Is ChatGPT Going Bankrupt?

Image credit: Yuichiro Chino via Getty Images

“The only winning move is not to play”. These were the words of investor Michael Burry when talking about the staggering rise of tech companies' stock market values caused by the revolutionary progress and promises of AI. At first glance, hearing the professional investor and former hedge fund manager discourage the general public from investing seems surprising. But Michael Burry is not only known as the cliché of a high-profit-seeking capitalist. For those of you who like cinema, he is at the centre of the 2015 movie “The Big Short” ,which tells the story of how he predicted the 2008 housing crisis, and how he hugely profited from it. Since then, and particularly thanks to this movie, Michael Burry has gained a prophet-like reputation of being able to decipher (some of) the inscrutable ways of the market.

Among the “heavyweights” of finance, he is not the only one to express concerns. Back in November, Jamie Dimon, the CEO of the world’s biggest bank, JPMorgan Chase, also warned about some assets which he considers to be “entering bubble territory”. The CEOs of Goldman Sachs, Citigroup and even the Bank of England further emphasized the risks. So what is happening with the stock market? What are these big bosses of the finance world afraid about? And are we on the verge of a new edition of a 2008-like crisis?

A capitalist’s safe place 

The source of all these fears of the growth of AI may be surprising at first. The stock market is indeed supposed to react to new innovation by increasing the value of the companies driving it. For example, Google and Meta are important companies creating and developing new AI models and their value has increased since the beginning of AI in 2022. Similarly, companies like Nvidia or Intel sell the microchips essential for the functioning of AI models and make more profit than ever. To economists, none of this is surprising. Innovation leads to growth and the companies behind it are the winners.

When ChatGPT, the first large language model (LLM), came out in 2022, it landed like a bomb, just like the beginnings of the Internet and smartphones did in the early 2000s. In both cases, investors saw the potential of a new technology to make them gain big. The growth of tech companies thanks to the web led to the “Dot-com bubble”, a big crash of the stock market leading to many bankruptcies and losses for investors. AI may therefore have the same important impact on the market and be the new  “Dot-com”.

Indeed, after 4 years of steady growth of the tech companies involved in the AI sector, some experts now start to question the rationality of these super-valuations and believe these companies may be overvalued. In 2000, many hopes were put on Internet companies and their growth in the upcoming years. Many of them merged, managed to buy ad spots for the Superbowl and made huge marketing investments. The ideas of “growth before profit” and “get big or get lost” dominated the landscape. But as the market kept reflecting a merely “expected” value, and this value kept rising without being anchored in the true revenues and profits of these companies, the bubble expanded. To put it simply, a bubble happens when the emotions of the market and the investors, who become over-optimistic and overly enthusiastic about the value of a stock, take over rational and intrinsic valuation of said stock. 

In the early 2000s, the bubble was bigger than ever and rested in an unstable equilibrium. In March 2000, the Nasdaq index, which reflects the global value of the technology sector, peaked at more than 5,000 points. However, a series of events in the following months finally burst the bubble and started a free fall until it reached 1114 points. A 78% fall in market value. The Nasdaq would only cross back the 5,000 points threshold in mid-2015.

Even though the numbers sounded catastrophic, the general public did not feel much of this crash on their grocery bills or on their general expenses. The big losers of the bubble were the investors who had granted a lot of capital to tech-related companies and the said companies, half of which went bankrupt due to the crash. Although the growth of the US economy slowed down, other indexes like the famous S&P 500, which reflects the value of the 500 leading US companies with the largest capitalization, actually gained as investors switched back to more established stocks. Seeing these relatively mild consequences of the crash, many economists tend to talk about a “dot-com correction” rather than a huge crisis. If the market was overvalued, it is more beneficial for it to reflect reality.

A small bubble never killed anyone. . . but a big one could

Today, the potential consequences of an AI bubble may be more worrisome. The bubble that may be before us is much bigger than the dotcom one. Warren Buffett, considered the most successful investor ever, imagined an indicator to assess this. The “Buffett indicator” measures the value of the S&P 500 as a percentage of the total US GDP. To take globalization into account, it considers that the market is overvalued if this number crosses 115%. During the dot-com bubble, it reached around 160%. The release of ChatGPT made it rise back to around 180% a year ago. On October 11, it reached a staggering 303% of GDP which, even if we take globalization into account, is disproportionately high. 

In its November 2025 article, The Economist dug deeper into the division of the shares of the market among its biggest companies. Their analysis gives another warning signal: 

“In 2000 the 20 biggest firms on the S&P 500 made up 39% of its total value. Eleven of those were internet-related companies. Today the top 20 account for 52%, with the same number deeply invested in AI. If the tech fails to deliver juicy returns they would all be hit hard, along with a large number of Americans. “

Beyond the fact that this bubble may be much bigger, its burst may be more devastating for American households, which now hold a larger part of their wealth in the stock market compared to 2000 (+ 4 percentage points). With the spread of online trading platforms like TradeRepublic, eToro or Revolut, investing has become easier than ever and this makes more people vulnerable to market variations. In addition, these numbers do not take into account indirect holdings such as what is invested in retirement funds or life-insurance companies. The growth of AI also puts pressure on the US energy system and may be the source of half of the growth in electricity demand in the country by 2030, compared to 8% for households, as Goldman Sachs and the Bank for International Settlements state. This pressure raises the risks of inflation in the economy and is set to become the main financial burden for tech companies in the future. The joint cumulative future cost of data centres is expected to reach $8 trillion over the next six years according to Stijn van Nieuwerburgh, economist at Columbia University. On the other hand, the productivity gains of AI have a deflationary potential that could offset this if they arrive.

A threatening Chinese needle

On the other side of the Pacific, the competition to American AI systems is getting more efficient and cheaper, although with less impressive market valuation. The rise of Chinese AI models like DeepSeek or Kim K3 is creating earthquakes in Silicon Valley.  First considered as largely underperforming American models, their only advantage was that they were cheaper . “On the 17 of July, we learned that Kim K3 is close to overperforming all international competition”, writes Le Monde diplomatique. The journal further underlines how the tokens (the virtual “money” used to execute tasks on AI models) of the latest Chinese models may be 110 times cheaper than American ones. This gap between the two superpowers may soon become, for some economists, the needle that will make the bubble burst. Indeed, companies like Apple are starting to look towards Chinese AI models precisely because of this cost gap. Some European companies may follow.

The Chinese financing of AI differs a lot from the American one, as it is largely state-controlled, which enables more long-term visibility and less risk to the health of these companies. On the other hand, valuations on Wall Street depend on rallying forces around growing yields (leading to potential bubbles) and run away from panics (leading to market crashes ). The choice for AI investors is therefore between political selectivity in the Chinese system or the selectivity governed by market yields in America. 

Neither system is better than the other, but uncertainty is the common denominator. Hence the words of Michael Burry: “ The only winning move is not to play”. Sure, but many mock such prophets of the AI Apocalypse: “Michael Burry successfully predicted 72 of the last two crises”. What if the hype is justified? The gains of AI may indeed be real. Investors keep on looking at the quarterly reports of companies like Nvidia. Every four months, the expectations are high. On the 26th of August this year, the Q2 report of Nvidia was strong, and investors were reassured. Is there a limit? After the numerous economic crashes in history, investors feel like they keep forgetting that there may be a limit to the sky. To see how high it is still remains to be seen.