Former Fidelity manager warns AI boom mirrors dot-com and subprime eras
- Former Fidelity manager George Noble compares AI boom to dot-com and subprime crises
- Goldman Sachs forecasts hyperscaler debt issuance to hit $420 billion in 2027, up 60%
- Five direct customers account for about 70% of Nvidia's accounts receivable
- Microsoft reports 90% of cloud revenue comes from non-frontier model customers

*this image is generated using AI for illustrative purposes only.
Former Fidelity fund manager George Noble compared the current artificial intelligence boom to a merger of the dot-com bubble and the subprime crisis, warning that massive capital deployment may not yield adequate returns for investors.
Noble made the remarks during an interview with Steve Eisman on The Real Eisman Playbook. He argued that while the technology may succeed, the financial structures supporting it are fragile. "Show me the ROI," Noble said, suggesting that funding could dry up if returns fail to justify the scale of investment.
Rising debt and market valuations
The warning coincides with a rally in AI-linked stocks. Advanced Micro Devices (NASDAQ: AMD) crossed $1 trillion in market value for the first time on Monday, driving the Nasdaq 100 to a three-month high. However, the financing burden on major tech firms is expanding.
Goldman Sachs projects that hyperscaler debt issuance will reach a record $420 billion in 2027, a 60% increase from 2026 estimates. This surge in borrowing raises concerns about sustainability if revenue growth lags behind capital expenditure.
Customer concentration risks
Steve Eisman highlighted significant concentration risks within the supply chain, particularly regarding Nvidia Corp. (NASDAQ: NVDA). Data from Nvidia’s latest quarter indicates that five direct customers account for approximately 70% of its accounts receivable. Eisman described this dependency as "very frightening."
| Metric | Figure | Context |
|---|---|---|
| Hyperscaler Debt Issuance (2027) | $420 billion | Record high; up 60% from 2026 |
| Nvidia Top 5 Customer Share | ~70% | Of accounts receivable |
| Polymarket AI Bust Probability | ~11% | By December 31 |
Noble estimated that roughly 70% of hyperscaler AI revenue derives from OpenAI and Anthropic. He warned that financial trouble at these key entities could reverberate through the broader ecosystem. Eisman agreed, noting that circular financing among suppliers, customers, and investors creates systemic vulnerability.
Microsoft counters concentration narrative
Microsoft Corp. (NASDAQ: MSFT) presented data suggesting AI demand is broader than just frontier-model companies. The company reported that nearly 90% of its Microsoft Cloud revenue in fiscal 2026 came from customers outside OpenAI and Anthropic. Additionally, Microsoft’s contracted commercial revenue backlog excluding OpenAI grew 25%.
Market sentiment remains cautious but optimistic
Prediction markets currently assign a low probability to an imminent AI bust. A Polymarket contract with nearly $3 million in traded volume places the chance of an AI-industry downturn by December 31 at about 11%. Despite this, the growing web of financing links continues to fuel debate over whether the boom depends excessively on continuous capital inflows.
How might the projected $420 billion in hyperscaler debt issuance for 2027 impact broader credit market liquidity and borrowing costs for non-tech sectors?
What specific regulatory measures could be introduced if circular financing among AI suppliers and customers is deemed a systemic risk to financial stability?
If OpenAI or Anthropic face funding constraints, how quickly could the ripple effects impact Nvidia's revenue recognition and the valuation of its top five customers?
























