Could the AI megacap bubble burst?

Tech stocks were at the forefront of a US stock market retreat on Tuesday (19 August) as the S&P 500 fell 0.6%. Artificial intelligence (AI) megacap stocks like Nvidia and Meta saw the biggest falls: Nvidia shares fell 3.5% while Meta’s fell 2.1%. AI bulls’ favourite, Palantir, fell 9.4%.

We’ve become used to the tech megacaps being the top stocks for investors over recent years, with the Magnificent Seven seemingly the sole drivers of stock market returns, but the market is showing signs of jitters over stretched valuations.

“Tech stocks came under real selling pressure last night [19 August],” said Steve Clayton, head of equity funds at Hargreaves Lansdown.

Subscribe to MoneyWeek

Subscribe to MoneyWeek today and get your first six magazine issues absolutely FREE

Get 6 issues free

Sign up to Money Morning

Don't miss the latest investment and personal finances news, market analysis, plus money-saving tips with our free twice-daily newsletter

Don't miss the latest investment and personal finances news, market analysis, plus money-saving tips with our free twice-daily newsletter

Sign up

The decline followed reports that Sam Altman, CEO of OpenAI, thinks the AI market is a bubble.

“When bubbles happen, smart people get overexcited about a kernel of truth,” Altman is said to have told reporters at a dinner on Friday 15 August. “Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes.”

Altman also said that he simultaneously believes that AI is “the most important thing to happen in a very long time”. Comments like these, though, from one of the AI boom’s most central figures, have only fuelled concerns that the AI sector is wildly overpriced.

Could a recovering US economy end megacap dominance?

As well as Altman’s comments, the sell-off was preceded by the publication of a report by Bank of America analysts that suggested a change in market narrative could be on the way.

The Bank of America report highlights that various leading economic indicators point to the US economy leaving its ‘downturn’ phase and moving into ‘recovery’.

It might sound counterintuitive that an uptick in the health of the US economy would be a negative for its biggest stocks. But Bank of America’s model, which breaks the market cycle into four phases (recovery, mid cycle, late cycle and recovery) shows that downturn phases tend to favour megacaps with dominant market positions and superior earnings power.

The recovery phase, however, tends to see companies like these lag compared to the broader market.

Historically, looser monetary policy from the Federal Reserve (Fed) “has been accompanied by Mega caps lagging more than leading, and higher inflation should support a broadening of the S&P 500 beyond defensives/secular growth”, said Savita Subramanian, head of US equity and quantitative strategy at Bank of America.

An interest rate cut from the Fed at its next meeting would strengthen the view that the US economy has shifted from downturn to recovery.

Is AI a bubble?

The Bank of America report also noted that the ‘nifty fifty’ – the fifty largest stocks in the S&P 500 – outperformed the rest of the index by 73 percentage points over the last ten years.

“The last memorable nifty fifty run was the 1990s lead up to the Tech Bubble,” said Subramanian. “It was about six years long, and the outperformance vs. the S&P 500 index was 71 percentage points.”

She added that “today’s nifty fifty is lower quality than most of the 1990s” and “sports average long-term growth expectations despite near-record valuations”.

This isn’t tech-specific. Subramanian highlighted that “tech is not the culprit” and that sector-neutral analyses showed that it is large market caps rather than sector alignment per se that underscores the similarities. All the same, though, as far as megacaps are concerned “the run may be done”.

That of course shifts focus onto the huge multiples at which tech and AI megacaps are currently trading. These valuations only make sense in a world in which these stocks continue to strongly outperform the market.

Bill Bonner, financial author and writer of the substack Bonner Private Research, recently highlighted that the market caps of the Magnificent Seven combined are worth more than China’s GDP. These companies are pouring money into AI projects that may not generate anywhere near sufficient revenue uplifts to justify the investment.

AI’s long road to profitability

If over-investment in a technological trend that doesn’t deliver the implied profits sounds familiar, that’s because we saw the same thing happen in the late 1990s.

The counterargument to comparisons between the AI boom and the dotcom bubble is that the latter was characterised by a proliferation of unprofitable business models, while the former is led by hyper-profitable companies like Nvidia and Microsoft.

This overlooks that Microsoft’s profits come predominantly from its hardware and cloud divisions. Nvidia, like Microsoft’s cloud arm, is driven by AI spend. But in both instances, it is companies developing AI that are doing the spending. No-one is arguing that lots of money can’t be made by selling the kit to build AI: it's the products of this buildout that have a profitability question mark.

A glance beyond the AI megacaps reveals a landscape that looks an awful lot like the 1990s. ‘AI-powered’ start-ups are everywhere, but are barely anything more than cash incinerators.

“Neither Anthropic nor OpenAI is anywhere near profitability,” says Kenneth Lamont, principal at Morningstar. “Growth has taken priority, and while both generate strong revenues, the vast costs of renting or building cutting edge infrastructure dwarfs their income.”

Altman hopes that OpenAI will reach profitability by 2029, but this implies top-line revenue increasing tenfold over the next four years, and for enterprise sales to increase even faster (ChatGPT’s consumer uses are, and probably always will be, loss-making).

That isn’t to say there won’t be any long-term winners from AI, even if it is a bubble. “Framing the debate as either a bubble set to burst or the dawn of a lasting transformation misses the point,” says Lamont. “History shows it can be both: the dot-com boom left behind plenty of failures, but also the tech giants that still dominate today.

“That AI will play an ever-larger role in the economy is not in doubt. What is less certain is which companies will capture the gains.”

Protecting against the AI bubble bursting

You are likely more exposed to any potential bursting of the AI megacap bubble than you think. Most of your portfolio, whether that’s investments in a stocks and shares ISA or your pension, will likely comprise index funds that track the global stock market. If sentiment towards these stocks dips your portfolio could take a hit, at least in the short run.

One of the drawbacks of passive investing using index funds is concentration risk. These funds track the movements of the broader market: they naturally put your money mostly into those megacap stocks that have the highest weighting within global indices. But that naturally exposes you to stretched valuations. The effect is also circular, as every index fund for a given market has to buy the stocks that gain the most (further pushing up prices).

You can reduce your concentration risk by buying equal-weight funds, which will cap every holding at a certain weighting of the portfolio. Lamont highlights Xtrackers S&P 500 Equal Weight UCITS ETF (LON:XDWE) as an example.

Or alternatively, you could buy a fund that excludes the megacaps altogether, such as Amundi MSCI USA Ex Mega Cap UCITS ETF (LON:XMGA).

Investors that want to retain some exposure to AI without being over-exposed to the stretched valuations of the megacaps could look for value opportunities in AI and its broader ecosystem, such as the suppliers to companies like Nvidia and TSMC.