Michael Burry Warns Anthropic IPO Signals AI Industry Vulnerability
Investor highlights complex debt structures and slowing growth as systemic risks for AI.
1Burry's Skepticism on Anthropic
Investor Michael Burry identified significant red flags in Anthropic’s recent IPO filing. He argues that the company’s dense web of financial arrangements creates a system that is catastrophically vulnerable. Burry specifically pointed to massive compute spending commitments and complex, circular deal structures that he believes mask underlying risks within the broader AI ecosystem.
2Massive Compute Spending Commitments
Anthropic has committed to purchasing approximately $537 billion in compute resources from major hyperscalers like Amazon, Alphabet, and Microsoft. These obligations, combined with additional agreements with firms like SpaceXAI, total roughly 12 times the company’s current annual run-rate revenue. Burry views these figures as evidence of unsustainable capital expenditure requirements.
3Circular Financing and Hyperscalers
Burry highlights that hyperscalers function simultaneously as lenders, vendors, channel providers, and shareholders for Anthropic. He describes these financing arrangements as garish and audacious, suggesting they are necessary because these tech giants lack sufficient cash flow to fund the AI buildout independently. This complexity mirrors vendor financing practices seen during the dot-com bubble.
4Slowing Growth and Token Trends
While Anthropic reported $16 billion in revenue for the first half of the year, Burry notes a deceleration in growth. Quarterly revenue growth slowed from 144% between the first and second quarters to 46% in the third. He attributes this partly to the declining trend of token-maxxing as AI models become more efficient.
5Structural Fragility in AI Deals
The investor emphasizes that the complexity of these deals is designed to offload debt from the balance sheets of involved parties. He warns that while this interconnectedness appears secure during market upswings, it leaves the entire organism highly susceptible to failure during a market pause or a downturn in the broader technology sector.
6Technological Efficiency vs Revenue
Burry observes that large language models like Claude are becoming more efficient, which ironically works to the detriment of their own growth rates. As companies shift toward using models more efficiently and adopting smaller language models, the demand for massive computing power may decrease, potentially undermining the revenue projections of AI infrastructure providers.
7Market Risks and Hardware Glut
Beyond financing, Burry is focused on the fragility of debt-heavy AI deals and potential hurdles in hardware deployment. He anticipates a possible crash in memory chip prices once current shortages ease. Furthermore, he suggests that the industry is overinvesting in microchips and data centers, ignoring the fundamental economics behind these aggressive expansion strategies.
8Historical Parallels to Bubbles
Drawing on his experience from the mid-2000s housing bubble, Burry suggests that market participants are repeating the same mistakes with AI. He remains skeptical of the current boom, citing aggressive accounting and hidden liabilities. His analysis serves as a cautionary note on the long-term sustainability of the current AI financing model.