price.

Amphenol (NYSE: APH) makes the connectors, cables and sensors used across data centres, communications networks, industrial equipment and aerospace. These components represent a small proportion of a system's overall cost, but they are critical to its performance and reliability. Customers value technical expertise and consistency over choosing the cheapest supplier, supporting long-term relationships and attractive returns. Demand is supported by investment in AI infrastructure and data centres. Amphenol is not conventionally cheap on a headline earnings multiple. However, relative value does not simply mean buying the stocks trading on the lowest price-earnings (P/E) ratios. We assess valuation relative to the durability of growth, cash generation and the opportunity to reinvest capital. We believe Amphenol's exceptional execution and potential for growth justify a higher multiple.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Insurer Admiral Group (LSE: ADM) is among several firms which earlier this year saw their share price slump because of fears that AI-powered rivals could capture most (or all) of their business. However, since then many of these stocks have bounced back, with investors deciding that such fears are overhyped.

Admiral Group's shares fell by 14% in January after US firm Lemonade, which uses AI to process claims, launched a cheap policy for self-driving cars. While the policy was aimed at US consumers, it fuelled fears about AI being used to undercut traditional insurers.

Investors also fretted that the better driving record of autonomous vehicles compared with those steered by people could reduce the need for car insurance. Some analysts, such as AJ Bell's Dan Coatsworth, wonder whether car insurance will eventually be purchased by car manufacturers rather than by individual drivers.

How Admiral Group is using AI to cut costs

Yet even if such fears come true in the very long run, it's worth noting that full self-driving for individual cars (as opposed to a relatively small number of taxis currently on the streets) is at least a decade away from mass adoption. In any case, Admiral Group has itself been using AI and digitisation to cut costs and give it an advantage over its main rivals.

Earlier this year, Admiral Group also bought Flock, a technology firm it had been working with. The purchase gives it full access to, and ownership of, Flock's technology, which uses AI and telemetry (the process of collecting data from remote sources and passing it to a receiving system) to judge how well people are driving.

Meanwhile, Admiral Group has been taking steps to diversify its business by branching out into household, travel and pet insurance. While these areas currently make up only a small proportion of overall profit, they are growing at an extremely rapid rate, which should improve the group's medium-term prospects.

Meanwhile, sales almost tripled between 2021 and 2025, and are forecast to keep growing over the next few years. While profits have been more volatile, they have increased since 2021. Admiral boasts strong margins, with a double-digit return on capital employed. This has allowed the group to raise dividends to record levels. The stock's valuation also looks attractive at 15 times expected 2027 earnings and a dividend yield of just under 5%.

Admiral Group's share price has plenty of momentum behind it, having beaten the overall UK market over the last one, three and six months. It is trading well above its 50- and 200-day moving averages, and has also been one of the best performers in the FTSE 100 over the last six months. I suggest that you go long at the current price of 3,772p at £1 per 1p. I would put the stop-loss at 2,800p, which would give you a total downside of £972.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

“Baby-faced” billionaire Yang Zhilin recently dealt the biggest shock to Western markets since DeepSeek, wiping hundreds of billions of dollars off the valuations of AI and chip stocks.

Kimi K3, developed by Yang’s Moonshot AI, is the most advanced “open-weight” large language model to emerge from China yet, topping many benchmarks with its capabilities “at a third of the cost”.

At a stroke, notions of Silicon Valley's technical dominance have been swept away, with its developer, Moonshot AI, challenging the likes of Anthropic and OpenAI at the frontier – prompting questions about their mega-valuations.

Moonshot's founder has a good story to tell too, says The Telegraph. Yang Zhilin is a prog-rock devotee who named his firm in honour of Pink Floyd's The Dark Side of the Moon, seemingly in tune with Western ideas and culture.

The 34-year-old has built a “mythology” that has “helped distinguish Moonshot from China's otherwise austere AI industry”, says the Financial Times. Employees describe an intense culture of long hours in Moonshot's headquarters in Beijing's Haidian district. “But Yang has also infused the company with his own... obsession with rock music... A white piano stands prominently in the office.”

Yang Zhilin heads to China's MIT

Yang Zhilin was born in 1992 and grew up in Shantou in the southern state of Guangdong – China's industrial heartland. Former classmates recall that he was always “unusually gifted”. Having started coding in high school, he won first prize in the National Olympiad in Informatics, earning him direct admission to Tsinghua University, often dubbed “China's MIT”.

Even in that specialised atmosphere, he was known as “Yang the genius” because of the way he managed to balance elite academic performance with his musical interests. He was a drummer in a campus band called Splay, organised music competitions and gained a reputation for being “romantic and idealistic”. Some reports suggest that he switched his undergraduate degree from thermal engineering to computer science, having been inspired by a Haruki Murakami novel.

In 2015, Yang Zhilin completed his PhD at Carnegie Mellon University, where he studied under AI gurus Ruslan Salakhutdinov and William Cohen, worked at Google Brain and Meta, and founded a retailer-focused start-up, Recurrent AI, before returning to China in 2019.

In 2023, he co-founded Moonshot AI with Tsinghua University classmates. “Recurrent AI taught him how to woo investors and scale a business,” says The Telegraph. “But he learned painful lessons too.” Moonshot's early years, when he attempted to build a “Chinese-first version of ChatGPT”, were marred by a lawsuit from investors in Recurrent AI that saw him hauled before the Hong Kong International Arbitration Centre before a settlement was reached.

Moonshot's chatbot Kimi was an instant hit, but also “plagued by outages and... overtaken by larger rivals”, says the FT. Rival DeepSeek's successful R1 model was another “existential test”. Yang Zhilin returned to the laboratory to focus on model training. He also made the pivotal decision to make Moonshot's AI models available to developers globally.

Moonshot's open approach has enabled China to “cast itself as a champion of low-cost, open-source AI”, says The New York Times. But that openness has raised concerns in Beijing about “the potential threats the technology might pose” to Communist Party rule. “[He] may be wise to moderate his views,” says The Telegraph. “Other tech billionaires in China have found... that the Party likes its economic champions on a short leash.”

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Semiconductors are the world's most valuable manufactured good. These tiny, intricately engineered pieces of silicon can perform more calculations in a second than a single person could complete in 30,000 years. This year has brought a semiconductor boom for the ages. The US PHLX chip index has nearly doubled over the past 12 months. Investors, noticing that big US tech firms are planning nearly $1 trillion in spending on AI data centres next year, followed the money to the chip stocks that provide AI hardware.

The global semiconductor supply chain is very concentrated. A handful of manufacturers and designers – Taiwan's TSMC, South Korea's Samsung and SK Hynix, America's Nvidia – capture the lion's share of profits. Yet expectations have run ahead of reality. This week, SK Hynix reported a 557% surge in operating profit, with margins of more than 80%. That Midas-like profitability still wasn't good enough for investors in Korea, who sent the shares tumbling 19%. The Korean Kospi slumped 11% on Tuesday and a further 6% on Wednesday. America's Nasdaq 100 technology index has fallen 9.7% from its peak, says Eva Roytburg for Fortune.

The immediate trigger was talk of new competition from China, where chipmaker CXMT listed on Monday. Those fears are probably overdone – China still doesn't have access to the cutting-edge extreme ultraviolet lithography machines required to make the world's best chips. But the chip stock selloff isn't irrational; for months, the “going trade” has been to sell the hyperscalers – firms such as Microsoft and Meta that appear to be overspending on data centres – and “buy the semis”, companies such as Samsung that are profiting from Silicon Valley's profligacy. Now investors have realised the obvious contradiction: if Big Tech's AI investments really are as wasteful as they think, then at some point that spending will be cut, which would be a disaster for chip stocks, too.

The semiconductor boom is based on very real profits, says Moses Sternstein for a16Z. Rising earnings have come alongside falling valuations – an unusual symptom for an alleged bubble. Micron, whose earnings are poised to rise 60% year on year, trades on a mere six times forward earnings. The wider US semiconductor complex trades on about 21 times forward earnings, a slight discount to the five-year average of 23.8.

The semiconductor industry is infamously cyclical

So are semiconductors cheap? In one sense, yes, but the industry is infamously cyclical. An acute shortage during the pandemic turned into a big bust in 2023 as demand returned to normal levels. “Investors are wondering whether semis can keep it up” this time. As laptop buyers will be well aware, dynamic random-access memory (DRAM), which is used for computer memory, is in acute shortage this year.

Samsung and SK Hynix have joint plans to invest as much as $1.5 trillion to double Korea's DRAM output within five years, say Song Jung-a and Michael Acton in the Financial Times. Yet there are risks of the chip cycle turning again. Should AI demand disappoint or Chinese supply surge, there could be a glut as soon as 2028.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Technology giants with an edge in artificial intelligence (AI) have dominated equity market returns in recent years, but enthusiasm is slowing down with chip stocks looking jittery as competition heats up.

There have also been warnings of a potential AI bubble, but as yet, the jury remains out over which companies will emerge the longer-term winners or losers. A recent survey by fund management group Natixis Investment Managers revealed that despite a number of global headwinds – ongoing US-Iran conflict, volatile energy markets and persistent inflation – 91% of the 33 strategists interviewed were optimistic that AI will be a driving force behind market performance in the second half of the year. It also found 88% expect the AI sector to accelerate with just 12% believing its bubble will burst in the second half of the year.

But should investors be seeing that disruption as an opportunity?

How to invest in AI beyond ‘big tech’

There are two distinct strategies the AI wave opens up. One is to aim to capture the growth potential of AI but without limiting yourself to the big names, such as the ‘Magnificent 7’ – Apple (NASDAQ:AAPL), Microsoft, Amazon, Alphabet (NASDAQ:GOOGL), Meta, Nvidia (NASDAQ:NVDA) and Tesla.

BlackRock’s Helen Jewell, international chief investment officer of fundamental equities, believes by looking at the AI story through a wider lens, investors could discover more upside with less of the volatility that comes with high valuations and market concentration.

One route is to look at infrastructure and the power investment needed to facilitate the AI boom and the broader shift towards electrification it has helped accelerate. Jewell said this trend is being “turbocharged” by governments focusing on energy independence.

“These sectors may offer exposure to structural growth trends while potentially providing more diversified return streams, attractive valuations and lower concentration risk than some of the most highly valued areas of the market,” she added.

Where are the next big opportunities in global equities?

In 2025, a handful of sectors led market gains – namely banks, aerospace and defence, and industrials.

All three areas are expected to continue to perform positively, as valuations are increasing. European banks in particular look promising; BlackRock's Jewell said they’ve shown resilient earnings despite interest rates calming down from recent highs.

She added that banks are increasingly adopting AI to modernise their own systems. Better integration across the European banking and capital markets system, alongside consolidation indicates a more profitable sector and, therefore, better likely returns for shareholders.

How to invest in contrast to AI

Another way to play the AI theme is in reverse. Concentration risk presents a problem if too high a share of your overall investments are gathered in one stock, region or sector – hence the ‘don’t have all your eggs in one basket’ analogy.

If there’s a correction in AI, and share prices fall (or the supposed bubble bursts), being exposed to different areas of the market that aren’t correlated will offer investors a degree of ballast to their portfolio.

Jewell cited healthcare as a strong diversification play, though it is also a sector that can benfit from AI. The sector has historically traded at a premium to the market but is now at a 15% discount, with earnings growth that has been second only to technology.

Elsewhere, she likes Latin America, which also has a low correlation to the AI trade. It’s trading at lower valuations than historical average, and while it makes up just 0.8% of the MSCI All Country World Index (ACWI), it accounts for 7% of global GDP.

In the UK, the FTSE 100 index has outperformed global stocks on a total return basis, without any direct AI exposure. Broadly, rising interest rates over the past five years and higher energy prices have boosted banks and oil companies, while defence has also returned to prominence amid the ongoing conflicts around the world.

The three income stocks picked below demonstrate the diversity of opportunities in the Aberdeen Equity Income Trust portfolio and the combination of income and capital growth that we look for.

The trust takes a deliberately index-agnostic approach, searching for companies undergoing change that the market under-appreciates. The idea is simple: the most compelling opportunities are often found in overlooked or under-researched areas. This leads to a portfolio that looks very different from other traditional UK equity income strategies.

With no sector constraints and a flexible approach to size, the trust can access a broader universe of income stocks, many offering attractive yields and the prospect of dividend growth. As the businesses gain wider recognition, valuation re-ratings can follow, supporting capital appreciation. With the macro backdrop starting to improve and investor attention moving beyond the FTSE 100, this approach is increasingly relevant.

Three income stocks to consider

We have held Chesnara (LSE: CSN) since 2014, reflecting our long-standing confidence in its business model. It operates as a disciplined acquirer of legacy life insurance assets, completing £440 million of acquisitions over the past five years. As large financial institutions streamline operations and dispose of non-core assets (often at attractive discounts) and private equity focuses elsewhere, Chesnara has carved out a niche by pursuing overlooked deals and refusing to overpay. This disciplined approach has built a strong record of earnings-accretive transactions. At the same time, the steady flow of acquisitions replenishes the book as older policies run off. Scale has increased meaningfully, with assets under administration rising from £8.5 billion to more than £20 billion, alongside expansion into Europe. With more than £100 million of available firepower, management sees further opportunities ahead. Chesnara generates cash through efficient management of existing books and delivers investment returns above the risk-free assumptions embedded in its actuarial models.

GTT (Paris: GTT) is a global leader in containment systems for liquefied natural gas (LNG), a market set for structural growth. Demand for LNG is expected to rise by around 60% between 2025 and 2040 as economies transition away from coal, driving the need for additional tanker capacity. GTT's membrane technology is critical to the safe transport of LNG, and decades of research and development have secured it a dominant market position. Barriers to entry are high, with shipowners and insurers reluctant to risk unproven suppliers, thus supporting pricing power and consistently high margins. Core growth should benefit from increasingly global LNG flows and a replacement cycle for an ageing tanker fleet. GTT is also building a digital services platform, with technology already installed on more than 15,000 vessels. This creates a valuable opportunity to cross-sell software and consulting services – an area that remains under-monetised, but offers high returns. A new CEO may accelerate this focus, while robust cash generation underpins both dividends and reinvestment.

The sharp correction in software stocks in early 2026 created an opening for investors hunting for income. UK IT reseller Softcat (LSE: SCT) plays a key role in connecting businesses with complex IT, partnering with more than 200 global technology providers. It has delivered consistent organic growth, expanding market share and securing a highly loyal customer base – 95% of revenues come from repeat business. The rapid adoption of AI is driving demand for processing power, storage, networking and security infrastructure – areas where Softcat is well positioned. This structural tailwind is expected to support continued earnings growth for the business, resulting in rising dividends.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

This interview is an excerpt from Andrew’s recent filmed MoneyWeek Talks podcast episode with Jeremy, available on all podcast platforms and YouTube.

Jeremy Grantham, originally from Yorkshire, is the co-founder and long-term investment strategist of asset management group GMO, based in Boston. Jeremy’s reputation is based on his long-standing ability to spot bubbles. He called the Japanese bubble in the late 1980s, and rightly refused to rush into the tech bubble in the late 1990s, a strategy that earned him notoriety as a permabear. But he also turned bullish in March 2009, just as the market bottomed after the crisis.

These episodes, and a great deal else, are chronicled in his memoir, The Making of a PermaBear: The Perils of Long-Term Investing in a Short-Term World, published earlier this year.

Andrew Van Sickle: Jeremy, we learn in your book that you weren't always a patient value investor. I enjoyed the section about the late 1960s. You say you became a “gunslinging nitwit” in an expensive market, but you came a cropper with a couple of stocks. Will you tell us a bit about that episode and how it became a formative moment?

Jeremy Grantham: Just after I got my first job in the investment business, I came up to Boston and joined this lunch club of gunslinging kids, fresh out of business school. Every time we met, someone would have a hot story. And typically these stocks would go up and come down very quickly. One was American Raceways, a motor-sports group that had Stirling Moss on the board. It was going to introduce Formula One to the US. I thought it would work. It seemed desperately American: power, noise, blood and death.

American Raceways bought one track in the middle of the country and everyone showed up. Expecting the races to catch on nationwide, I bought 300 shares at $7. I went to England and Germany for three weeks to get married; we came back and the stock was at $21. So I like to say that I did what any good value manager would do. I sold everything else I had and tripled up. I had 900 shares, a lot of them on borrowed money, at $21. And fate always teases you and wants to get you fully committed to a bull market, so the price was $100 by Christmas. All I had to do was sell and run. While my wife and were deciding whether to make a higher bid on a house we had our eye on, the market started to break and pretty soon American Raceways was slumping. And I scrambled out and got into another company equally far ahead of its time (Formula One is now doing well in the US).

This company was going to put a monitor on everyone's desk, which back then was hugely high-tech. And on this little screen they were going to have the option price of individual stocks. But the idea, a forerunner of Bloomberg terminals and the internet, didn't catch on at that stage, and the company never took off. It went belly up and I was lucky to scramble out with enough money to pay the banks back. From then on I thought I'd better revert to my Yorkshire instincts and be a cheapskate and a value manager.

Andrew Van Sickle: Having had your fingers quite badly burnt, you decided to take a very thorough look at stock market history and how human nature always ended up in these situations. And you constructed by hand, as part of your research, the first index for small-cap stocks?

Jeremy Grantham: Yes. We had to go back into the archives and put together an index going back to 1925. And what we saw were long periods of small-cap domination and long periods of large-cap domination. They ebbed and flowed in these multi-year cycles. And the interesting thing to me was that we were in a big “nifty-fifty” blue-chip cycle and small stocks' valuations had become extremely depressed relative to the rest of the market. And so we put 100% of our money into small caps, which was so original as to be totally unique back in the day. Institutions didn't dabble in small caps back then. They were beneath contempt.

And so we had a very strange portfolio that was difficult to sell. There were 99 rivals selling Coca Cola and there was one of us selling companies that no one had ever heard of. So it was at least entertaining. It amused the clients rather than anything else.

Andrew Van Sickle: So the idea was to seek out investments that people had overlooked, and you clearly enjoyed the number crunching – later your asset management group was one of the first to use a computer to keep doing so, wasn't it?

Jeremy Grantham: Yes, it was painfully expensive, filled the whole room and created a lot of heat. But it did give us a little advantage for a year or two. And what we found, by the way, was that the numbers we'd hand-crunched were pretty accurate. And what was nice about hand-crunching numbers was that no one else did it. Whereas once we got a computer, everybody else did. Pretty soon, a computer was simply a cost of doing business.

And no one made a killing by having one, you just had to have it and pay for it, whether you liked it or not. This is, incidentally, getting ahead of myself, very analogous to AI. In five or ten years AI will be a cost of doing business. But it won't be a way you get ahead. It will be a case of falling irretrievably behind rivals if you don't use it. New technologies confer an advantage on the early adopters. And when it's clear that they have an edge, everyone copies them and it goes away. But we'll come to AI later.

Andrew Van Sickle: The notion of an early lead being eroded brings us to the issue of mean reversion, one of the principal themes to emerge from your research. What goes up must come down. This applies to corporate profits, which get competed away. Similarly, asset markets get euphoric, human nature being what it is; we overdo things on the way up and on the way down.

Bubbles blow up and burst, but one can never really tell when things will revert to the mean. What did it feel like in late Japan and late 1990s America, knowing that you were right to be bearish – because mean reversion is something unavoidable – but standing practically alone for years on end?

Jeremy Grantham: Well, it gave you lots of time to do more research, particularly in 1998 and 1999. We didn't start to lighten up on US stocks until the end of 1997, when the trailing price/earnings (p/e) ratio reached 21, the same as at the peak in 1929. So it was an emotional point to reach. By the end of 1998, we were as light as we could get. And the p/e went all the way up to 35 by the peak. Japan was even worse, of course: the mother and father of all bubbles. The p/e had never eclipsed 25, and in 1989 it soared to 65.

Of course 35 wasn't 65, thank heavens, or we would definitely have gone out of business. But conveniently at 35 the market beat a magnificent retreat and we were positioned for it, having doubled and redoubled our ante until there was nothing left to do, and we actually made good money.

To give you an example of how much of a bargain value stocks were around the market peak, consider real estate investment trusts (REITs). Properties were selling at a discount to replacement cost, and Reits were yielding 9.1%. The overall S&P 500 yielded just 1.5%, a record low. When the overall market slumped, Reits jumped by 30% amid a flight to safety and value.

Andrew Van Sickle: For your business it sounds like a race against time, with clients no doubt increasingly frustrated that you were sitting out the big technology-led gains of 1998 and 1999. They would have had the same perspective as Chuck Prince, CEO of Citigroup, in the credit bubble in 2007: “[As] long as the music is playing, you've got to get up and dance.” Do you think that if the bubble had burst a year later, you would have gone bust?

Jeremy Grantham: Yes, I think so. Incidentally, coming back to Mr Prince, George Soros' take was that “Actually, the music had stopped, he just hadn't noticed,” which was typically cruel.

Andrew Van Sickle: Did lots of people you spoke to at the time think it was a bubble, too, but just didn't want to say so publicly?

Jeremy Grantham: That's exactly right. We were a purely institutional firm dealing with lots of big pension funds. The hired guns in the pension funds usually had a lively understanding of how risky the market was. But their committees – made up of private equity and VC investors, and all manner of people who'd made lots of money and were very, very confident – insisted on going with the flow, and that anyone who didn't was stuck in the past and should be fired.

The upshot is that the uncertainty surrounding the timing of bubbles is greater than the typical client's patience. And that is all you need to know about institutional investing. Most of the engine-room players saw the bubble. It's just that the marketing people and the bosses realised that scepticism was not a good business strategy.

If you are a big firm, you simply mustn't bet on the bursting of a bubble. You have to go with everybody else, you have to be willing to run off the cliff, and you have to be willing to be professional and slick and quick, saving some money on the way down and redeploying it. If you do that, you will thrive. If you try to fight the bubble, well, you may get lucky, you may win one. We, in a sense, won the great financial crash. We explained it in quarterly letters. We prepared for it and we got out in a timely fashion and everything worked well. But if you get it wrong, watch out.

Keynes, just about my solitary hero in the economics business, said the key to investment life is never be wrong on your own. So you can be wrong in company and you don't lose your job. Even being right on your own, he said, was dangerous in that they would pat you on the head if you won, but then describe you as an eccentric when you'd left the room. That's not a great reputation to have. And he said that if you're wrong on the downside, if the market doesn't break and you were positioned for a bear market, “you will not receive much mercy”.

Andrew Van Sickle: Turning to the bubble of the moment, what is your take on AI and the market's view of it?

Jeremy Grantham: In 100 years they'll be writing about this point in stock market history as they write about the South Sea bubble. It is simply magnificent. The SpaceX prospectus was of the order of the famous South Sea bubble equivalent: “An undertaking of such profound importance but cannot at this time be revealed.” Just give us your money.

Andrew Van Sickle: I was horrified to read that Isaac Newton went for the South Sea bubble. He should really have known that what goes up comes down.

Jeremy Grantham: He said something along the lines of: “I know a lot about the movements of heavenly bodies but nothing about human nature.” The SpaceX prospectus is unbelievable – mining asteroids and colonies on Mars and moving through space and a projection of revenue streams, 90% of which seem to relate to AI. And it's not clear, of course, that SpaceX's version of AI, which is at the moment having its bottom kicked around the block by Anthropic and the rest of the boys, is going to be around in the long term. Talk about tulips.

Andrew Van Sickle: You've said that you think excitement over the advent of AI essentially stopped the bubble of late 2021 deflating fully.

Jeremy Grantham: Yes, December 2021 met all the conditions of a bubble, we thought, and there was a slump in 2022. But for the first time in history, halfway through a bubble breaking, you come out with an idea that is so colossal and accompanied by so much capital expenditure that you change the game.

On that infamous day in late 2022, ChatGPT appeared and someone rang the bell and said: “All change.” And the rest of the market didn't believe it for ten months, drifting down a bit. But by then the Magnificent Seven had doubled, and it dragged the rest of the market up.

I used ChatGPT, telling it to please summarise War and Peace in ten points, and then do it in German. And that was enough to make me realise that this was going to be impressive. It's clearly better than anything else other than the railroads.

People don't realise that the more obvious and important the idea, the more likely you are to attract too much capital and have a capital bust and a market bust. The railroads transformed our lives, they added enormous productivity, and yet they were so obviously going to do that, that everyone built too many railroads, and everybody lost their money. And that will happen in AI.

At present there are seven companies plus another 15 snapping at their heels. All are aiming at the same market, AI. The one that gets there first, they believe, has a licence to make more money than you could shake a stick at. More than anyone has ever made on anything. And they are all saying the main risk is not spending enough. We will spend our vast cash flows, they are saying.

They're going to fight until someone survives. This could be the most vicious fight to the end that we have ever seen, starting now. In that sort of fight, they do not make lots of money and the stocks get crushed. And then they emerge out of the wreckage. Like the internet. Amazon declined 92% in the tech slump. And yet it then rose and inherited the Earth. The railroads rose from the ashes. And this will rise from the ashes.

Andrew Van Sickle: While investors take bets on who might survive, where are you finding value outside the US?

Jeremy Grantham: That was an easier question to answer at the beginning of last year. At that stage, valuations looked unremarkable and therefore priced to make a decent return. Since then, the S&P has gone up another 23% but that is nothing like the rest of the world, led by emerging markets. Emerging markets are up 60%. And value in Europe is up 45%. And these are very big gains over the S&P. The rest of the world now begins to look a tad overpriced, while the US has moved into “read all about it in 100 years” territory.

Andrew Van Sickle: We get the impression that Japan is certainly no longer cheap, although it's probably reasonable.

Jeremy Grantham: Yes, reasonable, and it's done very well. And again, it has handsomely beaten the S&P since the start of last year. The main thing that impresses a lot of people, however, is that even though the S&P may have been at the back of the pack in the last 18 months, it still went up handsomely, and so it's created the impression that, therefore, it defies the pull of gravity.

And this feeling that it will go on forever is, of course, absolutely classic. That's exactly how people wrote and thought in 1929 and how they wrote and thought in 2000. This time is not different.

Of course, people hate you if you say this. They are so involved in making money they loathe the notion that the whole thing is a mass delusion. Perhaps they hated me more in 2000, but it's getting close. In the comment section of a recent podcast, three people said that my ears were big. And of course they are big. It's just that people don't usually say so after the age of seven or eight.

Andrew Van Sickle: I don't suppose the boy pointing out the emperor was naked was very popular either.

Jeremy Grantham: I don't know. The episode wasn't recorded.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Investors in Japanese stocks – like investors everywhere – need to be alert to concentrated exposure to the AI story. Over the past few months, it has gone from important to indispensable.

Wherever you look, the stocks that are doing best are linked to AI demand, while those that are AI-agnostic or an “AI loser” are mostly lagging. The emerging markets index is now trading like an AI play, due to TSMC, Samsung Electronics and SK Hynix, but this is by no means the only example.

Among Japanese stocks, the top performers are firms such as Kioxia, a chipmaker that soared from ¥1,455 when it floated in December 2024 to a high of ¥112,700 in June. Other big hitters include tech conglomerate Softbank and firms involved in chipmaking and testing such as Advantest, Murata Manufacturing and Tokyo Electron.

Implausible-sounding companies such as food-seasonings firm Ajinomoto and toilet manufacturer Toto have also been carried along: their core businesses make them leaders in materials that play a role in the chip-supply chain.

Japanese stocks have made a strong start to the year

“The result has been an unusually narrow, yet powerful market,” note Alex Bowles and Brett Moshal of the Japan equity team at asset manager Orbis. As of the end of June, the Topix index has made a strong start to the year (up 19%), yet only a third of Japanese stocks have beaten the benchmark. A basket of 67 AI companies accounts for 14 percentage points of that return.

This has been a headwind for any investors underexposed to AI, although Bowles and Moshal argue that it is also creating contrarian opportunities. They point to Nintendo, which has halved amid fears of a memory crunch hurting hardware sales in the short term, but also the threat that AI poses to its competitive advantage in game development. This is overdone given the strength of Nintendo's intellectual property, they argue.

Foreign investors held back by a weak yen

Diversification between regions may not be much protection if the AI boom ends badly. That said, for now, the market is still doing well, and the drag for foreigners is the currency.

The yen keeps weakening and now stands at ¥163 to the US dollar and ¥218 to the pound. There has been little sign of this bottoming out, notwithstanding talk of “appropriate and bold action” by the finance minister this week. The result is that Japan has become one of the cheapest developed-market countries to live in, note Jim Reid and his team at Deutsche Bank. In purchasing power parity terms, with price levels measured on the basis that the US is 100, Japan now comes in at 60; in 2012 it was at 125.

In theory, the yen is deeply undervalued. Yet this has been near-consensus and it keeps sliding. Our exchange-traded fund (ETF) portfolio is invested in Japan through Vanguard FTSE Japan (LSE: VJPN) and this has done fine, but clearly a currency-hedged ETF would have done better. We are sticking with the unhedged position since we expect the yen to rally eventually – but iShares MSCI Japan GBP Hedged (LSE: IJPH) or UBS Core MSCI Japan hGBP (LSE: UB0D) are other options to cut the risk it falls further.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Taiwan Semiconductor Manufacturing Company (TSMC) (Taipei: 2330 and NYSE: TSM) may be the most important business most people have never heard of. Right now, you're probably carrying products that it has made. Most consumers recognise names such as Apple and Nvidia. Yet behind many of the products they sell sits a Taiwanese manufacturer responsible for turning their designs into reality.

Every day, billions of people rely on devices powered by chips produced by TSMC. The company's influence stretches far beyond smartphones. From artificial intelligence to consumer electronics, much of the modern digital economy ultimately depends on a business with headquarters on an island roughly 100 miles off the coast of China.

What makes TSMC remarkable is not simply its scale, but the way it achieved it. Unlike most technology giants, it did not become dominant by creating the best consumer products or developing a monopoly over software. Instead, it positioned itself as a neutral supplier to an industry filled with fierce competitors. In effect, TSMC became the Switzerland of the semiconductor world, doing business with everyone and doing so in secrecy.

How Morris Chang founded TSMC

That strategy was the brainchild of Morris Chang, a veteran semiconductor executive who spotted a flaw in the industry's business model and built an entire company around solving it. Nearly four decades after it was founded, his insight sits at the centre of the global technology industry.

Chang never set out to build one of the world's most important firms. For 25 years, he worked at Texas Instruments, rising high to run its global semiconductor business. During those years, Chang noticed a problem. Brilliant engineers regularly designed innovative chips, but turning those designs into products required vast sums of money.

In the 1970s and 1980s, semiconductor firms were expected to do everything themselves. Designing chips was only half the job. Companies also needed expensive factories, specialised equipment and the expertise to run them. The result was an industry dominated by a handful of large, vertically integrated firms.

Then Chang's own career took an unexpected turn. In 1983, aged 52, he was passed over for the top job at Texas Instruments and left the company. After a brief spell in a senior role at another American chip company, he received an unusual offer. The Taiwanese government wanted to build a domestic electronics industry and was looking for someone with Silicon Valley experience to lead the effort. Chang accepted.

He arrived in Taiwan with decades of semiconductor experience and a conviction that copying America would be a mistake. Taiwan lacked the design expertise, customer relationships and global brands needed to compete. But Chang had spent years watching another problem unfold. The industry was full of talented chip designers who could not afford to manufacture their ideas. What if somebody built the chips for them?

That simple question led to the creation of TSMC in 1987. At the time, the idea looked absurd. Bringing a chip to market required access to a fabrication plant, or “fab”. The industry believed serious companies should own these factories themselves. In practice, that meant chip designers relying on one of the industry giants.

That created another problem. The company manufacturing your chip was often also a competitor. Handing over your most valuable intellectual property required a leap of faith. Chang's solution was that TSMC would make chips for anyone willing to pay, but would never design products of its own.

In 1987, that sounded like madness. When Chang went looking for investors, many of the industry's biggest names rejected him. Texas Instruments and Intel both declined his offer. A factory without its own products looked like a recipe for bankruptcy. How could a manufacturer survive without guaranteed demand?

In the end, Chang persuaded the Dutch electronics group Philips and several wealthy Taiwanese families to back the venture. Even then, enthusiasm was limited. Philips largely viewed the investment as a way of supporting the Taiwanese government's ambitions rather than as a compelling commercial opportunity. It intended liquidating its investment early. Potential customers were hardly more enthusiastic. Many designers saw little reason to outsource manufacturing. A company that only made chips for other people seemed unnecessary.

By now, Chang was a 56-year-old executive pitching an untested business model in an industry convinced it could never work. Then, fortune presented an opportunity. In 1988, Intel found itself short of manufacturing capacity. Faced with the prospect of disappointing customers, it reluctantly turned to TSMC for help. Intel's engineers arrived in Taiwan expecting a low-cost, unsophisticated subcontractor. Instead, they found a world-class operation run by one of the industry's most experienced executives. Passing Intel's quality standards was not easy, but once TSMC secured the American giant's approval, attitudes across the industry changed quickly. If Intel trusted TSMC, others reasoned, perhaps they could too.

That endorsement transformed the trajectory of the company. Designers no longer needed to spend billions building factories before launching a new product. Instead, they could focus on what they did best – designing chips, and letting TSMC handle the rest. Without TSMC, it's unlikely that Nvidia could have existed, nor could a host of other chip companies.

A new generation of semiconductor firms emerged, freed from one of the industry's biggest barriers to entry. While rivals competed to design better chips, TSMC focused on becoming the best manufacturer in the world. By choosing not to compete with its customers, the company turned neutrality into a competitive advantage. That decision would prove far more powerful than anyone imagined. But the success of TSMC's model created an obvious question: if it was such a good idea, why didn't somebody copy it?

Many tried, but almost all failed. For years, Samsung looked like the most credible challenger. The South Korean giant had deep pockets and decades of manufacturing experience. The problem was that Samsung was also a competitor. Unlike TSMC, Samsung sold smartphones and consumer electronics under its own brand. That created a dilemma for customers. Why hand your most valuable chip designs to a firm that might one day compete against you? No customer wrestled with that question more than Apple.

During the early years of the iPhone, Samsung made many of Apple's processors. The arrangement worked, but it became increasingly awkward as the two companies emerged as fierce rivals in the smartphone market. By the early 2010s, they were fighting a series of patent disputes. Apple found itself in the strange position of relying on one of its biggest competitors to make some of its most important components.

TSMC offered an escape route. With the launch of the A8 processor in 2014, Apple shifted production to Taiwan. The move was risky, but Apple concluded that the benefits outweighed the costs. TSMC's neutrality had become one of the most valuable assets in the technology industry. Today, many of Silicon Valley's biggest rivals manufacture their chips at TSMC. Apple, Nvidia, AMD and Qualcomm all rely on the same company, despite competing aggressively in their own markets.

Samsung's problem was a conflict of interest; Intel's was something different: success. For decades, Intel dominated the semiconductor industry. Its factories were among the most advanced in the world. However, the company became increasingly focused on its own products. When Apple approached Intel in the mid-2000s about supplying chips for what would become the first iPhone, Intel declined.

Management believed the opportunity was too small to justify the investment. It was one of the most expensive misjudgements in the history of the technology industry. By the time Intel recognised its mistake, Apple had moved on and TSMC was becoming the manufacturing partner of choice for a new generation of chip designers. When Intel later attempted to open its factories to outside customers, its manufacturing systems had been built around Intel's products, not the needs of third-party designers.

Other competitors couldn't keep up with the investment needs. GlobalFoundries, an American rival, spent years trying to keep pace before effectively giving up on leading-edge manufacturing in 2018. The company concluded that each new generation of chip technology required so much investment that the returns no longer justified the risk.

China's national champion, SMIC, faces a different challenge. Western export controls have restricted access to advanced manufacturing equipment, making it difficult to compete at the industry's frontier.

TSMC's greatest advantage

TSMC's greatest advantage is not its technology, because that can be copied. Its real advantage is the business model Morris Chang created nearly four decades ago. The company sits at the centre of the semiconductor industry, serving customers that often compete with one another. That position generates enormous scale, which in turn funds the next generation of factories and equipment.

The most advanced chips require ultraviolet lithography machines built by the Dutch company ASML. Each cost more than £275 million. A state-of-the-art fab may contain dozens of these machines, helping to push the cost of a new facility beyond £15 billion before production even begins. That creates a problem for potential rivals.

Customers will not trust an unproven manufacturer with their most important products, especially if they don't have advanced fabs. Yet building a state-of-the-art factory requires billions of pounds before those customers appear. Having already achieved enormous scale, TSMC now largely escapes this trap. The company controls roughly 92% of advanced chip manufacturing and generates the cash needed to fund the next generation of technology.

In 2026 alone, TSMC expects to spend nearly £45 billion on new factories and equipment. Few companies in the world could contemplate spending that much. None can do so with the same confidence of earning a return. The result is a powerful feedback loop. Scale attracts customers. Customers generate cash. Cash funds new factories. New factories attract even more customers.

Every year that cycle turns, TSMC becomes harder to catch as the price of entry rises ever higher. That scale gives TSMC another advantage: it allows customers to help fund its expansion. Most manufacturers have to build factories first and hope demand follows. Today, TSMC often works the other way around. Some of its largest customers commit billions of pounds years before new facilities begin production, effectively helping to finance the next generation of capacity.

At the end of 2024, TSMC held more than £7.3 billion of customers' deposits. As production ramped up on newer technologies, some of that money was recognised as revenue, but the balance remained substantial. Technology companies are willing to tie up enormous sums because access to TSMC's manufacturing has become critical to their own growth plans. This arrangement shifts much of the risk away from TSMC.

When companies such as Nvidia sign long-term agreements worth billions of pounds, they provide “visibility” – confidence in management forecasts – that few industrial businesses can match. New factories can be built with a high degree of confidence that demand will be waiting when they open. That helps explain why TSMC can continue investing through industry cycles.

AI is a game-changer for the semiconductor industry

For years, Apple was the company's most important customer. The iPhone generated the predictable demand that allowed TSMC to refine successive generations of manufacturing technology and steadily expand its lead. Now a new force is reshaping the industry. AI has become the biggest driver of demand for advanced semiconductors. Training and running large AI models requires vast quantities of computing power, creating an arms race among technology companies desperate to secure enough chips.

The biggest beneficiary has been Nvidia. In 2025, Nvidia overtook Apple as TSMC's largest customer, generating more than £18 billion of revenue for TSMC and accounting for roughly a fifth of total sales. The shift says a great deal about how quickly AI has altered the economics of the technology industry, but the opportunity extends beyond chip design.

Producing cutting-edge AI processors is one of the most demanding manufacturing tasks in the world. The chips themselves are larger, more complex and more difficult to assemble than those used in smartphones. As demand has exploded, bottlenecks have emerged throughout the supply chain. For TSMC, that has translated into even greater pricing power.

The world's largest technology firms are competing for a limited supply of advanced manufacturing capacity. Many have little choice but to accept TSMC's terms because there are few credible alternatives. AI has reinforced the advantages of specialisation. Developing a leading-edge AI chip already costs hundreds of millions of pounds. Building the factory to make it would require billions more. As AI pushes the technological frontier forward, the advantages of specialisation are becoming even more pronounced.

But TSMC's dominance creates a problem. Most of the world's most advanced semiconductor manufacturing remains concentrated in Taiwan. That has become a concern for governments, particularly as tensions between China and Taiwan have intensified. A disruption to TSMC's operations would ripple through the global economy.

Smartphones, data centres, AI systems and countless other technologies depend on its chips. Under pressure from the US and other governments, it's begun expanding overseas. The largest investment is a vast complex in Phoenix, Arizona. Similar projects are underway in Japan and Europe.

Building advanced factories in the US is estimated to be roughly 50% more expensive than doing so in Taiwan. Labour costs are higher, experienced engineers are harder to find and supply chains are less developed. TSMC has reportedly had to transfer experienced staff from Taiwan and create thousands of new operating procedures to support its US operations. Yet even these higher costs have not weakened the company's position.

Customers are willing to pay a premium for chips manufactured on US soil. For many, securing a politically safer supply chain is worth the extra expense. In an ironic twist, efforts to reduce dependence on TSMC have largely demonstrated how dependent the world has become on its expertise.

The future looks bright for TSMC

Whether the company can maintain its current position forever is another question. The semiconductor industry has a long history of dominant firms losing their edge, while geopolitical tensions surrounding Taiwan remain an ever-present risk. Governments are spending heavily to build alternative sources of supply and rivals continue searching for ways to close the gap.

However, history suggests writing off TSMC would be unwise. For nearly 40 years, the company has repeatedly adapted to changes in technology, customers' demands and the structure of the industry. It has survived downturns, outlasted competitors and continued strengthening its position at the heart of the digital economy. The story of TSMC is ultimately the story of how a company became indispensable. In an industry defined by relentless change, that may be its most remarkable achievement.

None of this means TSMC is a bargain. Investors are well aware of the company's strengths and the shares have performed exceptionally well over the past decade. As a result, the stock trades on a valuation that reflects high expectations for future growth. Still, TSMC has qualities that are difficult to find elsewhere. It occupies a dominant position in one of the world's most important industries, enjoys deep relationships with many of the largest technology companies on the planet and continues to invest heavily to maintain its lead.

Most importantly, investors do not need to predict which company will ultimately win the AI race. Whether the future belongs to Nvidia, AMD or some future challenger, there is a good chance that their chips will still be manufactured by TSMC. That does not guarantee attractive returns from today's share price. But betting against the company has rarely been a profitable strategy. For investors seeking exposure to long-term growth in technology and AI, TSMC remains one of the highest-quality businesses in the market.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

In May 2025, Dario Amodei, the CEO of AI company Anthropic, said that the technology his company is helping push forward could drive unemployment up to 10%-20% in the next one to five years and wipe out half of all entry-level white-collar jobs, as Josh Tyrangiel points out in The Atlantic.

Jim Farley, the CEO of Ford, has estimated that AI will eliminate half of all white-collar jobs in a decade. Sam Altman of OpenAI has opined that it is just a matter of time before we see a billion-dollar company staffed by just one person.

That the advent of a new technology has given rise to predictions of disastrous consequences is hardly new. But that the prophets of doom come not from the ranks of the usual suspects, but from the makers of the new technology and those most in a rush to adopt it, is.

So, are they right? AI is clearly already transforming work, says Tyrangiel. Companies including Meta, Amazon, Walmart, and JPMorganChase have recently announced lay-offs due to “automation”.

Three academics from the Stanford Digital Economy Lab have found that entry-level jobs that are exposed to disruption from AI have already seen a 13% decline since late 2022. So the transformation may already be under way, even if it's too early to be sure (other factors could explain the decline and the evidence is sparse and mixed).

If that transformation unfolds slowly and the economy adjusts quickly, then we may, as economists reassure us, be fine, or even better off in aggregate. But if AI instead triggers a rapid reorganisation of work, compressing years of change into months, affecting roughly 40% of jobs worldwide – as the IMF projects – then the consequences could be huge.

Is AI actually any good for us?

Which will it be? Let's remember that humanity has been automating work for 250 years, as technology analyst Benedict Evans has pointed out. History shows that every wave of automation has destroyed whole classes of jobs and created new ones. The process may be painful for some, but over time and in the aggregate the result has been greater prosperity.

Two concepts from economics give us confidence that this time is unlikely to be different. The first is the “lump of labour fallacy” – the misconception that there is a fixed amount of work to be done and that if some work is taken by a machine then there will be less work for people. But if it becomes cheaper to use a machine to make a pair of shoes, say, then the shoes are cheaper, more people can buy shoes, and they then have more money to spend on other things, and we discover new things we need or want, and new jobs get created.

The second concept is Jevons Paradox. In the 19th century, the Royal Navy ran on coal and people worried about what would happen when the coal ran out. Don't worry, said the optimists: steam engines are getting more efficient, so they'll use less and less coal. Not at all, said economist William Stanley Jevons: if we make steam engines more efficient, then they will be cheaper to run, and we will use more of them and use them for new and different things, so more efficient steam engines means we will use more coal.

That paradox has been at work in relation to white-collar work for a long time, says Evans. In the 1880s, typewriters and carbon-copy paper meant that clerks could produce more than ten times the output of the days when they copied out documents one at a time by hand. The result for clerical employment? Far more clerks were hired. If one clerk can do the work of ten, then perhaps you might want to do more of the work that clerks do – more analysis, or manage more inventory, say. You might build a different and more efficient business that is only possible because of the new technology.

It was the same story when, much later, digital spreadsheets were introduced that could do at the click of a button what might previously have taken a whole team of accountants all week. Employment for accountants went up.

The most recent study into what AI is doing to jobs seems to confirm that this is indeed what is happening this time, as Noah Smith reports on Substack. A study by Ara Kharazian, Lisa Simon and Ryan Stevens, researchers at US technology start-ups Ramp and Revelio Labs, examined private data to determine what happens when companies start using generative AI. The answer is that they hire more humans. The number of entry-level jobs rose, too. So it seems that AI is “still mostly a complement to human labour rather than a substitute” for it, says Smith. For now at least, AI is “behaving pretty much like a normal technology”.

AI is just software

That's the usual pattern, and if AI did indeed start to progress at the rates feared and with the consequences predicted, it would be “unprecedented in human history”, says The Economist. New technologies have never spread fast enough to make large numbers of people unemployed for long periods of time because the diffusion of the technology always proceeds slowly.

To see why that is unlikely to be different this time, remember that AI is just software, as Tyrangiel points out. And the thing about software is that “people hate it almost as much as they hate change”. Before AI can transform a company, it has to access data and be woven into existing systems. A “trade secret of most Fortune-500 companies is that they still run critical functions on lumbering, industrial-strength mainframe computers that almost never break down and therefore can never be replaced”. Integrating such legacy tech with AI would mean big changes involving lots of people with strong opinions about the “right” way to proceed. Meanwhile, months pass, then years – and “the CEO still can't understand why the miracle of AI isn't solving all of their problems”.

Indeed, the idea that “one magic piece of software” will change everything instantly and override all the complexity of real people, real companies and the real economy “sounds like classic tech solutionism, but turned from utopia to dystopia”, says Evans. The reality looks rather different, as Zeynep Tufekci shows in The New York Times. Firms that have experimented with fully automating functions such as customer service have been burned. The result has been scammers talking chatbots into handing over control of key functions, promising refunds or incredible deals, such as a new car for $1. The bot taking orders at McDonald's proved “wildly dysfunctional”.

The key thing to understand is that these incidents are not the result of errors, but of the technology functioning as it is designed to do. Currently existing AI technologies are “not reasoning machines” – they simply produce answers that are probable based on the data they've been trained upon. They have no common sense or intelligence. AI can “do many things with astounding efficiency”, especially if those things are formal and structured and can be tested and checked in real time. Most jobs are simply not like that and still require “good old-fashioned human intelligence”.

This doesn't mean the “job apocalypse” definitely won't happen, says The Economist. Maybe this time will be different. Perhaps the technology will transform in ways we cannot yet predict. If so, you may know the apocalypse by these signs: sharply rising productivity combined with weak real-wage growth in the US, the world's frontier economy. This would show up as an increase in GDP per person above the 2.5% upper limit that is the historical norm in frontier economies and a simultaneous jump in corporate profits, reflecting that the gains from higher output were flowing to capital, not labour. Another sign would be big job losses in lots of industries, showing up in a recession. Which jobs vanish in the next recession will “give a hint about the shape of the AI world to come”.

Is there actually any sign of any of this happening? Not really. The labour market “certainly is not cracking yet”, says The Economist. “The share of the OECD's working-age population with a job keeps breaking records, unemployment across the club of mostly rich countries is just 5%, and America employs more people than ever in ‘AI-exposed' industries, such as law.” American graduates have been struggling to find jobs since before the launch of ChatGPT fired the starting gun on the AI revolution in late 2022. Many economists foresee relatively little disruption ahead. Those at America's Bureau of Labour Statistics think the country will add 5.2 million jobs between 2024 and 2034, increasing total employment by 3%.

Robots can't do your job

There are broader reasons for scepticism. The heaviest users of AI have recently been scrambling to curtail its use as the cost of using it outweighs the gains. Surprisingly few people use the technology on a regular basis and the share of companies in the OECD that have adopted AI remains small (about 20% for the latter, although figures for both individual use and company uptake vary widely across different studies, depending on what is deemed to count.) The basic problem here is that most people just don't know what AI is supposed to do for them, as Evans has argued. There's a box you can type stuff into, and you get text in response. Often the text is roughly right, but precisely wrong. For how many people will that be life-changing? As Pablo Picasso perceptively saw in 1968, “Computers are useless. They can only give you answers.”

The likelihood is that AI will not so much replace jobs, as make certain tasks easier and quicker for some people. Generally, says Evans, jobs are a complex mesh of things that we might not even be able to explain explicitly. You may have a good idea of just why a chatbot is never going to be able to do your job, for example, but will be impressed if someone says that it can of course already do the job of a lawyer or a doctor. The blunt truth is we do not know just what is involved in jobs we are confidently predicting will be gone tomorrow, nor do we know how AI will change them, if at all.

What we should most fear is fear itself. A recent poll found that 70% of Americans believe that AI will reduce their employment opportunities, says Robert Shiller, also in The New York Times. That fear could in itself have economic consequences. When millions and millions of people make economic decisions based upon negative expectations, there is a risk that the fear can actually “help birth the reality”. The leaders of Silicon Valley should learn to do better than peddle alarmist narratives in the hope that the resulting media attention will highlight how powerful their latest AI model is. They will find it harder to sell their wares in future if the result is an economy paralysed by fear and recession.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Artificial intelligence (AI) could be embedded into every aspect of a financial services business by 2030 as millions of savers and investors are already making use of the tools, research by the Financial Conduct Authority (FCA) has found.

The City watchdog asked executive director Sheldon Mills to review how advances in AI could transform retail financial services.

The Mills Review, published this week, found one in five UK adults - equivalent to 11 million UK adults - are already open to AI making decisions for them in areas such as pensions and investments but there are concerns about trust and control.

The review found that while AI has the potential to improve access, personalisation and efficiency, it could also amplify risks associated with fraud, cybersecurity, consumer harm and market concentration.

Mills said: “Artificial intelligence will transform financial services by 2030. It creates significant opportunities for consumers, firms and the wider economy. This report sets out a roadmap for how industry regulators and government can prepare for the next phase of AI-driven change in our world-leading financial services sector.”

Here is how the FCA expects AI to reshape financial services.

Changing roles in financial services

The regulator suggests human roles in financial services will change.

It highlights that many firms are already piloting and rolling out AI tools and by 2030 they could be more independent and cover every function from customer support and underwriting to compliance, claims and product design.

AI may become the main method by which they process information, serve customers, and evidence outcomes, the FCA suggests.

This could mean the role of people within firms changes from operators close to each decision towards collaborators, approvers and, eventually observers who monitor outcomes and step in when systems move outside agreed parameters.

The FCA said: “This is a substantial organisational shift, requiring new skills and a clearer account of what human oversight actually involves.

“Firm governance will extend existing model risk management to cover more complex systems and deeper reliance on third-party providers. Successful AI deployment should lift productivity and support economic growth, though the benefits will reach consumers only where firms remain accountable and markets stay competitive enough to pass them on.”

The review suggests the human role becomes one of challenge, judgement and review rather than direct production of every output.

The rise of agentic AI

Consumers are increasing using AI applications to act on their behalf and automatically follow preset instructions, known as agentic AI, and the FCA predicts this could grow in financial services.

This may involve easier bank switching, embedding insurance into other platforms, auto-rebalancing in savings and investments and pension pot consolidation.

The FCA said: “Overtime, AI systems will move beyond offering information and recommendations towards trusted AI agents that can act continuously for consumers within agreed limits, providing ongoing financial management and optimising people’s financial lives.

"If done well, this could help consumers achieve more while doing less, addressing long-standing problems such as low switching, advice and protection gaps, and improving outcomes for people with lower financial capability.”

The FCA warns that consumers will still need to be able to oversee, understand and challenge AI-driven decisions, especially when things go wrong, the report adds: “Unequal access to high-quality applications risks widening inclusion gaps - but well-designed AI systems also present an opportunity to radically improve outcomes for those who need more support.”

Changes in market power

The rise of AI could reshape who holds the power in financial services.

Investors and savers may flock to well-known investment platforms or providers now but the FCA says AI has the potential to drive greater beneficial competition in financial services and to support new entrants.

This could make the suppliers more powerful and there are risks of dependance on a few technology firms.

The FCA said: “Control of the AI-mediated customer interface may become a major source of market power.

"As consumers rely on agents to search, compare and transact, the owner of that AI layer may influence which products are visible, how choices are ranked and where value is captured, shifting the customer relationship away from financial services providers.”

AI risks

While AI could help consumers manage their finance more effectively, the FCA review wants that there will also be more fraud risks.

The report said: “Deepfakes, synthetic identities and personalised social engineering are taking fraud and cyber risks into a new era and changing how fraud and cyber-attacks are conducted. Existing weaknesses can be exploited far more quickly than before, and defenders will need to keep pace.

“Defensive, supervisory and enforcement capability must evolve at least as quickly as the threat. To remain effective, firms, regulators and their partners will need access to many of the same AI capabilities as those used by attackers.

"They will also need to share the right information with those best placed to act, when it matters and before harm escalates.”

Is AI regulated?

Artificial intelligence isn’t regulated but Mills suggests that existing rules such as the Consumer Duty and Senior Managers Regime should cover some of the risks associated with how savers and investors may use AI.

The review does add that regulation may have to evolve though to focus on shared models between firms though rather than focusing on individual conduct.

It also suggests that the FCA review AI tools such as ChatGPT and Claude to assess if there are regulatory overlaps and risks in the results generated.

Commenting on the report, Amal Jolly, chief executive of the AI company Saturn, which specialises in financial advice, said: "AI brings new opportunities to close the advice gap, improving the financial lives of millions of adults, but as this report shows it also brings huge risks.

“In financial services, AI is the new Wild West: consumers are left with no protection. Only 9% of people have access to regulated human financial advisers, but 100% of people have access to ChatGPT and other AI platforms. This is not just a theoretical problem, but can cause real harm to people who are entrusting major life-changing financial decisions to unregulated AI.”

SpaceX has joined the Nasdaq 100, meaning passive funds that track the index will now automatically hold positions in the company, which listed on 12 June.

SpaceX (NASDAQ:SPCX) joined the index today (7 July), a week after it was added to the Russell 1000 Index (29 June).

Bloomberg reported SpaceX could look forward to an estimated $5.4 billion of inflows as a result of ‘forced’ buying by index funds that track these two indices.

Elon Musk’s space exploration company was fast-tracked for inclusion following rule changes by the index providers, put in place to reflect the unprecedented size of some initial public offerings (IPOs) coming to market.

Nasdaq’s new rules now allow freshly listed companies to be included in as few as 15 trading days, rather than its previous minimum period of three months after an IPO.

What will SpaceX index inclusion mean for flows?

Nasdaq says globally, there is around $1.4 trillion in assets tracking its component companies’ combined market capitalisation (market cap) of $31.5 trillion, around half of which do so through exchange-traded funds (ETFs). The other half is in derivative products, such as futures and options.

The Nasdaq 100 index represents the largest 100 companies, excluding financials, listed on the Nasdaq Stock Market. Often described as a tech-focused index, it contains all ‘Magnificent 7’ names – Alphabet, Amazon, Apple, Tesla, Meta, Microsoft and Nvidia. But it also contains many other companies with a value of $100 billion or more from healthcare, industrials and materials, for example, with representation across 10 of the 11 standard industry classification sectors.

When a stock joins an index like the Nasdaq 100, funds tracking that index are effectively forced to buy its shares so that they still reflect the index. This creates additional demand for a stock and could push up its share price.

The UCITS version of Invesco’s Nasdaq-100 ETF (LON:EQQQ) is the largest Nasdaq-tracking ETF available to UK investors. Barclays Smart Investor platform lists it as the seventh most popular purchase during the week of 26 June to 2 July.

Alongside the uplift from index fund inclusion, several investment banks have issued positive analyst statements on SpaceX, marking the end of the ‘quiet period’ that typically follows an IPO. Morgan Stanley, Goldman Sachs, UBS and Bernstein Research are among the names backing the stock with ‘buy’ recommendations or equivalent, based on asset strength and long-term growth prospects.

Baltimore-based Constellation Energy, a $90 billion company, generates electricity on a vast scale. And AI's voracious appetite means that electricity is now becoming a very valuable commodity, and the companies that can generate it reliably, cleanly and at scale will see a lot more attention than they're currently getting.

Investors have re-priced entire industries, assuming AI will transform the global economy. Electricity gets less attention, yet the chips, AI models and data centres are all useless without power. Vast data centres consume enormous quantities of power to train and run increasingly powerful models. Electric vehicles, battery factories, semiconductor plants, air-conditioning systems, industrial re-shoring and electrification more generally are all pulling in the same direction.

Tap into the great electrification with Constellation Energy

Constellation Energy (Nasdaq: CEG) owns the largest fleet of nuclear reactors in the US and has more nuclear power stations than anyone else at a time when hyperscalers are searching for reliable and cost-efficient power. Investors increasingly view Constellation less as a utility and more as the owner of scarce infrastructure – an essential asset – which explains its appeal as a long-term growth stock.

Constellation Energy currently generates about 10% of US clean energy. It produces enough electricity to power about 27 million US homes. It is the largest nuclear-energy company in the country and its power stations sit alongside a fleet of gas, hydro, wind, solar, geothermal and oil-fired assets. These produce 55 gigawatts (GW) of capacity.

To put that into perspective, the UK's recent winter peak demand for electricity was typically around 60GW, according to the National Energy System Operator. Constellation has 2.5 million customer accounts across the US and counts 80 of the country's 100 biggest firms by revenue among them.

Results in May showed first-quarter sales up 64% from $6.8 billion to $11.1 billion year-on-year. While impressive, roughly $2 billion-$3 billion of that reflected the acquisition of Calpine, a largely gas and geothermal power generator. Constellation Energy generated roughly $25 billion of revenue over 2025 as a whole. Earnings per share were $2.74 in the quarter, up 28% over the year and beating analysts' estimates by 14 cents.

The firm's nuclear fleet achieved an excellent 92.3% capacity factor, meaning its reactors were producing electricity at close to their maximum potential for almost the entire period. Management isn't seeing any slowdown in demand from the hyperscalers, with projected spending levels continuing to rise to reflect the growing need for computer processing.

Management has reaffirmed its expectation of 2026 earnings per share of around $11, up from $9.39 and $8.67 in 2025 and 2024 respectively. It's targeting 20%-plus annual earnings per share through to 2029 led by higher prices and rising demand, including improved long-term contracts.

Analysts have $13.50 pencilled in for 2027. Their 12-month share-price target is $362, about a third higher than now. Of course, in a world hungry for electricity, existing generation capacity may prove considerably more valuable than investors currently assume.

Don't chase the chips

The curious thing is that investors can currently buy a company expected to grow earnings by more than 20% annually at a valuation broadly in line with the wider market. Usually, investors are asked to pay a substantial premium for that combination of growth and strategic importance. The market's comfortable paying premium valuations for businesses that consume computing power. But it's a lot less interested in businesses that sell the vast amounts of electricity that makes such computing possible now and in the future. Look beyond that “utility” label and there is an opportunity to be had.

The immediate objection is that electricity is hardly scarce. If demand goes up, the power generators can build more capacity. But new power stations need planning permission, endless environmental reviews, financing, engineering expertise, political support and years of construction. And then transmission networks need upgrading.

This is where Constellation Energy's nuclear fleet becomes particularly interesting. Investors spend a great deal of time discussing technological moats. Yet there may be few barriers to entry that are more formidable than a collection of fully operating nuclear reactors. The market's growing interest in Constellation Energy reflects a simple reality: it already owns large-scale electricity infrastructure that is built, connected and delivering cleanly at scale. Building more is possible, but doing so quickly is another matter.

Nobody can yet say with certainty which company will dominate AI. What already seems clear, however, is that the modern economy wants far more electricity. Investors have spent the first phase of the AI boom chasing the chips. The second phase may belong to those pumping the power. Investors may find it easier to back the latter than gamble on the eventual winners of the AI race.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Dr Douglas Williams is a veteran senior executive and board member for several biotech companies. During his time at Biogen, ZymoGenetics, Amgen Immunex and Seattle Genetics, he was involved in the development of several multibillion-dollar treatments, including Enbrel, Tecfidera and Spinraza.

Matthew Partridge: The time and cost of clinical trials has been seen as one of the big stumbling blocks to the emergence of new drugs. Do you see any developments that could help speed up the process?

Douglas Williams: Anything you can do to speed up the process is beneficial – after all, time is money. I think the real benefits will come as we become more efficient at targeting better-defined populations of patients [those who meet highly specific and uniform criteria, thus making it easier to gauge the exact effect of drugs]. In a field with a high rate of failure, improving the chances of success by even a small percentage can have a huge impact on the bottom line.

Matthew Partridge: Have the recent changes at the US Food and Drug Administration (FDA), the regulator of federal health, helped or hindered the clinical-trial process?

Douglas Williams: It's all a bit chaotic at present. You're seeing reversals of long-standing policy with political interference in what should be scientifically driven decisions. The “willy-nilly” nature of the Department of Government Efficiency's (DOGE) cuts has also led to a brain drain at the FDA itself, with some of the most experienced staff leaving.

Matthew Partridge: The FDA is seen as a gold standard when it comes to getting drugs approved first. Could there come a day when drug companies start looking to regulators in Europe or elsewhere?

Douglas Williams: This is already happening at the front end of the clinical-trial process, where Australia has become a key destination for early phase-one studies [the first of three stages of clinical trials, when scientists test the safety of the drug]. The regulatory process there is relatively quick and streamlined, making it a cost-effective place to run studies. And Australia's consolidated healthcare system makes the process of finding patients and enrolling them in studies more efficient.

Similarly, I've been working with Chinese companies and the system's low cost of capital, overall efficiency and rapid trial process are remarkable.

Matthew Partridge: The Trump administration has been threatening tariffs on drugs. Do you see geopolitical issues as a major risk for the drug and biotechnology sectors?

Douglas Williams: I do think there is logic in wanting to onshore some of the manufacturing for crucial drugs. There has been this enormous migration offshore from the US on the manufacturing side. So to try to bring some of that back, certainly for vital components of key drugs, makes a lot of sense. But there are surely better ways to achieve this than tariffs, which are a blunt instrument.

Matthew Partridge: Do you think America's dominance of the biotech and drug sectors is at risk?

Douglas Williams: I think that it's very much at risk for a variety of reasons. There definitely need to be some major changes to the FDA to bring the regulatory regime closer to what's happening in China and Australia. Firms are moving to the latter for early-stage studies, although people will still want to enrol patients in later-stage trials in the US and Europe.

The other thing that's happening in the US that I worry about from a longer-term perspective is that the reduction in the National Institutes of Health's funding for basic science grants is chasing away a whole generation of PhD students and postdoctoral candidates. This is already starting to create a hole in the pipeline for talent, and the longer this goes on, even if it's just for the four years of the current administration, the longer it will take to rebuild.

The engine driving the innovation that creates new companies in the sector and allows for new intellectual property to be created and new breakthroughs to take place is being eroded.

Matthew Partridge: What should the UK do to make itself more friendly to biotech and pharma companies?

Douglas Williams: The UK can streamline the process of starting studies and enrolling patients quickly and easily through the NHS, and raising patients' awareness of the trials on offer. More specifically, it could learn a lot from what the Chinese have done around streamlining the rules governing which particular regulatory bodies you need to secure approval from to begin a trial.

Matthew Partridge: Turning to the wider sector, GLP-1 drugs are changing the way we deal with weight-loss, diabetes and perhaps other conditions as well. Do you think the firms that pioneered GLP-1s are going to be able to stay ahead of the competition, or will it be like the computer industry, where firms such as IBM were unable to maintain their control of the industry?

Douglas Williams: The biotech industry is based on the expectation that there will be a rotation of dominance, as patents only last a certain amount of time before rivals are allowed to produce generic versions of a drug, causing prices and profits to collapse. But until that happens, the first-movers in this area, such as Novo Nordisk and Eli Lily, will dominate it. They've also pursued new approaches for delivering the drug – shifting from injectables to oral tablets, for instance. So they're creating scope for multiple waves of innovation, which could extend their dominance.

However, biotech is ultimately all about building a better mousetrap: there are other young companies coming in that are attempting new methods that don't come with the side effects, such as muscle loss, that are associated with the GLP-1s, for instance.

Matthew Partridge: Are there any other big leaps forward that could take place in the next five years or so?

Douglas Williams: I find the work around the role of sleep in dementia and brain conditions very interesting, and the idea that deep sleep can help combat those conditions is certainly an elegant theory. Neurology, in general, has become much hotter from an investment perspective. I'm involved with several companies in the neuropsychiatry sector, including being chair of Draig Therapeutics, a Cardiff-based company developing treatments for major depressive disorder.

The analogy people have used is that neurology is going to become the next oncology, where the precision approach to well-defined populations of patients is going to dominate drug development. So, you'll essentially be treating slices of the populations that have a particular broad definition of a disease.

Matthew Partridge: What about advances in medical imaging, such as MRIs and CT scans?

Douglas Williams: During my career, I was involved in the early development of some of the first approved drugs in the amyloid reduction arena and what really turned the tide was being able to understand what these drugs were doing inside the brain; it's hard to do without some way of looking at the target. I think faster and more effective scanning technology has already fed back into neurology-drug development.

Matthew Partridge: Do you think in five years' time the range of treatments for neurological conditions, things like dementia, could be radically different?

Douglas Williams: Yes, there's so much activity in this area, a reflection of the problems posed by ageing populations.

Matthew Partridge: How is AI going to change drug development?

Douglas Williams: It depends on your definition of drug development. Taking the all-encompassing view, where a fully integrated company does the discovery, drug development, manufacturing and sales and marketing, it will change the whole process. One example is in manufacturing. Already, we can take real-time data from the bioreactors that are used to manufacture proteins, and AI can use this to tweak the process to make sure that you maximise productivity. There's an increasing amount of work now on using AI in the clinical-trial process, both in terms of designing the trials and dealing with back-office operations.

I've been in this field for 40 years, and it's remarkable what some of the young companies I'm involved in are doing with AI. I'm optimistic that in the next ten years, we'll start to see the impact of AI on designing new molecules and finding new drugs too.

Matthew Partridge: Do you think that there will still be a need for actual scientists, rather than AI alone, to be involved?

Douglas Williams: Without question. If you ask ChatGPT or Claude a question, how the question is worded matters a lot, and that's where the role of the scientist really shows itself – the better the question, the better the answer. So, there will always be a place for the human element in terms of driving science.

Matthew Partridge: Have investors in the sector learnt to tune out political noise and focus on the long-term growth story?

Douglas Williams: I think so. Stock market valuations have risen while mergers and acquisitions have proliferated, which is always healthy because it gives investors capital to put back to work. So a virtuous circle has developed. There's never been a more amazing time in terms of the tools that we now have for drug development, while our understanding of biology continues to expand.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Some analysts have proclaimed that AI is more important and will be more transformative for human development than the creation of the railways. With the world's largest technology companies now spending more than $1 trillion a year expanding their presence in the market, there's no doubt this theme will dominate for the foreseeable future.

However, as investors focus on the so-called hyperscalers – Alphabet, Microsoft and Amazon – which are building their AI infrastructure at an alarming rate, as well as chip manufacturers such as Nvidia and Micron, which are supplying the industry, not much attention is being directed to the infrastructure that will power this revolution. This is where bottlenecks are now starting to throttle growth.

The energy grid needs an upgrade to power AI

The critical one is the power grid. In the US, for example, the grid is around 50 years old and was not designed to handle the current level of rapid growth in demand. The graphics processing units, or GPUs, that underpin AI data centres today are vastly more energy-intensive than their previous counterparts.

Research compiled by Goldman Sachs and JPMorgan estimates that by 2027, AI server racks will require 50 times more power than the equivalents that formed the backbone of cloud infrastructure five years ago.

The computing power of any facility consumes only around 60% of the total energy requirement. The rest is taken up by cooling systems and other infrastructure.

As the hyperscalers expand, they are learning that Silicon Valley moves much faster than the rest of the world. GPUs have become 50 times more energy-intensive over the past five years, but global energy output has risen by just 1%-3% per year. In the real world, it can take five to seven years just to secure permits and sign initial contracts to build the power infrastructure. This has started to change in the past two years, but there's still a long way to go. According to the International Energy Agency (IEA), global electricity consumption by data centres will double to 945 terawatt-hours (TWh) by 2030, representing roughly 3% of global demand for electricity. That's roughly the same as adding 34 Hinkley Point C-scale nuclear-power plants to the global grid. Between 2025 and 2030, data-centre electricity consumption is expected to grow by 15% per year, four times the growth rate of total electricity consumption across all other sectors.

Electricity consumption from accelerated AI data centres, the most intensive units that train AI models, will rise by 30% annually. In the worst-case scenario, the IEA estimates that global data-centre demand for electricity could exceed 1,700 TWh by 2035, nearly 5% of global demand for electricity. If the industry becomes more efficient at utilising power, that figure could fall to 970 TWh. If the electricity industry fails to rise to the challenge, demand could be limited to 700 TWh by 2030, nearly 25% below the base-case scenario.

This bottleneck is most apparent in the US and China, where the most time and energy are being spent on AI development. China and the US will account for nearly 80% of global data-centre electricity consumption growth to 2030, according to the IEA. The US, in particular, is facing a projected power access shortfall ranging from 10.4 gigawatts (GW) up to 49GW by 2028, even though projections from the US Energy Information Administration (EIA) show the grid adding 86GW of new utility-scale electricity-generation capacity in 2026, the largest single-year rise since 2002.

Rise of the ‘Bring Your Own Power’ model for data centres

To get around some of these issues, data-centre providers are increasingly seeking to innovate. The “Bring Your Own Power” (B-Y-O-P) movement, for example, is bypassing grid-connection bottlenecks by building on-site microgrids, utilising utility-scale batteries, solar panels, fuel cells and wind and gas turbines. Elsewhere, data-centre operators and hyperscalers are working with utility providers to purchase and install natural-gas power stations.

Data-centre providers are also shifting their attention to gas-rich zones such as the Permian Basin in Texas and New Mexico, where natural-gas pipeline capacity is severely constrained (gas prices recently dropped below zero despite the war in the Middle East). Companies are building off-grid data centres directly at these extraction sites, monetising otherwise stranded gas that would have zero economic value.

For example, Microsoft (Nasdaq: MSFT) and Chevron (NYSE: CVX) are partnering to build a $7bn, 2.5GW off-grid natural-gas power complex in Pecos, Texas, specifically to supply Microsoft's AI data centres under a 20-year agreement. Williams (NYSE: WMB), a pipeline company transporting a third of the natural gas moving across the US, is developing “Neo”, a $2.3 billion project utilising gas turbines paired with battery energy storage systems (BESS) for a major hyperscaler. This is the company's fifth BYOP agreement.

Some providers are also turning to AI to help mitigate AI's impact on power grids. A recent report from the World Economic Forum notes that “power-flexible” AI factories can dynamically modulate their electricity use, throttling energy-intensive tasks such as model training during periods of stress for the grid and routing more mundane tasks (such as answering simple questions on ChatGPT) to other locations. This flexibility enables data centres to capitalise on the volatile nature of renewable-energy generation.

Energy prices take the strain

As the utility market has struggled to adapt to the surge in demand for electricity, prices have responded. Wholesale electricity prices have jumped across all global markets, and the impact is particularly acute in the US. In some eastern US states, prices have risen 76%. According to the Bureau of Labour Statistics, across the country, electricity prices are rising nearly 61% faster than general inflation. The entire supply chain is feeling the pain. Lead times for the production and delivery of grid equipment have skyrocketed. Standard electricity transformers now take 128 weeks to deliver, compared with just 16 weeks in 2019. In some cases, specialist transformers are being delayed for nearly three years.

The production of highly efficient combined-cycle gas turbines can take up to four years, more than double the length recorded in 2022, and across the entire supply chain analysts put the average price rise at 30% across all grid equipment. There's also been a dramatic shortfall in the number of construction engineers and electricians, with the figure put at nearly 300,000 construction engineers and electricians in the US over the next decade. There are no quick solutions to any of these problems. While producers try to scale up output to meet rising demand, it looks as if they will continue to hold all the cards for the next five years at least.

There are three ways for investors to play this trend. There are the companies that generate power, those that make equipment for power stations, such as gas turbines, and those that manufacture cables and equipment to transmit electricity from A to B.

Tap into the AI energy boom with power players

One of the hottest plays is GE Vernova (NYSE: GEV). Created as part of General Electric's break-up, GE Vernova specialises in designing, manufacturing and maintaining equipment for the power-generation industry. Its technology provides roughly 25% of the world's electricity and the group has an order backlog of $163 billion, or 3.5 times sales. Its order backlog for gas turbines sits at around 100GW – around 2.5 times the UK's total daily electricity consumption. UBS has modelled 14% annual organic sales growth for the group through 2028 based on its current order backlog, with a 22.7% Ebitda margin by 2028, up from 8.4% in 2025.

Unlike the GPUs that power data centres, which have an estimated average life of around five to eight years, gas turbines can last up to three decades, locking in a multi-decade service contract for Vernova. The firms also offers kit for firms running older units (20 years and upwards) to help improve reliability and efficiency. Despite this growth and its key market position, there's a lot priced into the stock at a mid-30s price-earnings (p/e) ratio, but UBS argues that the valuation is worth it given the revenues and potential for margin growth.

Rolls-Royce (LSE: RR), which came close to a government bailout in the pandemic, is now one of the world's most sought-after power engineers. For the year to the end of 2025, the company reported a 12% jump in underlying revenue to £20 billion and underlying operating profit rose by 41% to £3,462 million, equating to a margin of 17.3%. Profit growth was driven primarily by the power-systems arm (25% of revenue), where the divisional margin expanded by 430 basis points to 17.4%. The firm put this down to “growth driven by data centres” and it's hoping its “power-dense” next-generation diesel and gas engines will continue to drive growth. Its technology is in demand as data-centre providers seek alternatives to bypass ever increasing queues for power-grid connections. The FTSE 100 company recently noted that orders across gas and diesel engines in the first quarter was around 50% higher than last year and March was a record month. Power Systems' order backlog was £7.3 billion at 31 March.

But Power Systems isn't just about data centres. The business also produces battery energy-storage systems and engines for Leopard tanks. What's more, last year Rolls-Royce conducted the world's first successful test of a high-speed marine engine running on pure methanol. There's also the company's nuclear business. A long-time supplier of nuclear reactors to the Royal Navy, Rolls-Royce has begun moving into the civil market with its small modular reactors (SMRs). In June last year, Rolls-Royce's SMR was chosen as the sole provider in the Great British Energy – Nuclear competition to build three SMR units in the UK.

Rolls-Royce SMR also received a strategic investment from CEZ Group, alongside a commitment for up to six units in the Czech Republic. In mid-June, the division was selected to deliver three SMRs on Sweden's west coast in partnership with Videberg Kraft. Based on current estimates, Rolls-Royce is trading at a forward p/e of 38.2, falling to 32.7 in 2027, according to average analysts' estimates. Those have ticked higher after the company's latest upbeat trading update and Berenberg has pencilled in a forecast of £8 billion of share buybacks over 2026-2028, split by £2.5 billion in 2026, £2.7 billion in 2027 and £2.8 billion in 2028, with scope for more cash returns if cash flow beats projections over the coming months.

A US peer of Rolls-Royce is BWX Technologies (NYSE: BWXT). Like its UK counterpart, BWX has the backstop of a US Navy contract in its back pocket to support its general operations – it has been the sole nuclear-fuel provider to the US Navy for more than 70 years. It's now seeking to grow in the civil market, where it provides specialised, complex, high-precision equipment used in nuclear reactors, including steam generators, reactor-pressure vessels and piping. It has an order backlog of $8.7 billion (around 2.2 years of revenue) bolstered by the recent $1.4 billion set of contracts through the US Naval Nuclear Propulsion Programme. However, at nearly 50 times forward earnings, there's a lot baked into the current share price.

How to invest in the undersea cable kings

The energy-transfer market is more concentrated than the other two potential investment segments. The renewable-energy transition has triggered an unprecedented global demand for ultra-high-voltage subsea cables to transport power from offshore wind and solar sites to urban centres, a market that did not exist 15 years ago. It's currently dominated by a European oligopoly consisting of Prysmian (Milan: PRY), Nexans (Paris: NEX), and NKT (Copenhagen: NKT). These companies emerged as the winners in what was an incredibly competitive market, with lots of smaller players that couldn't keep up with the capital-spending commitments required to manufacture vast undersea sea cables.

High-voltage direct-current (HVDC) cables can be thick, and they must be kept completely straight during manufacturing, which often requires companies to hang them inside skyscraper-high warehouses. The capital required to build this infrastructure runs into the billions. For example, the 500-kilometre Eastern Green Link 2 (EGL2) project in the UK, the single largest ever investment in electricity-transmission infrastructure in the country, has a price tag of £4.3 billion, with £2.7 billion of that for the cable itself.

The global high-voltage submarine cable market is expected to grow at a compound annual growth rate of 17.3% over the next decade. Production hit 7,000 kilometres in 2025, an all-time high, and the major players are rapidly ramping up production. Prysmian is drawing on its experience in this market to expand in the DC inside-building segment – essentially wiring up the power inside data centres. The company believes it will become a one-stop shop for data-centre construction contracts, building long-haul subsea connections and shore-based transmission infrastructure, and then for infrastructure throughout the building to power GPUs and air-conditioning units.

Management has estimated that overall global demand for DC power will expand at a compound annual growth rate of 33% over the next five years, with the bulk of this coming from AI-related data-centre growth. Analysts have pencilled in earnings growth of 25% for 2026, followed by 23% for 2027, with a net profit of €1.7 billion projected for 2027, up nearly ten times from 2020. Based on these projections, the shares are trading at a 2027 p/e of 24.9, which doesn't seem too demanding for a high-growth business operating in an oligopoly.

Prysmian is around three times the size of its smaller peers, both of which are using their growing profitability and cash flow to expand into newmarkets. Of the two, Paris-listed Nexans is the cheapest, trading at a 2028 p/e of around 13 based on management's growth targets. The group has laid out a road map to achieve an adjusted Ebitda of €1.2 billion by 2028 (up from about €750m) through growth in its three main businesses: PWR-Transmission, PWR-Grid and PWR-Connect. Sales are already locked in with a backlog of €7.9 billion by early 2026, enough to cover sales through to 2028.

Deals will also be a major part of the future growth plan. It recently added US-based Republic Wire to the stable to bulk out its US arm (about 15% of revenue). Republic reported sales of €52millionmn in its latest fiscal year and will be a key conduit for Nexans to enter the US data-centre market. Nexans plans to use Republic Wire's established channels to sell its own comprehensive offering of medium-voltage and grid technology into premium US end markets. The acquired business is currently finalising a significant expansion programme, which will increase its production capacity by about 30% by the end of 2026.

How to play coal

Another FTSE 100 company that's strategically well placed is National Grid (LSE: NG). Although still small compared with the US and Chinese markets, the UK data-centre market is the largest in Europe. National Grid believes demand for electricity in the UK will increase by 30% by 2035 to 290GW with a 90% increase in installed generation capacity to 370 TWh. To meet this demand, the company is investing £41 billion by 2031 to expand its regulated asset value by 60% to £60 billion. It is also going to invest £29 billion to expand its US business to a regulated asset value of £45 billion, with a focus on its key markets of New York and Massachusetts. The shares currently look cheap, selling at a forward p/e of 13.9.

One sector investors could also consider is coal. According to Global Energy Monitor, more than 2,200GW of coal-powered generation still operates worldwide, with another 710GW under development. China is scaling up its coal-output market to meet increased demand for energy and a total of 32 countries are proposing, or building, new coal plants to meet the growing need for power. At the beginning of June, Donald Trump announced plans to build two new coal plants in Alaska and West Virginia under the Defence Production Act, adding to the US coal fleet, which supplies 15% of the country's demand for power. Alliance Resource Partners (Nasdaq: ARLP), Peabody Energy Corp (NYSE: BTU) and Warrior Met Coal Inc (NYSE: HCC) are three left-of-field plays worth considering here.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

The invisible hand of the facilities management (FM) industry operates in almost every large commercial building. Someone is maintaining the chillers and the fire-suppression system. Someone is cleaning the floors. Someone, in theory, knows whether the heating, ventilation and air-conditioning (HVAC) unit on the fourth floor is three months from failure. This industry, sprawling, unglamorous and rarely covered by analysts, generates north of $4 trillion in annual global revenue. It is also in the early stages of a bifurcation that will create some genuinely interesting investment opportunities and destroy a remarkable amount of value for those who pick the wrong horse.

The core problem with facilities management is that it has spent decades solving the wrong problem. It has been focused on fixing things rather than understanding why things break in the first place. It has been reactive when its customers need it to be predictive. It has been operational when the most sophisticated clients are desperate for something more strategic. And it has been unable to provide evidence of the value it affords. Every contract renewal thus defaults to a conversation about cost that facilities management companies are badly placed to win.

Technology is now changing this, but not in the way most industry observers have assumed. The narrative for many years has been that some form of unified digital platform, a so-called single pane of glass, would allow facilities managers to own the data and therefore control the strategic conversation with their clients. That narrative is broadly correct. What it has missed is who actually ends up controlling that glass.

The problem with facilities management companies

The smart money is on the original equipment manufacturers (OEMs). Siemens, Schneider Electric, Johnson Controls and Trane Technologies have all made aggressive acquisitions of integrated workplace-management software businesses in the last three years. They control the mechanical and electrical systems that generate the core telemetry. They are now buying the platforms that interpret that data.

CBRE and JLL, two giant US commercial property services and investment groups, have responded with their own investments in technology, and CBRE in particular has built something genuinely differentiated. Its technology stack, running from raw asset data through AI-driven performance optimisation to a generative AI interface for facility managers, is meaningfully ahead of most traditional facilities management rivals.

More importantly, CBRE has made a strategic choice that I think is correct and underappreciated: it self-delivers the engineering and maintenance work where risk and complexity are high and it subcontracts almost everything else. This keeps the business focused on what it does best, avoids the diseconomies of running enormous low-margin cleaning and catering workforces, and keeps the conversation with customers at the level where CBRE's technology and insight capabilities actually add value.

The integrated model that FM firms ISS, Coor, Mitie and ABM Industries have pursued, employing vast workforces across subsectors from cleaning and engineering to food service, has struggled with low margins, volatile earnings and weak cash generation.

It is not that these companies are badly run. It is that the model is structurally disadvantaged. Every dollar of capital reinvested in innovation or process improvement flows through to a smaller share of the overall business when that business is simultaneously managing electricians, cleaners, security guards and caterers. The benefits of scale are harder to capture. Best practice is harder to standardise. The most talented engineers would often rather work for a pure-play technical services company than be one service line among eight.

Compass Group found the right path

There are exceptions and they are instructive. Compass Group has built one of the most impressive records in global services by staying almost entirely focused on food. Its management and performance framework is a masterclass in what happens when a large, decentralised services firm imposes a common operating language and a small number of clearly defined drivers of value across an entire organisation.

The framework ties every decision to one of five levers determining profit or loss – from client retention and consumer participation through to labour scheduling and overhead control. It sounds almost boring in its simplicity. It has produced two decades of best-in-class margin delivery at scale.

My preferred name among facilities management stocks is Bravida (Stockholm: BRAV), the technical services group that installs and maintains the electrical, HVAC and plumbing systems in buildings across Sweden, Denmark, Norway and Finland. It does not try to do everything.

Bravida focuses almost entirely on facilities management engineering delivered through a network of 330 branches providing the local density and proximity to customers that makes the economics work. When you build genuine scale in a single technical discipline, you can standardise ways of working, invest properly in training and certification, attract the best engineers, and compound efficiency gains year after year.

Bravida has been through a difficult patch, hit by a Swedish construction downturn, a governance failure in one branch that has since been closed and prosecuted, and some bad debts from a large customer. The share price has derated significantly. I think that creates an opportunity. The underlying business model is sound, the structural drivers for technical building services are strongly positive, and the company's internal focus on operational excellence is exactly the kind of self-improvement culture that separates durable compounders from cyclical operators.

The privately owned CFS, or Churches Fire & Security, is another business worth watching. It operates in the UK fire safety and electronic-security market, an arena driven by tightening regulation, historic underinvestment and alarming fragmentation that sees roughly 2,000 small operators competing with essentially no scale advantages. CFS has now absorbed over 70 businesses, each integrated fully within three to six months. Revenue has jumped to £100 million at attractive margins. The model is replicable, the regulatory tailwinds are real and the market is large enough to sustain further consolidation.

Depth beats breadth in the facilities management industry

Focused, scalable business models with genuine density economics outperform diversified ones over time, by a wide margin. The temptation to add services and geographies is understandable in an industry where large contracts look attractive from the outside. But every incremental service line is also an incremental distraction. Peter Thiel has noted that you cannot run dozens of start-ups simultaneously and hope one works out. The same logic applies to a low-margin services business with finite capital and management bandwidth. Depth beats breadth, almost every time.

The FM industry is also at a genuine technology inflection. Advances in building sensors, AI-driven predictive maintenance and integrated data platforms are removing the ability of mediocre operators to hide. Buildings that were opaque are becoming transparent. Clients that once relied on SLA compliance reports are now demanding energy dashboards, asset-lifecycle forecasts and sustainability documentation.

Operators with weak processes and inconsistent data capture will be exposed. Operators with strong processes, standardised ways of working and an ability to translate data into value for customers will find themselves able to charge for something other than just labour. The buildings around us are getting smarter. The companies that service them need to be smarter too. The ones that are will be interesting to own.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Korea is still an emerging market, or so MSCI reckons. On Tuesday, the most important provider of global indices – the MSCI World and the MSCI Emerging Markets matter much more than the equivalents from FTSE Russell and S&P Dow Jones – once again declined to put it on the watch list for upgrade to developed status.

On one hand, this situation feels increasingly ridiculous. Korea is a very advanced, high-tech economy, home to key tech players such as Samsung Electronics and SK Hynix. GDP per capita measured at purchasing power parity is higher than the UK, France, Japan and many other heavyweights. How can this be an emerging economy in any meaningful sense?

Yet there are aspects to Korea that feel like an emerging market. The ones that MSCI cites are certain limitations that bother institutional investors (restrictions on trading the Korean won offshore is a key one) – although FTSE Russell has classed Korea as developed since 2009, so the importance of these is not cut and dried.

However, perhaps more significant for the long-term future of the Korean stock market is the dominance of large business conglomerates (chaebols), of which the Samsung group is the biggest. The founding families of these groups still control them – often using a series of shareholdings between different listed entities – and frequently make decisions for their own benefit to the disadvantage of minority shareholders.

Generational changes are happening in Korea

Corporate governance is a major reason for the “Korean discount” – the fact that Korean stocks trade at lower valuations than peers elsewhere – but there are signs that this is changing. Policymakers have been pushing reforms, inspired by what governance changes in Japan have done for that market, with some success.

Generational changes also mean a structural shift in attitudes is inevitable, suggested a Korea manager at a recent conference. The individuals who built up chaebols in the 1960s and 1970s put huge importance on passing on control to their heirs as cheaply as possible (Korea has very high inheritance tax). They are now largely dead, the handovers are being completed, and tax bills are being settled. Their heirs will have different priorities that may often be better served by unlocking the full value of their businesses.

So the bull case for Korea sounds easy to make. It does not depend on MSCI one day acceding to the obvious, although being added to the developed index would result in significant inflows from tracker funds. And on the face of it, Korean stocks look very cheap – the MSCI Korea stock market index is on a forecast price/earnings ratio (p/e)of eight.

Yet this reflects the huge weight in Samsung and SK Hynix (65% combined) and how fast they are expected to grow. The MSCI Korea Equal Weight is on a forward p/e of 15, which is not cheap. Most of all, note the market is up by 260% in won terms in a year. If the AI boom continues, it will go higher, but have no illusions. Right now, a Korea stock market tracker is not a valuation play or a reform play – it is entirely an AI play.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

A new world calls for new stocks. In the VT De Lisle America Fund, we use the paradoxical combination of value plus momentum to find winners for the next decade. For 40 years, declining interest rates and disinflation created a powerful tailwind for steady growth stocks such as McDonald's, Nike and Procter & Gamble. Their predictability was rewarded with rising price-to-earnings (p/e) multiples, thus they strongly outperformed the market. That all changed in 2021 with the return of higher interest rates and inflation, combined with the launch in 2022 of ChatGPT, which funnelled capital into AI infrastructure.

Higher rates and inflation tend to push down p/e multiples and put pressure on profits due to rising costs of materials and labour. At the same time, the urgency to spend on building out AI pushed capital into different, previously unloved, parts of the economy: construction, manufacturing and blue-collar jobs. Investors like to find a comparison with past cycles. We think the 1970s provides the best precedent, but with the addition of AI. So how will that play out today?

Companies that own scarce real-world assets and have growing order backlogs gain pricing power in this new capital-expenditure-driven cycle, positioning them to thrive in the new industrial economy. Within these big macro themes, minor ones are also emerging. One we like is the break-up of old industrial conglomerates via “spin-offs” – the packaging of overlooked industrial assets into attractive new listed companies.

Three stocks for your portfolio

In 2025, Honeywell spun off its speciality chemicals division, Solstice Advanced Materials (Nasdaq: SOLS). Solstice holds an oligopolistic position in refrigerants (its cash cow, increasingly necessary for data-centre cooling) while expanding capacity in its specialist chemicals business, supplying America's semiconductor supply chain. But the hidden crown jewel is its stake in the only US uranium conversion facility – one of only seven in the world – and a scarce asset during a nuclear renaissance. Although expensive-looking at a high 20s p/e, its earnings power is underappreciated as its chip exposure grows and higher-priced uranium contracts begin to roll in.

Forum Energy Technologies (NYSE: FET) makes high-tech parts for oil wells and subsea exploration. It is a picks-and-shovels play on rising global energy needs due to AI and on the increasing complexity of extraction. Forum operates in a high-value niche and is a leading player in all its product lines. The company prides itself on its patented technical know-how, which makes it hard to compete against. Low reinvestment requirements also provide high levels of cash generation. Even after a near tripling over a year, we think Forum's stock is a buy given its cheapness relative to cash flow.

Demographic trends offer another type of steady growth. Pennant Group (Nasdaq: PNTG) owns, leases and operates care facilities for the elderly and sees the ageing population in the US as a steady tailwind to increase sales for years. Its ability to renovate old buildings to add value in a market where supply is constrained by the cost of new-builds is impressive. Growing demand and scarcer assets, plus entrepreneurial local management teams, are giving it the chance to execute its vision. Pennant trades on a low 20s p/e and has a high double-digit growth rate, giving it an enviable p/e to growth ratio of not much above one.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Alphabet, Google's parent company, is listed in the US with a total value greater than that of the entire UK stock market. Billions of people use its search engine every day – indeed, the Google name has become so ubiquitous that it is now used as a verb around the globe. Yet there is more to Alphabet than Google.

Beyond search and advertising revenues lies an empire that includes everything from deep-sea cables to self-driving cars and energy storage. The business uses the billions harvested from search advertisements to fund massive bets on the future and is fast becoming one of the most influential companies in history.

The Alphabet name reflects a corporate mission to fund independent bets that produce “alpha” – the term in finance for an investment that outperforms the broader market. Alphabet wants to be the structure underpinning countless future innovations. The name signalled to the market that the firm was no longer just a search engine, but an incubator of new technology.

Alphabet's rise from Google search to global dominance

Search remains Alphabet's largest source of profits. Its enormous scale explains why the company can afford to fund so many broader ambitions. When the business launched from a garage in 1998, it was just one of many experimental search engines competing on the early World Wide Web. Its rapid rise to dominance was driven by a proprietary algorithm called PageRank. Unlike rival systems that mainly counted how often a keyword appeared on a page, Google ranked pages based on the quality and importance of links pointing towards them. A link from a respected university or major news site carried far more weight than one from an obscure blog. This breakthrough produced far more useful search results, triggering a wave of adoption that quickly led to dominance. Put simply, Google search worked much better than everything else.

Today, Google remains the search engine used by most. It controls more than 90% of the worldwide search market and processes billions of queries every day. Its closest rival, Microsoft's Bing, holds only a tiny share by comparison. Google's reach also extends far beyond its own homepage. The company provides the underlying search infrastructure for countless browsers and software applications around the world. Competitors struggle to replicate what Google has built because search engines improve through users' behaviour. The more people who use the platform, the more data it collects and the better the system becomes. By capturing most of the world's search data, Google continuously improves, creating a self-reinforcing cycle that keeps competitors behind.

This constant stream of searches is transformed into revenue through a system of paid results. When a user searches for a term with commercial value, the engine places sponsored links at the very top of the page, positioned directly above the information. Google avoids charging businesses a flat fee simply to display these links. Instead, it operates on a pay-per-click model, collecting a fee when a user selects a sponsored result. Because millions of consumers use the search box to find products, services and local businesses every second, these small fees accumulate into billions of dollars of highly predictable revenue.

This is so profitable because the underlying mechanics require little human involvement. Traditional advertising agencies only grow by hiring armies of account managers and media buyers to manage campaigns. Google removed much of this by building an automated, self-service advertising platform to run its pay-per-click business. Advertisers simply log into a dashboard, set their budgets and bid against one another for visibility tied to specific queries from users. Valuable searches, such as those related to legal or financial services, command extremely high advertising prices. This allows Google to generate enormous profits from everyday internet traffic without relying on large numbers of highly paid employees.

At the end of last year, Alphabet employed roughly 191,000 people worldwide. However, those workers are spread unevenly across the business. Most do not work directly on the core search or advertising operations. Instead, they are concentrated in labour-intensive divisions such as Google Cloud and areas such as compliance and other administration. The core systems and software that power Google's search engine require only a small group of engineers to maintain and monitor it. By the end of 2025, Alphabet was generating annual revenue equivalent to more than $2.1 million per employee, although the figure within search alone is probably far higher, perhaps as much as $10 million per employee. This ultra-low headcount relative to sales creates a self-operating engine that supports the rest of the organisation, funds Alphabet's broader ambitions and produces vast profits – thought to be $1 billion every two to three days.

Branching out into Google Cloud and beyond

The fastest-growing large-scale part of the business outside of search is Google Cloud. This division sells computing power and data storage to large corporations and public-sector organisations, providing a platform for businesses to build, host and run their own software applications. Unlike the search engine, the cloud business is inherently labour-intensive, requiring a global sales force. By the end of 2025, the unit had exceeded $70 billion in revenue, driven by demand for machine-learning applications. This segment spent years burning cash to build physical data centres, but has now matured into a highly profitable operation, generating billions in quarterly operating income. The third large division within Alphabet is subscriptions and devices. This includes premium, advertising-free access to YouTube, digital storage upgrades through Google One, which pools together personal consumer storage for Google Drive, Gmail and Google Photos. This is distinct from the corporate cloud, focusing instead on individual consumers' hardware, such as Pixel smartphones. Total consumers' subscriptions have climbed past 325 million globally. This division generates more than $50 billion annually.

What ultimately cements Alphabet's dominance is how seamlessly it intertwines these separate businesses. Google Video's early failure was solved by acquiring YouTube, for example, which was then deeply integrated into core search results. Google Maps was built to serve local search needs, but is now embedded directly into the Android operating system and Android Auto vehicles' dashboards. This interconnected web provides built-in, zero-cost marketing across the entire portfolio, making the user's experience smoother while locking consumers firmly inside Alphabet's systems.

The relentless flow of cash allows the parent company to invest on a scale few companies in history can match. Rather than returning profits to shareholders through dividends or share buybacks, Alphabet operates like a vast venture-capital fund. Its investment strategy is divided into three tiers. For near-term product improvements, it uses Google Labs, a fast-moving environment where software teams test early features, such as improved AI systems, directly with the public. For medium-term time horizons, the company focuses on strategic acquisitions, buying external platforms and scaling them over time. Finally, the long-term horizon is handled by X Development (formerly Google X), the “Moonshot Factory” created to back speculative technologies ranging from self-driving cars to grid-scale energy storage. In these ways, Alphabet channels its search, cloud and subscription revenues into tomorrow's cutting-edge technology.

Alphabet's formula for acquisitions

Alphabet has acquired more than 250 technology companies over the years. Each deal has followed a similar formula: acquire a promising but financially constrained technology and scale it using the company's engineering expertise and vast profits. Google Maps originated from Keyhole, a struggling start-up founded in 2001. Keyhole developed a 3D digital globe called EarthViewer 3D and even received early backing from America's Central Intelligence Agency. The technology was impressive, but the business model weak. Keyhole sold its software on physical CDs to real-estate firms and defence agencies. Google recognised that around a quarter of all web searches were location-related and acquired Keyhole in 2004 for roughly $35 million. It removed the expensive pricing, introduced a cleaner, more accessible interface, and relaunched the platform as Google Maps. In the process, it transformed a niche military-style tool into a free utility that has become almost as recognisable as the search engine itself.

The acquisition of YouTube in 2006 stemmed from Google's own failure in online video. Its in-house platform, Google Video, was losing ground to its rapidly growing rival. YouTube succeeded by offering a simple interface that allowed anyone to upload and stream videos easily. However, by the summer of 2006, the company was struggling under the weight of its own popularity. Hosting costs were soaring, while copyright lawsuits from traditional media companies threatened its survival. Realising Google Video had lost the battle, management stepped in with a $1.65 billion acquisition. The takeover rescued YouTube from likely insolvency and allowed Google to secure the leading destination for online video before legacy media companies could shut it down. By the end of 2025, YouTube was generating more than $40 billion in annual revenue.

The 2005 purchase of Android is probably the most successful acquisition. As a start-up, it had been developing an operating system for mobile handsets, but ran short of cash to fund engineering salaries. At the time of its purchase, it was a small company employing eight people and was bought for just $50 million. This was such a small sum at the time that it wasn't even disclosed to the stock market. The goal of the acquisition was to prevent competitors from blocking its search engine on mobile devices. By making Android free, Google rapidly came to dominate mobile software, eventually capturing more than 70% of the global smartphone market. This comparatively small investment helped ensure that the search business continued to grow even as smartphone usage overtook computer usage.

The 2014 acquisition of DeepMind secured Alphabet's lead in AI. The laboratory, which was founded in London by Demis Hassabis, Shane Legg, and Mustafa Suleyman, had assembled one of the world's strongest machine-learning research teams. DeepMind focused on neuroscience-inspired AI and deep reinforcement learning. Yet cutting-edge AI research is very expensive, requiring vast computing resources and highly paid engineers while producing little immediate revenue. Much of Hassabis's time was spent raising venture capital. Recognising that DeepMind needed the support of a company with deep pockets, the founders agreed to a £400 million sale to Google, with Hassabis taking on the role of CEO of the renamed Google DeepMind. The deal kept the research group based in London and provided the resources needed to pursue foundational scientific breakthroughs. That long-term backing ultimately paid off. Among other things, Hassabis's work on protein folding using DeepMind recently won the Nobel Prize in chemistry.

How Alphabet is shooting for the moon

Where Alphabet stands apart is its willingness to invest in technologies that may take decades to mature. The X Development division filters all ideas through a demanding three-part screening process to protect capital. Projects must address a global problem affecting millions, propose a radical breakthrough solution and rely completely on technology that does not yet exist. Incremental improvements are rejected outright. To encourage bold experimentation, X is also designed to reward failure. Teams are expected rigorously to test their ideas and can even receive bonuses for proving a project is technically or economically unworkable before significant resources are wasted.

This strategy has produced a trail of discarded technologies, including mysterious, giant floating barges intended to be high-end, floating marketing showrooms; a technology for storing renewable energy by pumping electricity into massive tanks of molten salt and chilled liquid; high-altitude, helium-filled balloons designed to float in the stratosphere, creating a shifting network to beam wireless internet down to remote rural communities. But the crown jewel of the moonshots to date is Waymo, the autonomous-vehicle division that began life in 2009. Waymo shows how a massive cash cushion allows a company to outlast an industry cycle. While some car makers promised self-driving fleets by 2018 only to scale back their ambitions when machine learning proved too difficult, Alphabet simply maintained its multi-billion-dollar funding trajectory. By refusing to rush out unproven systems to market, the division solved the major challenges.

Waymo has now achieved scale, with roughly 3,700 vehicles operating, servicing half a million paid rides per week. Its fleets of autonomous vehicles operate robotaxi networks across major American cities, including Phoenix, San Francisco and Los Angeles, completing passengers' trips without human drivers. In September of this year, it is due to launch in London. What began as a highly speculative experiment has matured into a genuine advancement in transportation.

Deep underwater, Alphabet is also building global subsea cable infrastructure. This is an ongoing project that has so far created a total of 60,000 miles of armoured cabling crisscrossing the oceans. To support the growth of its cloud services and advertising, Alphabet shifted from renting space on third-party telecommunications networks to owning its own. These subsea lines are the plumbing of the internet, moving vast amounts of data across the world at the speed of light. In owning this infrastructure itself, Alphabet ensures its consumer services operate with lower latency than that of competitors.

Is Alphabet worth owning?

Turning digital advertising revenue into real-world infrastructure requires enormous investment. For investors, the key question is whether these assets will create lasting value or simply become an expensive distraction. The shares of Alphabet rarely look cheap on any conventional valuation metric, but waiting for a deep-value entry point has been a fool's errand. Ever since the company listed on the stock market in 2004, there has never really been a bad time to buy its shares.

Not that the shares have risen consistently. Alphabet's shares have fallen quite significantly a few times over the years, dropping roughly 50% during the 2008 financial crisis and weathering several 20%-30% declines since. Every single decline has proved to be an exceptional buying opportunity, as the underlying earnings have moved relentlessly higher. This compounding has generated astonishing wealth, transforming its founders into centi-billionaires and ranking them among the wealthiest individuals on Earth. The model does not just reward senior executives. In 1999, when Google employed just 40 people, massage therapist Bonnie Brown joined the company part-time. Her pay was only $450 a week, but she also received stock options. Five years later, she retired a multi-millionaire and went on to create her own charitable foundation. Had she kept her shares, she would now be a billionaire.

Despite a massive market valuation of about $4.5 trillion, Alphabet is still growing remarkably quickly. A simple heuristic for evaluating growing companies is to ask whether the business can generate enough operating profit within five years to make its current enterprise value look cheap. Specifically, can its future operating profits reach a tenth of that valuation? For Alphabet, that means aiming for roughly $450 billion in annual operating profit. Last year it made about $190 billion and is forecast to grow at 20%-25% per annum for the next five years. At that level of growth, the company's current trajectory makes $450 billion perfectly feasible.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Private debt went through a “golden moment” after the rapid post-pandemic rise in interest rates, said Jonathan Gray, president of alternative-asset giant Blackstone, in 2023. The question now is whether that golden moment is past. With interest rates expected to stay higher for longer, sovereign yields rising sharply, and cracks appearing last summer in US business development companies (BDCs), some investors wonder whether private debt is entering its first real test as an asset class – or even facing a day of reckoning.

Private debt is a broad label. It includes syndicated leveraged loans, direct lending, asset-backed finance and even fund financing. These distinctions matter. Syndicated loans are liquid and volatile, but trade at tighter yield spreads (ie, they promise lower returns). By contrast, direct lending – where investors such as funds lend directly to borrowers – is illiquid and assets are rarely marked to market.

Direct lending is often presented as the core of “true” private debt, due to its potential for higher risk-adjusted returns. So far, that record has proved hard to ignore for institutional investors. Spreads of around 550 basis points over base rates remain appealing, while reported default rates are still modest. As a result, it attracted large inflows from pension funds and insurers. Assets could reach $2.8 trillion by 2028, up from $1.8 trillion currently, according to private-markets data firm Preqin.

However, this influx of capital and its potential ability to skew industry fundamentals has meant that some participants and regulators have begun to voice concerns about the risks of a private-credit crisis. Some signs of stress have emerged. US BDCs made the headlines last year after a rise in the number of investors who tried to redeem their holdings. Open-ended funds that invest in illiquid assets can enter a vicious cycle when redemption requests rise: as funds sell their most liquid assets (often assets of better quality) to meet redemptions, that news creates an incentive for other investors to also redeem so as not to be the ultimate “bagholders” of the riskiest and less liquid assets. In order to avoid the proverbial rush for the exits, some BDCs were obliged to cap redemptions – which obviously does not improve investors' sentiment.

A more challenging macro-economic environment, slower growth, persistent inflation and rates that are higher for longer also raises the risk that the asset class could be entering a more difficult phase. A further concern for investors trying to understand these risks is that valuations are opaque and largely controlled by managers. However, as the asset class is becoming more institutionalised, index providers such as S&P and Bloomberg are developing private credit benchmarks. Along with BlackRock's acquisition of Preqin in 2024, this suggests private markets will become more standardised and scrutinised despite some fund managers fighting this trend.

Private debt may like high interest rates, but borrowers don't

However, the primary concern at present is that the same high interest-rate environment that makes private debt attractive to investors is also increasing pressure on borrowers. The probability of default is growing, while recovery rates in private debt have never really been measured. Defaults can be obscured: for example, the restructuring of US educational technology firm Anthology last year did not trigger a formal default, as its lenders allowed flexibility in payments. Payment-in-kind (PIK) – where interest is added to the loan principal to be paid at maturity – can similarly mask stress by deferring payments out of cash flow. The covenant-lite nature of some lending offers less protection than investors might expect, while recovery rates in restructurings may be weaker than for senior debt that requires regular repayments in cash.

Meanwhile, diversification in some private-credit vehicles looks weaker than it first appears. A significant share of portfolios are exposed to software and/or software-as-a-service (SaaS) businesses, creating hidden concentration risk. These companies benefit from recurring revenues, predictable cash flows and healthy earnings before interest, tax, depreciation and amortisation (Ebitda) margins, but there are fears that artificial intelligence (AI) fundamentally threatens their business models. AI may not disrupt these companies overnight, but it could gradually erode pricing power, compress margins and weaken exit valuations, making refinancing more difficult.

The sell-off in these software stocks earlier this year, combined with the impact on private debt as investors realise how exposed some lenders are to the sector, has highlighted hidden correlation risks in supposedly diversified portfolios. All lenders now have their eyes on business-software firm Visma, which has reportedly delayed its planned initial public offering (IPO) until 2027 – a decision that perhaps suggests growing caution among investors despite strong operations and cash generation. This highlights another important feature of the market: concentration in a small number of large credits backed by private-equity sponsors. Jitters in private debt can – by definition – be less publicly visible, but the performance of HgCapital Trust, the specialist private-equity investor in software and services that has 13% of its portfolio in Visma, testifies to how sentiment has changed: it is now trading at a 25% discount to net asset value (NAV), having been on a premium in 2024.

Momentum remains in the private debt market

And yet, the market remains active. Despite slower mergers and acquisitions (M&A) activity in Europe during the first quarter, two large refinancings – of TK Elevator (€1.8 billion) and Global Sports Group (€2.2 billion) – drove solid deal volumes. “The prominence of these large-cap deals also suggests direct lenders have successfully taken advantage of the volatility seen in Q1 to reclaim some market share from the syndicated markets,” notes Debtwire. The amount of funds raised in 2022-2024 that are not yet invested (known as “dry powder” in the industry) will allow the market to keep momentum.

The fact that direct lending can offer 100-500 basis points of additional yield over what is available in the traditional syndicated loan market (depending on the amount of leverage used by the vehicle, or if it acquires subordinated debt and junior capital) – combined with the ability to negotiate better structural protections than broadly syndicated loans – should continue to attract strong interest.

Dry powder will help lenders to inject capital if needed and protect their initial investments, tempering the risk of a private-debt crisis. It will be used to address the “refinancing wall”: a large stock of debt raised in the low-rate era now needs to be rolled over at significantly higher cost. So far, this has been managed through extensions and amendments – euphemistically renamed as Liability Management Exercises (LMEs). Widespread defaults are rare.

We also need to keep in mind that there are important regional differences in private debt. The US market is more mature and increasingly competitive as banks re-enter leveraged finance, given an incentive by the Trump administration's new regulations. In Europe, private debt is still expanding, supported by a more fragmented banking system and structurally lower penetration. Note too that public markets are treating private credit as a homogeneous bet on leveraged buyouts. However, outcomes will depend less on the asset class itself and more on managers' capabilities: origination, underwriting discipline and balance-sheet flexibility. Scale is an advantage in stressed environments. Large managers such as Ares, Apollo and Blackstone have the ability to amend loans, provide follow-on capital and manage restructurings internally. Origination is the real moat, because it separates lenders that focus on financing commoditised leverage buyouts (LBOs), which are often syndicated loans, from those that source bilateral or semi-bilateral deals with pricing power and better protections. Collateralised loan obligations (CLOs) – securities backed by a pool of loans, often from LBOs – have held up so far and offer diversified exposure to leveraged loans. However, their lack of control over the underlying loans and consequently over the outcomes when borrowers come under pressure may become more apparent as the cycle turns.

The easy part – that “golden moment” – is over. Higher rates are a source of stress. As refinancing pressures build and dispersion rises, returns will be less a function of a rising tide lifting the whole asset class and more driven by selection, discipline, structure and scale. A broad opportunity is turning into a more challenging game.

High-performing private-debt funds generally require high minimum commitments (ie, institutional size – maybe €2 million - €5 million to a private-debt fund such as CVI with strong exposure to central Europe). Some managers are starting to target individual investors as a source of further capital, but one should always be aware that individuals are less likely to get the best opportunities. Avoid products offered by private banks where layers of fees accumulate.

More interesting are listed funds and managers exposed to the sector. Some merit caution, but fears of a crisis may create buying opportunities in the best ones. In Europe, there is Tikehau Capital (Paris: TKO), which has robust exposure to special situations, ICG (LSE: ICG) and CVC Income & Growth (LSE: CVCG). In the US, Ares Management (NYSE: ARES) and Blackstone (NYSE: BX) have developed the strongest platforms in terms of origination. Apollo Global Management (NYSE: APO) has the strongest exposure to direct lending, but readers may prefer to wait for the dust to settle.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

RentGuarantor Holdings (Aim: RGG) has been given a boost by the Renters' Rights Act, one of the most significant pieces of legislation to hit the UK rental market in decades. It's only been in force since the beginning of May, but the Act is already driving a complete rewriting of the market.

Under the new law, fixed-term tenancies have been abolished, “no fault” evictions are no longer allowed, rents can only be raised once a year, and during the first 12 months of the tenancy, the landlord cannot serve notice to move back into the property or sell it. These changes, far from protecting tenants, have forced landlords to become more defensive.

The changes have made it much harder for landlords to evict tenants who can't or won't pay their rent, piling pressure on a system that's already on the verge of collapse. According to professional body Propertymark, due to lengthy court backlogs, the average time from claim to repossession has risen to more than 68 weeks, compared with just over 20 weeks in 2019. At the point of eviction, average unpaid rent stands at £12,708 across England and Wales and £19,223 in London.

Landlords have responded by demanding that tenants provide a guarantor before they agree deals. According to multiple reports, around 40% of landlords now require guarantors for both new and existing tenants. This is where RentGuarantor comes into play.

How RentGuarantor works

The firm is a rare example of how effective London's capital markets can be for early-stage growth businesses. Founded in 2016 by Paul Foy, a property investor since the mid-1980s, RentGuarantor does what it says on the tin – guarantees rents. Tenants pay a fee (£20) for an initial background check and the firm uses tools such as Open Banking and AI to calculate how much the tenant can afford and if they're able to maintain payments. If the tenant passes the check, which should be completed the same day, RentGuarantor can offer the guarantee.

This incurs a further fee, usually around three to five weeks' rent, depending on the underlying risk profile. When the tenant has paid and signed, RentGuarantor provides a legally binding guarantee of rental payments to the landlord or letting agent. Unlike traditional guarantors, such as parents or grandparents, this provides an extra layer of protection for the landlord. RentGuarantor passes the risk to a panel of insurers while collecting the origination fee and remaining the key point of contact for customers.

Five years of RentGuarantor Holdings on the London market

After spending five years building the foundations, Foy and his team took the company public in 2021. It listed on the Aquis exchange in 2021 with hardly any revenue and moved to the Aim junior market in the second half of 2025. The new listing raised £4 million in 2025 to support its growth efforts and it ended the year with revenue of £2.4 million, up 87% year-on-year. The founder has remained a key shareholder with a 30% stake.

RentGuarantor hasn't charged into the market seeking break-neck growth and drawing down shareholders' goodwill to fund spending. There's a very tight grip on marketing spending, which totalled just £200,000 in 2024 and £500,000 in 2025 against revenue of £2.4 million, or around £165 per contract (based on the year-end figure of 3,123 contracts). The focus over the past five years has been on getting the offering right and putting in place the right technology and team to scale up effectively.

The firm has now reached the point where this hard work is beginning to pay off. In May, the month the Renters' Rights Act came into force, RentGuarantor recorded a 115% increase in unaudited revenue compared with the average for the first four months of the year. Moreover, revenue per contract was up 24%. The group also recorded its first positive monthly Ebitda earnings since its admission to trading – well ahead of the board's expectations.

The challenges facing RentGuarantor Holdings

The key risk for the group here will be scaling up without falling flat on its face, as so many firms do when they encounter a sudden surge in demand. The Act is driving demand for guarantees, but it'll also lead to a surge in disputes.

To help, RentGuarantor is looking to AI and has an expert on the matter in its orbit. The AI strategy is being led by Dave Cliff, a non-executive director and professor of computer science at the University of Bristol. He previously worked at MIT's artificial intelligence laboratory, so unlike many other businesses, which seem to be turning to AI with little actual understanding of the benefits, drawbacks and costs, RentGuarantor looks well-placed to exploit the benefits of the technology fully. Management estimates the group can process 20,000 contracts per year, but that will rise to 100,000 with AI's help.

According to house broker Shore Capital, RentGuarantor could agree 7,000 contracts this year, 13,000 in 2027 and 62,000 by 2030. Revenue could hit £6 million in 2026, rising to £19 million by 2028 and £54 million by 2030. Even if it achieves this lofty growth, it would still leave the group at only 3.4% of the potential total market.

Now that the firm is essentially self-funding, there's scope for marketing spending to rise. Shore Capital expects a ten times rise by 2030, easily covered by the firm's 79% gross margin. The broker has pencilled in adjusted earnings per share of 3.6p by 2028. As with all early-stage firms, these forecasts are likely to be wrong, but they illustrate the growth potential if the firm manages to scale up over the next 12 months. This is a high-risk play, but one with a huge and growing market to support it.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

The emerging market (EM) universe is very diverse in terms of what drives individual economies. What does China have in common with India (other than being populous and in Asia) or either of them with Brazil? Yet they are treated as a block, and recent trends are stretching these contradictions further than ever.

A top-down investing strategy often involves assigning things to groups, then buying the most compelling groups or choosing the most attractive within a group. These groups can seem arbitrary – the difference between members can be as big as the similarities. Yet in the investment business, classifications that seem easy to understand can stick around well past the point where they make sense.

Standard rules of thumb for emerging markets would tell you that the last few months have been difficult. Many emerging markets are energy importers, so will suffer from higher oil prices. Markets also tend to be affected by inflows and outflows from foreign investors. If global investors get more nervous, they would be expected to cut emerging-market exposure first and take their money home. Yet the MSCI Emerging Markets index is up by 20% in sterling so far this year. How?

AI stocks are over-represented in emerging markets indices

The explanation hinges on two points. The first is that two of the biggest markets in the index are emerging markets only in one very specific sense. South Korea and Taiwan retain certain restrictions, mostly around their currencies, that MSCI deems incompatible with being in the developed markets group. Yet in many respects, they are both wealthier and more advanced than many developed economies.

The second is that a few huge companies – Taiwan Semiconductor (TSMC), Samsung Electronics, SK Hynix – are huge beneficiaries of the AI boom and are driving their markets even more than the Magnificent Seven drives the US market. Those three stocks account for almost 30% of the MSCI Emerging Markets index. Taiwan and Korea together make up 50% of the index. In turn, TSMC is 55% of the MSCI Taiwan, while Samsung Electronics and SK Hynix account for 60% of the MSCI Korea.

These are eyebrow-raising numbers. They have worked out very well for any broad emerging-market investor. Still, we must remember that if the AI boom ends and the US market slumps, the emerging market index will do the same – it's been a play on the same theme.

If you want diversification, you will only find it in funds whose mandate does not bring in these stocks – BlackRock Frontiers (LSE: BRFI) or Barings Emerging EMEA Opportunities (LSE: BEMO), for example. Of course, these funds have lagged in recent months, held back by the lack of tech exposure or battered by the Middle East crisis. I would not say it is yet time to rotate out of broader emerging market funds. But it is something to keep in mind if the crisis passes and the AI boom falters.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Helium is a commodity that is key to every major growth theme in the global economy – space, AI, and healthcare. But almost nobody talks about it – and the helium supply picture has just become dramatically more complicated.

The launch of a single Falcon 9 rocket consumes roughly 14%-18% of the world's daily helium production in a single ignition sequence. SpaceX launches Falcon 9 rockets dozens of times a year, with ambitions that stretch well beyond that frequency. The satellite industry is preparing for annual launch volumes of between 3,700 and 5,000 by 2030, as mega-constellations reach full deployment. Goldman Sachs anticipates 70,000 low Earth orbit (the region between 160 and 2,000 kilometres into space) satellite launches globally between 2025 and 2031. Every single one of them needs helium. There is no alternative.

So surely someone is producing more of it? They are not, at least not at anything close to the rate the market requires. Helium is not manufactured. It is extracted as a by-product of natural-gas processing in a small number of locations where underground concentrations happen to be commercially viable. The US and Qatar together account for more than 75% of global supply. Russia produces a significant share, but that supply is unavailable to Western markets. Algeria contributes a modest fraction. Everyone else is a rounding error.

Qatar's share, 30% of global supply, flows out of a single industrial complex at Ras Laffan. In March 2026, Iranian strikes forced QatarEnergy to cease production of liquefied natural gas (LNG) and associated products, including helium. Almost one-third of the world's helium supply was removed from the market overnight. Spot prices doubled. The south site at Ras Laffan took direct hits and will not restart before late summer 2026. Permanent capacity reductions, analysts say, will take years to recover fully.

The Strait of Hormuz is a trigger, not the underlying problem. Even before the first missile flew, this market had endured four recognised major shortages over the past two decades, each lasting two to three years before any equilibrium was restored. The conflict simply made the market's structural deficit visible. Compounding the supply picture is the nature of helium itself. It is the second-lightest gas on Earth. It escapes containment at a rate that makes strategic stockpiling impossible. You cannot build a meaningful reserve. When supply breaks, the market has no buffer.

Now consider who is competing for that inelastic pool of supply. Semiconductor makers use helium at nearly every stage of wafer production, and there is no substitute for its role in extreme ultraviolet (EUV) lithography, the process that makes the most advanced AI chips. The AI-driven chip market is set to double by 2030 at a compound annual growth rate above 11%. MRI machines make up 20% of the global demand for helium, each requiring an initial fill of 2,000 litres of liquid helium and continuous top-ups throughout their operational life.

As India and other large emerging markets rapidly expand healthcare infrastructure, that demand grows structurally. And then there is the quantum-computing sector, scaling fast and entirely dependent on liquid helium cooling to reach the cryogenic temperatures that make quantum processors function at all.

Rocketry, AI chips, medical imaging, and quantum computing. The three or four fastest-growing sectors in the global economy all compete for the same gas from the same small group of producers, and one of those producers has just been taken off the board for an indeterminate period.

The best helium stocks to buy now

Investors should note that SpaceX's S-1 registration statement, filed ahead of what promises to be one of the most significant listings in a generation, conspicuously omits any reference to risk associated with helium – even though the company has already acknowledged, through its own CEO, that helium supply represents a binding operational constraint on its ambitions to grow. Elon Musk stated plainly that there is not enough helium produced on Earth to sustain a high-flight-rate Starship programme, a constraint significant enough that the vehicle had to be redesigned around it. That is not a footnote. That is a material operational risk.

Consider instead the industrial gas companies Linde (Nasdaq: LIN), Air Products & Chemicals (NYSE: APD) and Air Liquide (Paris: AI), the obvious primary beneficiaries of tighter supply and rising prices, with pricing power that will compound through long-term supply contracts. The UK angle deserves particular attention. Britain has no domestic helium production, no strategic reserve, and no formal critical-minerals designation for helium by the government. The NHS scanner estate, the National Quantum Computing Centre at Harwell and the defence-electronics supply chain are all exposed to a commodity that receives no policy attention in Whitehall. That is a gap waiting to be filled, and investors who identify it before policymakers do will have done so at the right time.

Helium has been treated as background infrastructure for too long, considered too cheap, too abundant, too boring to warrant serious analysis. That era is over. The commodity no one talks about is the one underpinning the launches everyone is watching, the chips powering the AI everyone is funding, and the scanners keeping hospitals operational. The question for investors is not whether helium matters. The question is how long the rest of the market takes to realise it.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

Anthropic, one of the world’s leading developers of artificial intelligence (AI) models, is expected to make its stock market debut next month having filed for its IPO prospectus on 26 October.

The company, which creates the Claude family of large language models (LLMs), is expected to trade on the Nasdaq next month.

Anthropic’s IPO would follow in the wake of SpaceX’s IPO in June, which saw Elon Musk’s space exploration and AI company post the largest stock market debut in history, raising over $75 billion at a valuation of $1.77 trillion.

Anthropic’s IPO is expected to be one of the largest-ever IPO with markets anticipating it could raise as much as $100 billion and with a valuation of around $2 trillion. This would make it the biggest stock market debut in history.

Anthropic's IPO will test investor appetite for AI – we will soon see whether AI is a bubble or a boom market.

"The gloss is beginning to come off some of the AI investment boom, as investors take a harder look at whether the eye-watering sums being poured into the technology will ultimately generate the returns needed to justify valuations," Susannah Streeter, chief investment strategist, Wealth Club, said.

The Claude developer reported revenue of around $16.3 billion in the first half of 2026, but has seen a net loss of £81 billion in the first six months of this year. Plus it has warned that it may need to increase its spending, with commitments amounting to $800 billion until 2045.

"Anthropic is asking ordinary investors to help bankroll one of the largest spending commitments any company has ever made, at a price of around $2 trillion, while warning in the same document its growth may slow and it could need more money sooner than expected," Nigel Greene, CEO, De Vere Group said.

“Frankly, it’s nuts."

When will Anthropic’s IPO take place?

Reports suggest Anthropic will list by mid to late November.

The AI developer filed its IPO paperwork on 26 October and will list on the Nasdaq, under the ticker symbol 'ANT'.

It is not expected to pay dividends and the founders will hold just over 50% of the voting power.

“Public shareholders will be putting in the money without getting control, income or certainty," Greene added.

How to get pre-IPO exposure to Anthropic

Anthropic is a private company and until it lists, you can’t buy its shares directly.

But there are ways to gain indirect exposure pre-IPO by buying shares in companies or investment trusts that have already invested in it.

Amazon (NASDAQ:AMZN) and Google’s parent company Alphabet (NASDAQ:GOOGL) both hold a stake in Anthropic. While Alphabet’s is larger in absolute terms, Amazon’s is higher as a percentage of its own total market capitalisation, so putting your money into Amazon gets you a bigger stake in Anthropic than doing so into Alphabet.

But you’d get an even bigger cut by investing in a surprising company: Zoom Communications (NASDAQ:ZM). While Zoom’s surge to prominence during the Covid pandemic was short-lived, it invested some of the windfall into Anthropic in 2023.

“Zoom has a $3.13 billion stake in the AI company, which is equal to 11% of the communication group’s entire market value,” said Coatsworth. “Therefore, any future valuation uplift in Anthropic could move the dial for Zoom’s share price.”

Coatsworth also identified three investment trusts with meaningful stakes in Anthropic: Baillie Gifford US Growth (LON:USA), Schiehallion (LON:MNTN) and Scottish Mortgage (LON:SMT).

Of the three, Scottish Mortgage has the lowest allocation to Anthropic – with the AI firm accounting for 2.8% of its portfolio as of 31 July, compared to 6.6% for Baillie Gifford US Growth and 5.9% for Schiehallion.

Ceres Power Holdings (LSE: CWR) is one of many companies to have recently got caught up in the global AI boom. Shares in the fuel-cell provider have charged higher by 240% year-to-date. Over the past 12 months, the shares have risen 850%. Despite this performance, the stock is still down around 50% from its ten-year high in February 2021.

UK-based Ceres Power is a developer of solid oxide fuel cells (essentially mini power stations) and hydrogen-power technologies. For many years, the company and its technology were relatively unloved due to high costs and low demand. That has changed over the past 12 months thanks to the massive energy demands of AI, leading to what can only be described as an arms race for power. Data-centre operators and the so-called hyperscalers – Alphabet, Microsoft, Amazon and Meta – are seeking their own power sources as capacity-constrained electricity grids are unable to meet their needs.

Electricity demand from data centres soared by 17% in 2025. The power demand from the average newbuild data centre is forecast to rise to almost 110MW by 2030, from almost 47MW in 2025, according to S&P Global. Globally, data centres could consume 1,000 terawatt-hours (1,000,000GWh) of electricity by 2023, up 100% from 2030.

In November 2024, US utility American Electric Power entered into a landmark supply agreement with Bloom Energy to procure solid-oxide fuel-cell products capable of delivering 1GW of power. Designed to plug directly into AI data centres, the fuel cells can run on natural gas, biogas or hydrogen blends and, most importantly, can be delivered and turned on in months, not years. Data-centre company EdgeCloudLink claims it can construct a data centre and have it running within nine months, partly thanks to fuel cells removing dependency on power utilities.

A turnaround for Ceres Power?

The market seems split on whether Ceres can replicate the success of its US peer. Its record of delivering is spotty to say the least – revenue hasn't grown over the past five years. It is holding out on its next-generation Endura technology, a solid-oxide stack platform that can run on both natural gas and hydrogen. A solid-oxide stack uses ceramic electrolytes to convert fuels into electricity, and Ceres claims its technology can do so more efficiently than its competitors.

At scale, manufacturing costs are expected to be one-third lower than that of its competitors, according to Ceres, and its output aligns perfectly with the electrical requirements and standards of modern data centres. Berenberg analysts describe this as a “gamechanger” that will allow the company to monetise its research and development into a scalable platform.

Peel Hunt analysts believe this is nothing more than a branding exercise, repacking the company's existing technology into a new product. But ultimately, the company's success will depend on its ability to sell licences for its products. Over the past few years, Ceres has transitioned away from a manufacturing model to a licensing model and has outlined a goal of securing at least one new manufacturing licence agreement per year. That doesn't look like a particularly high bar considering the company has identified a 22GW market opportunity for its products by 2030, driven by demand from industry and data centres, particularly in Asia and the Americas.

A single partner scaling up to 1GW of production using Ceres' new Endura technology could generate £50 million to £100 million in annual royalty revenue. Berenberg believes that would generate $1 billion in shareholder value at current sector valuation multiples. Compared to Ceres's current market capitalisation of £1.4 billion, it's easy to see why investors have raced into the stock.

Should you invest in Ceres Power?

The next few years will be key. Analysts have pencilled in revenue of around £80 million for 2028, up from £52 million in 2024 when the company started its transition to a licensing model. This isn't enough to justify the current valuation. Still, these figures were compiled before the company announced a partnership with Centrica (the owner of British Gas) and Delta Electronics (a Ceres manufacturing partner) for off-grid energy generation.

The partnership will offer customers “competitively priced, on-site power generation, significantly reducing exposure to wholesale electricity market volatility and grid capacity constraints”. The plan is to deploy a demonstration site within the next 12 months and scale up capacity over the next three to five years. Other partnerships signed in 2025 are expected to ramp up and start delivering results in 2026, including partnerships with Shell in India, Doosan in South Korea and Weichai in China.

Five years is a long time. The firm is still loss-making when measured by Ebitda and is burning through cash (Ebitda of £7.8 million is projected for December 2028). The shares are trading at 17.5 times forward sales. Still, with £83 million of cash in the bank at the end of 2025, Ceres has the resources to sustain itself for the next three years until it reaches break-even point. The company burned £20 million of cash in 2025 and costs are expected to be down by around 20% this year.

Ultimately, Ceres is a high-risk play and the company's valuation does not leave much room for error if sales fall below expectations next year. However, it's clear there's a large and growing market for the fuel-cell technology Ceres has spent years developing. With 50GW of additional electric supply needed in the UK alone to meet demand from AI data-centre projects in the pipeline, even a relatively small order could transform the company's fortunes.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

By European stock market standards, the size of the SpaceX initial public offering (IPO) will be breathtaking. The company is expected to be valued at between $1.75 trillion and $2 trillion, and given how frothy Wall Street is right now, it would hardly be a surprise if it went to a substantial premium on its first few days of trading. We can all question the valuation. The Starlink business that now provides internet access on flights is a clear money-spinner and it may be able to break into domestic broadband as well, but the plans for a colony on Mars look, to put it politely, a little optimistic. Even so, this is a huge business and a very successful one, and it has created a huge amount of value in a very short period of time.

It is far from alone. Anthropic, the company behind Claude AI, is reported to be planning an IPO in October, with a valuation of $1 trillion or perhaps more. Its rival OpenAI, the company behind ChatGPT, is also expected to list later this year, with a value of close to $1 trillion. There are slightly smaller companies just behind it. Last week, Cerebras, which makes AI chips, made its debut on Nasdaq, and after a first-day premium, saw its value soar to $95 billion. On the US market, incredible amounts of wealth are being created at dizzying speed. Anthropic is only five years old, OpenAI is ten (its profit-making unit only five) and although SpaceX was founded in 2002, it only really got going a decade ago.

The contrast with European stock markets is painful. SpaceX by itself will be worth almost as much as the whole of France's CAC-40 (valued at €2.6 trillion and falling rapidly as the value of LVMH slumps). It will be getting close to the entire value of Britain's FTSE 100, currently valued at £2.4 trillion, and SpaceX and Anthropic combined will certainly be worth more than all of the UK's 100 largest companies put together.

European stock markets need more mavericks like Elon Musk

The reason is clear. Very few new firms are being created. If you exclude mergers, the newest company on the CAC-40 is Eurofins Scientific, which was formed in 1987. Even where there are new companies, the best ones choose to list on Wall Street – the Cambridge-based chip designer ARM, for example, is now worth $220 billion, which would rank it as the third largest in the FTSE 100 if it had decided to list here.

Europe, including the UK, needs to realise how far behind it has fallen and start working out how to turn that around. First, it should radically reduce the taxes on start-ups to encourage more entrepreneurs. Britain has scaled back the break on capital-gains tax that anyone who started a new company used to benefit from, and most of Europe never had any concessions to start with. Instead, there is a constant stream of new wealth taxes and capital-gains taxes, with the Netherlands extraordinarily planning to tax capital gains before they have even been cashed in. No wonder there are far fewer start-ups and hence fewer giants ever emerge.

European stock markets should also roll back restrictions on growth industries such as AI and space. While the US has a booming space industry, Europe has a Space Act; while huge new AI businesses are created on the other side of the Atlantic, Europe is stuck with an AI Act. But there is no point in having a regulator if there isn't an industry to make rules for. There is still little sign that politicians in either Brussels or London realise how much damage has been done by trying to regulate industries before they have even begun.

Finally, Europe should relax the listing rule for entrepreneurs such as Elon Musk who want to keep control of companies. SpaceX will come in for a lot of criticism for allowing Musk so much control over the business and the $1 trillion pay package if he manages to create a thriving human colony on Mars. It doesn't follow Europe's governance rules. But so what? Entrepreneurs are often a little odd, and they are often control freaks, but they also have the drive and ambition to create huge new businesses. Europe could use fewer rules and more mavericks if it is to avoid turning into an investment backwater, with nothing more than a dull collection of very old companies.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

What force will shape the world in the next 20 years? The answer is China's drive to financialise, according to Diana Choyleva, founder and chief economist at Enodo Economics.

In this episode of the podcast, Diana speaks to MoneyWeek's Cris Sholto Heaton about how the AI race differs in China versus the West, the transformation of the country's equity market, and the breakdown of globalisation.

You can watch this episode on our YouTube channel or subscribe to it on any podcast platform.

About the podcast

MoneyWeek Talks is a podcast that helps you unlock the secrets to financial success. Editors Kalpana Fitzpatrick and Andrew Van Sickle are joined by influential guests – from CEOs and entrepreneurs to economists and policymakers – to share their top tips on managing money, investing wisely and building wealth.

Subscribe to the MoneyWeek Talks podcast and get ready to make it, keep it and spend it with confidence.

In February, major US brokers such as Marsh & McLennan , Willis Towers Watson, Aon and Arthur J. Gallagher & Co. declined by 8%-11% in a single day on news that OpenAI, the owner of ChatGPT, had approved the first insurance-focused AI app designed by US intermediary Insurify. OpenAI's tool arrived a week after Claude, the model developed by Anthropic, also released a new plug-in to automate sales, legal and financial analytical functions.

As well as this risk of obsolescence, investors have had to try to factor in the growing challenges of a so-called “soft” insurance market. Since 2017, the insurance market has been in a “hard” phase, where prices have been rising, and profits have jumped. However, two years ago, prices started to flatten and then fall as the market turned from hard to soft. The hard market was very good to Marsh & McLennan (NYSE: MRSH). Between 2017 and the beginning of 2024, shares in the broker and global-intelligence company rose more than 200%, but since topping out in the first half of 2025, they have fallen by around 28%.

This decline was needed. Insurance is a highly cyclical business and, coming into the soft cycle, Marsh was trading at about 22 times forward earnings, a multiple that left little, if any, room for error. After the re-rating, the shares are now trading at just 16.8 times forward earnings, according to UBS. That is still a bit high for this stage of the market cycle, but there's a good argument that you should start buying the shares at this level.

Marsh & McLennan plays an important role in the insurance market

Marsh & McLennan is the parent firm of Marsh Inc, one of the world's largest risk-management and insurance-services firms. The group also owns Guy Carpenter, a risk-management and reinsurance specialist; management consultancies Mercer, plus Oliver Wyman and Jardine Lloyd Thompson Group (JLT), an insurance, reinsurance and employee-benefits broker based in London.

The group's largest division is Marsh Risk, which generated $14.4 billion in revenue in 2025. The second largest is Mercer, with revenue of $6.2 billion, and Marsh Management Consulting, at $3.6 billion. The total consulting business turns over $9.8 billion a year. The risk-management and insurance businesses (including Marsh Risk) generated $17.3 billion in revenue.

Marsh Risk plays an important role in the insurance market. Large, complex risks are often underwritten by pools of insurers and reinsurers bought together by brokers. The insurers like it because they're not overly exposed to a single risk, and the buyer of insurance likes the security of multiple parties underpinning the contract. Marsh Risk helps navigate this process. The company also manages the claims process, which most large insurers and reinsurers don't have the capacity to handle, as it can be costly and time-consuming.

For example, Marsh has helped set up a clean hydrogen insurance facility, where developers pay an insurance premium, and Marsh negotiates insurance coverage with insurers to transfer the risk from the project owners' balance sheets. Investors (in this case, the insurers) provide capital investment for the projects, with their risk exposure mitigated by tailored insurance coverage. In the event of insured incidents, Marsh manages the claims process. The single platform helps lower costs for all involved.

It's hard to imagine a world where AI disrupts this process. It will certainly help streamline the paperwork, but the human touch of the Marsh brokers will always be required to navigate deals among key stakeholders. This business is highly profitable and cash-generative. The insurance arm booked an adjusted operating margin of 32% last year, compared with 21.1% for consulting. Of the $7.3 billion in adjusted operating income, $5.3 billion fell to the bottom line as operating cash flow. The company's return on invested capital, a measure of profit for every £1 invested in the business, is 25%.

How Marsh is embracing AI

The real AI threat is to Marsh's consulting arm, but even here, the company claims it is addressing the potential risk. In the company's first-quarter results, it said Oliver Wyman's AI Quotient, which helps firms deploy and scale AI tools, has advised on upwards of $50 billionin AI investment. This helped the consulting arm outperform in the first quarter, with Oliver Wyman recording revenue growth of 6%, ahead of group top-line growth of 4%.

Management is deploying these tools internally to help reduce costs. It's targeting a total of $400 million in savings over three years and has logged a 20% improvement in efficiency through AI-powered document processing. Other tools have saved an estimated one million hours of the team's time in the first year. UBS estimates this could help drive Marsh's return on invested capital to near 30% by the end of the decade.

So while the market frets about the risk AI poses, the company is quietly leveraging the technology to enhance its own services. This suggests that, if anything, the firm is an AI play.

Marsh's most important assets are its people and technology, and while it spends heavily on both, overall capital spending requirements are low. As a result, most of the cash generated from operations converts to free cash flow. Management has set out to return as much cash as possible to investors. At the end of last year, management authorised a $6 billion share buyback, with $750 million deployed in the first three months of the year. While the market was selling, Marsh was buying its own shares.

Cash flow is the firm's most attractive quality. While the shares might not look too cheap on a price-earnings basis, according to UBS, the shares are trading at a forward free cash-flow yield of 6.2% for 2026, 6.7% for 2027 and 8.1% for 2030.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

At Janus Henderson's North American Income Trust (NAIT) we focus on US income stocks – quality franchises that consistently generate cash and have disciplined capital-allocation policies focused on investment in the business to sustain competitive advantage while paying a progressive, covered dividend. Surplus cash beyond this may be used for bolt-on mergers and acquisitions, or to repurchase shares when the stock is dislocated from long-term assessments of fair value. The NAIT has a strong record of paying a progressive dividend and growing revenue reserves since the fund's inception in 2012 (it was converted from the Edinburgh Tracker Trust). The average dividend in the portfolio is 3% and dividend growth averages an attractive 6%-7%.

Our revenue reserves can comfortably cover one year of payouts and may be used if needed. However, there was only one small dividend cut during the 2020 pandemic period and none since then. Many UK investors may not automatically think of US income stocks, but there are several that offer attractive and growing dividends. The US has a history of superior earnings growth, which can often translate into higher dividend growth, too.

How to gain exposure to US income stocks

Dell (NYSE: DELL) is a technology infrastructure company uniquely positioned to grab a slice of the next wave of corporate spending on AI applications. Its scale, global supply chain and deep relationships with customers from the private and public sectors make it a preferred supplier of AI servers and data-storage technology. As enterprises move from experimentation to deployment, Dell will benefit from recurring technology update cycles. Growing profitability is supported by the company's shift toward higher-value technology infrastructure and its disciplined cost management. Debt has been cut and capital returns support the yield. We believe Dell's valuation fails fully to reflect the durability of demand and the firm's exposure to capital expenditure on AI.

Johnson & Johnson (NYSE: JNJ) is another US income stock that offers a rare combination of earnings quality and durable growth. Following the spin-off of its consumer-health division in 2023, it is a focused, innovation-driven pharmaceutical company and a leader in medical technology that should comfortably deliver mid-single-digit revenue growth. It has a diversified drug pipeline, which reduces risk, and its franchises in oncology, immunology and cardiovascular treatments are best-in-class, which will support cash flows in the long term. The medical-technology sector is growing strongly and the worst seems to be behind the company when it comes to legal issues. This is restoring investors' confidence and valuations. With a strong balance sheet, consistent free cash flow and a long record of dividend growth, Johnson & Johnson remains a core holding in volatile markets.

Morgan Stanley (NYSE: MS) is a global leader in the capital markets. Its earnings have become more resilient following a strategic pivot toward wealth and investment management, which generates stable, fee-based revenues. These annuity-like income streams provide downside protection while preserving upside exposure to appreciation in the markets and net asset inflows. The firm's strong capital position is enabling it to buy back shares and grow dividends. We believe Morgan Stanley's improved position will deliver impressive gains tied to long-term growth in the financial markets.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.

“It's a market of stocks, not a stock market” is an old cliché, intended to remind us why investing is ultimately about how well individual companies are doing from the bottom up and not a top-down view of the FTSE 100 or S&P 500.

I don't entirely agree with this thinking, at least in the modern world. The growth of index investing has meant that many people now invest in the whole market or in broad sectors and don't care about the companies they hold. Money flowing in and out of funds can do more to determine whether valuations rise or fall than real changes in a business's fundamentals.

Still, it is always important not to let big-picture fears blind us to how well individual stocks are doing. If most companies are seeing robust earnings growth from the bottom up, it is likely that the overall index will keep going up. And right now, the reality is that earnings growth remains very strong in the US.

The year-over-year blended growth rate (ie, including both results reported so far and latest estimates) for the S&P 500 is currently 15.1%, according to FactSet – the sixth successive double-digit quarter. The index is expensive: at just over 7,100, it's on a trailing price/earnings ratio of 28. Yet if earnings keep compounding like that, it's not really a stretch to stay bullish.

The greatest threat to earnings growth

In the medium term (maybe three to five years), one has to wonder whether giant companies can continue to earn such high margins: the S&P 500 net margin is once again setting a new record of 13.4%. The geopolitical and political trends that let businesses – especially multinationals – become ever more profitable over several decades are shifting. Maybe this goes into reverse. But a few years is a lifetime in the markets and we are obviously not there yet.

In the shorter term (maybe a year or two), the extent to which AI mania is underpinning this boom cannot be ignored. In the tech sector, earnings growth is at 46%. There is a very fine line to be walked here: if all this investment does not bring huge productivity gains, it will grind to a halt. If it puts too many people out of stable employment, the political backlash could be equally dangerous. Yet all investors care about is what will happen in the next couple of quarters, and there is no sign of the boom letting up so far.

So what is the greatest ultra-short-term threat? Energy. The amount of oil at sea when the Middle East crisis started means that the consequences of the closure of the Strait of Hormuz and the shutting in of millions of barrels a day of crude is not really translating into shortages yet. Even if supplies resume tomorrow, there will be a lag and the effects will still show up over the next couple of months. But if they do not resume soon, the crunch is going to become very evident. A market focused on historic earnings and understandably upbeat forecasts is not pricing that in.

This article was first published in MoneyWeek's magazine. Enjoy exclusive early access to news, opinion and analysis from our team of financial experts with a MoneyWeek subscription.