AI's Financial Fever: Roscongress Report Warns of 2026 Reckoning

A new report from the Roscongress foundation sounds a stark warning: the global market for artificial intelligence companies shows clear signs of a financial bubble, with the odds of a significant downturn this year now above 50%. The analysis, titled 'The AI Economy: Stock Market Hype, Project Payback, and Energy Deficit,' suggests only the relatively soft monetary policy of the U.S. Federal Reserve has kept the bubble from bursting already.

The study applies classic bubble criteria—inflated asset valuations, excessive concentration, over-investment, and high leverage—and finds the AI sector meets at least three. Evidence points to a market frenzy. Since early 2024, the S&P 500 index gained about $7.5 trillion in value; a staggering $4.9 trillion of that came from just 17 AI-related firms. This surge has pushed U.S. tech stocks to a historical threshold where, based on patterns from the 1970s gold rush to the dot-com boom, a crash becomes more likely than not.

Investors continue buying, a behavior the report's authors describe as a kind of 'financial mania.' They highlight a dangerous cycle: chipmakers invest in their clients, who then use that money to buy more chips, while cloud providers take on debt secured by equipment that may become obsolete before loans are repaid. A significant portion of these investments, the report predicts, will never pay off.

The first half of 2026 is pinpointed as a critical period when AI firms must demonstrate sustainable revenue growth to justify massive capital outlays. While historical precedent suggests a spike in interest rates typically triggers a collapse, current U.S. policy remains accommodative. A leadership change at the Fed in May 2026, with incoming Chair Kevin Warsh potentially viewing AI as a deflationary force, could extend this softer stance.

Yet even with favorable money, physical limits loom. The pool of quality training data is finite, and scaling computing power is slamming into electricity shortages and grid capacity. These hard ceilings, the report concludes, could be hit by 2028—around the start of the next U.S. presidential cycle—making the sector's underlying problems impossible to ignore.