Michael Burry's AI Reckoning: Warning on the Next Great Write-Off

Core Thesis

Michael Burry predicts that "return on investment will continue to fall, almost all AI companies will go bankrupt, and much of the AI spending will be written off." He suggests this reckoning could manifest as a "Panic of 2026" or 2027, though he acknowledges timing remains uncertain.

Primary Concerns

Return on Investment Decline The analysis highlights a fundamental economic disconnect: companies are deploying extraordinary capital ($80 billion from Microsoft alone in FY2025, with combined Big Tech AI spending approaching $200+ billion annually) while generating minimal returns. ChatGPT reportedly operates at a loss on most interactions, and enterprise solutions command only modest premiums relative to infrastructure costs.

Widespread Bankruptcies Burry anticipates that most AI startups will fail due to: - Concentration of resources among largest players only - Winner-take-most market dynamics - Constant margin compression from open-source alternatives - Big Tech's ability to bundle features and eliminate standalone competitors

Massive Write-offs Expected Anticipated losses include: - Stranded GPU infrastructure - Failed acquisitions - Goodwill impairments as valuations compress - Debt defaults on leveraged infrastructure investments - Venture capital losses when portfolio companies fail to achieve profitable exits

Depreciation Accounting Concerns

Burry estimates hyperscalers will understate depreciation by "$176 billion" between 2026–2028: - GPU product cycles are 2-3 years - Companies are extending useful life assumptions to 6-7 years - This artificially inflates reported earnings by 20%+ - "Customers are funded by their dealers" — circular revenue recognition

Historical Parallels

The analysis compares current dynamics to the dot-com bubble, where approximately $500 billion in fiber-optic infrastructure investment exceeded near-term demand capacity, causing widespread bankruptcies despite eventually enabling future technologies.

Counterarguments Acknowledged

The analysis concedes that accelerating productivity gains, efficient monetization pathways, continued cost declines, or paradigm-shifting value creation could invalidate Burry's warnings.