When someone constantly predicts that the market will fall, it’s only natural that they’ll be proven right at some point, since stock market cycles are characterized by ups and downs. This may happen years later, and therefore their prediction doesn’t carry the same weight.There is no doubt that the current stock market rally is based to a significant extent on the expectations generated by AI. However, a report from MIT last year revealed that 95% of pilot programs involving generative AI were not turning a profit. Ray Dalio, the founder of Bridgewater Associates, echoed this sentiment, claiming that we are at about 80% of the level seen during the 1929 stock market mania. The last time he mentioned this was in January 2026.
He has pointed out that in the final stages of a stock market bubble, there is a noose. For proponents of the AI stock market bubble theory, as they call it, the revenue noose is obvious. Major tech companies are investing in AI, and AI companies are buying cloud services from these major tech companies—and sometimes chips as well. The tech giants claim their growth is driven by AI, and their stock valuations are skyrocketing.
Microsoft invested $13 billion in OpenAI (ChatGPT), and OpenAI returned the investment by purchasing cloud services through Azure. Microsoft reported the amount as revenue. This is the “loop,” as it’s called. Something similar happened between Google and Anthropic (Claude), this time involving $40 billion. No one disputes the technology (AI) and its capabilities. However, money circulates, and the investment community interprets this as demand.
The legendary “short-seller” Jim Chanos revisited the issue of actual demand behind the AI boom, drawing a parallel between the period from mid-1998 to mid-2000 and the S&P 500’s gains 500’s earnings and the current situation. What did Chanos argue? The S&P? 500’s earnings rose by 30% during that period—faster than in the early 1990s—and the investment community believed a new era had dawned. However, S&P 500 earnings plummeted by 40% over the next 12 months because many purchase orders—for routers, for example—were canceled, and corporate revenues evaporated while costs remained high.
The collapse in earnings was similar to that of the Great Depression, when the U.S. economy—and others—came to a standstill. According to Chanos, the market misinterpreted the investment boom by telecommunications companies and others as genuine demand, but the actual demand for the internet was much smaller—for example, it was doubling almost every year rather than every quarter, as the market had assumed.
Chanos argues that the AI boom exhibits similar characteristics. Are S&P 500 earnings rising because many of the billions of euros spent on investments are returning as revenue and profits to a small number of companies? It’s a trap. The market rewards these companies on the stock market because their growth is impressive. However, a significant portion of that growth stems not from long-term demand but from their capital investments.
Just as the internet succeeded, so will AI. The question is whether such massive investments, based on optimistic demand scenarios, are really necessary. Once again, investors are assuming that growth proves everything is going well, but the biggest stock market bubbles formed because of unrealistic expectations. Jim Chanos and Ray Dalio seem to believe that the same mistake is being repeated.