Former Fidelity fund manager George Noble warned that the AI bubble burst could cause 17 times more damage than the dot-com bust, which wiped about $5 trillion from the Nasdaq.
According to Polymarket, the probability of the AI bubble bursting in 2026 has risen to more than 17% after recently dropping from 30% to 14%. The odds of a contract using different resolution criteria were between 16% and 24%, as traders weighed a decline in technology stocks, earnings concerns and overall weakness in global markets.
Noble tied his prediction to the huge influx of money into AI infrastructure, arguing that the financial fallout could extend far beyond technology companies if they don’t deliver the expected returns.
“The impact of this could actually be more significant,” Noble said while discussing increased AI capital spending.
AI bubble probability recovers to over 17%
New pressure on semiconductor and technology stocks is adding to these concerns. The Wall Street Journal reported that U.S. stock futures fell on Thursday as AI-related concerns spread from Asian markets, with SK Hynix and Samsung Electronics falling nearly 9%.
Both South Korean chipmakers plan to spend billions of dollars on semiconductor factories and AI capabilities. The decline in these companies occurred as investors questioned whether the revenue generated by AI services was worth the industry’s growing infrastructure costs, the report said.
IBM added to the worries earlier this week after its stock price fell almost 25%, the biggest daily decline since 1968. IBM closed another 2.7% lower at $211.20 on Wednesday, taking its decline in several sessions to more than 26%, according to market data cited in the report.
IBM said in its warning that spending on AI infrastructure is shifting enterprise budgets away from software and contributing to weaker-than-expected revenue growth. The decline wiped tens of billions of dollars off IBM’s market value, weighing on other software and information technology stocks, the report said.
The U.S. Treasury Department’s draft report also considers how a downturn in AI could ripple through the economy. The report, which cited research from the University of Texas at Austin cited by NOTUS, found that AI companies are more closely tied to the U.S. economy than Internet companies in the dot-com era.
Under the report’s downside scenario, disappointing productivity and profits could hurt private credit, chip makers, cloud providers, power companies and companies that finance data centers. The Treasury Department did not predict an imminent collapse, but cited power shortages, lending restrictions, supply chain disruptions and geopolitical tensions as risks facing the sector.
Demand for cash could reveal inflated AI valuations
Separately, Ray Dalio has argued that liquidity, not weak technology, could destroy the AI boom. In a television interview reported by Bloomberg, the Bridgewater Associates founder explained that investors often mistake rising asset values for ready money.
Dalio used the example of privately held companies to explain the risks. Companies can raise billions of dollars in valuations while raising much less actual capital, but shareholders cannot tap into that paper wealth without selling it. In his assessment, stress would arise if too many investors tried to convert these valuations into cash at the same time.
Mr. Bernstein and Mr. Cummings noted that new pressures have been building under the boom. Economists wrote in a recent post on Substack that the AI bubble “remains inflated,” while technology investment has reached nearly 5% of U.S. GDP, exceeding levels recorded during the dot-com era.
Their analysis also found that big tech companies are putting enough capital into AI projects to reduce their cash reserves. These numbers, along with Noble’s warning and Dalio’s liquidity concerns, have investors watching to see if AI returns can keep up with the money already being pumped into the sector.
