ChatGPT turns two: how has it impacted markets?

ChatGPT was launched to the general public on 30 November 2022. For two years since, the generative artificial intelligence (AI) chatbot has had the attention of the world’s investors in a vice grip. The world’s top investment funds have scrambled to add companies involved in AI to their portfolios in the meantime, and sent their valuations through the roof.

“ChatGPT’s launch marked a paradigm shift in AI adoption,” Rahul Bhushan, managing director at ARK Invest Europe, tells MoneyWeek. “We project that generative AI could drive a step-function increase in productivity, contributing significantly to global economic growth by 2030.”

The impact of ChatGPT on the stock market has been similarly profound. While OpenAI, the start-up that develops ChatGPT and related tools like DALL-E, is private and therefore not traded on stock exchanges, the increased public awareness of AI that ChatGPT triggered has catapulted companies in the space to new heights.

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

For no company is this more true than Nvidia (NASDAQ:NVDA). Nvidia’s earnings have increased by 2,600% in the two years since ChatGPT’s public launch, thanks to its status as the leading developer of high performance graphics processing units (GPUs) that are used to train the cutting edge AI models that underpin platforms like ChatGPT.

This has led to Nvidia’s share price gaining 700%, and it becoming (temporarily) the most valuable company in the world by market cap. (At time of writing, Nvidia is second behind Apple (NASDAQ:AAPL), though Nvidia has held the top spot several times over recent months.)

“The speed and scale at which generative AI has been embraced by companies and the general public is remarkable,” says Dan Coatsworth, investment analyst at AJ Bell. “Investors have made a lot of money from backing AI enablers over the past few years and there are now many examples of companies employing generative AI with great success.”

While the launch of ChatGPT has created some behemoths, it has threatened to disrupt others.

“One of the most important developments is ChatGPT creating the first serious challenge to Google’s search empire in a long time,” says Coatsworth. Shares of Google’s parent company Alphabet (NASDAQ:GOOGL) have fallen 3.3% over the past six months, rocked on the one hand by a string of antitrust cases, but also by the threat that ChatGPT poses to its core search engine dominance.

“While Google’s general purpose search engine is by far the dominant source of obtaining information online, ChatGPT recently introduced a new service that provides a different approach to searching,” says Coatsworth. He adds that Apple’s incorporation of ChatGPT into its own AI services “is another blow to Google”, as Apple devices have historically been a reliable source of traffic to its search engine.

How does ChatGPT work?

ChatGPT is a software program based on GPT, which stands for Generative pre-trained transformer. GPT is a type of large language model (LLM), a machine learning model that, effectively, learns to interpret language by analysing the statistical relationships between words in massive amounts of input text.

There are various versions of GPT. The initial version of ChatGPT launched two years ago was based on GPT-3, but two new versions – GPT-3.5 and GPT-4 – have been released since.

All these models thrive on data. The more that is fed in, the theory goes, the better they become.

As such, the success of ChatGPT “has catalysed the integration of generative AI into workflows across industries”, says Bhushan, “creating a powerful data flywheel that continues to refine OpenAI’s models”.

He adds that ChatGPT now has its own ecosystem through the GPT Store, which “enables developers to expand use cases, creating a virtuous cycle of innovation and demand for gen AI solutions”.

What is the next version of ChatGPT, and when will it launch?

OpenAI is working on GPT-5, but it isn’t expected to be released this side of the new year.

OpenAI’s CEO, Sam Altman, told an AMA on Reddit that the increasing complexity of the models OpenAI is building had led to delays in release. He said the company is working on some “very good” releases, but “nothing that we are going to call GPT-5”.

However, OpenAI has soft-launched a model that could give clues as to the next generation of ChatGPT’s capabilities. In September, it launched OpenAI o1, a new series of AI models which, it says, can reason through complex tasks and solve advanced problems in science, coding and maths compared to previous models.

This is a big deal in terms of the applicability of ChatGPT, and generative AI more broadly, to pressing real world problems. According to the MIT Technology Review, while previous iterations have demonstrated the ability to interpret and generate text, “such LLMs have failed to demonstrate the types of skills required to solve important problems in fields like drug discovery, materials science, coding, or physics”.

OpenAI o1, though, represents “one of the first signs that LLMs might soon become genuinely helpful companions to human researchers in these fields”.

How much is OpenAI worth?

In early October, OpenAI was valued at $157billion in a funding round that raised $6.6billion.

Unfortunately you can’t invest directly in OpenAI as an individual investor, as it is currently a private company.

OpenAI might seek to list on a stock exchange through an IPO at some point in the future, though because it is currently technically a non-profit organisation it would first need to change its structure in order to do so. OpenAI has promised its present investors that it will achieve this change within two years, or return capital to them.

Altman has, however, expressed concern that being a public company could restrict OpenAI’s ability to develop “superintelligence”, as investors might not agree with some of the decisions it took as a company in doing so.

There are less direct ways of investing in OpenAI. US-based investors can access the stock through one of several venture capital funds, though at present none of these are available to UK investors.

However, Microsoft (NASDAQ:MSFT) is a significant stakeholder in OpenAI, having invested significantly into it over the last two years. Buying Microsoft shares will therefore give investors some indirect exposure to OpenAI.

What are ChatGPT’s limitations?

First and foremost, ChatGPT is incredibly expensive to build.

While OpenAI’s revenue has increased 1,700% since the beginning of 2023, to $300million per month (as of August), the New York Times reported in September that it expects to lose an eye-watering $5billion this year. This is due to the cost of running the data centres and other services that train the AI models, as well as paying some of the most talented software engineers in the world to design them in the first place.

“It’s right that investors have asked when we’ll see positive financial returns from AI investments and we’re now getting a clearer picture of how the technology can improve earnings,” says Coatsworth. He adds that the onus on this front is for the companies using ChatGPT to quantify how its usage is impacting their business performance.

On this front, though, there are concerns that the productivity gains it promises essentially amount to displacing human labour.

In February, buy now, pay later group Klarna said that its AI assistant, which is powered by ChatGPT, was doing the work of 700 full-time customer support agents. Given the ability of OpenAI’s latest models to reason deductively, newer versions could begin to encroach more and more on skilled, white-collar workers.

“One might assume [ChatGPT] has been responsible for countless job losses, but many companies say its use has merely freed up staff to work on other roles rather than causing redundancies,” says Coatsworth. “Whether that’s the case in five years’ time remains to be seen.”

There are also environmental concerns over the vast amounts of energy and water that AI data centres require to power and cool, as well as over its ability to aid hackers and cybercriminals.

Finally, not everyone is bullish on the abilities of ChatGPT. Two years on from its initial release, ChatGPT still includes factual errors in its responses.

Further, a study published by Forbes in July found that 77% of employees using AI said that it had added to their workload and hampered productivity. Altman and co will of course hope that these results will improve with more advanced models, but at present, the promise of generative AI might not quite live up to reality.

As ChatGPT said when we asked it to summarise the market reaction to its first two years since launch:

In summary, the market reaction seems to be a mix of optimism regarding AI’s potential and cautious consideration of its challenges and risks. The next few years will likely continue to see this dynamic evolve, with more breakthroughs as well as calls for responsible deployment.