Former Fidelity fund manager George Noble has warned that the current artificial intelligence boom resembles a combination of the dot-com bubble and the subprime mortgage crisis. Speaking on "The Real Eisman Playbook," Noble told investor Steve Eisman that the massive capital being deployed into AI may fail to generate adequate returns, a sentiment he summarized by asking, "Show me the ROI." Noble argued that while technology adoption may prove broadly correct, this success does not guarantee profitability for the companies building the infrastructure, drawing a parallel to the internet boom where internet traffic exploded while companies like Global Crossing went bankrupt.
The warning comes as AI stocks rally, with Advanced Micro Devices Inc. (NASDAQ: AMD) crossing the $1 trillion market value mark for the first time on Monday. Noble suggested that funding could eventually dry up if returns fail to justify the amount of capital being deployed. He noted that the financing burden is growing, citing a Reuters report stating that Goldman Sachs expects hyperscaler debt issuance to reach a record $420 billion in 2027, up 60% from 2026 estimates.
Steve Eisman also highlighted concerns regarding Nvidia Corp. (NASDAQ: NVDA), noting that five direct customers accounted for approximately 70% of the chipmaker’s accounts receivable at the end of its latest quarter. Eisman described this figure as "very frightening." Noble estimated that roughly 70% of hyperscaler AI revenue comes from OpenAI and Anthropic, warning that trouble at OpenAI could reverberate through the ecosystem. In response, Eisman agreed with the assessment. However, Microsoft Corp. (NASDAQ: MSFT) has offered evidence that AI-related cloud demand extends beyond these specific frontier-model companies, stating that nearly 90% of its Microsoft Cloud revenue in fiscal 2026 came from customers outside frontier-model companies, while its contracted commercial revenue backlog excluding OpenAI still grew 25%.
While prediction markets are not yet pricing Noble’s scenario as imminent, with a Polymarket contract assigning an 11% chance of an AI-industry downturn by Dec. 31, Noble’s argument suggests that being right about the technology does not necessarily mean investors are paying the right price to build it.
