Why michael burry's 2026 AI Collapse Prediction Misses the Mark

michael burry, the legendary investor immortalized in the film “The Big Short,” built an unassailable reputation by correctly predicting the 2008 financial crisis—turning $100 million in personal profit and $700 million for his Scion Capital investors into a financial legend’s calling card. Yet as we move deeper into 2026, his latest bearish thesis on artificial intelligence reveals a critical flaw: it’s fighting yesterday’s battles with yesterday’s data.

The question isn’t whether michael burry was right before. The question is whether his playbook still works when the fundamental conditions have changed.

The Legend and the Letdown

michael burry’s greatest triumph overshadows his recent record. Since the 2008 windfall, his track record has been decidedly uneven. As markets rallied over the past decade-plus, burry has repeatedly issued premature bear calls that failed to materialize. His consistent misjudgment of timing and market dynamics led him to close his hedge fund in the past year, citing misalignment with market movements. This isn’t a minor detail—it’s a warning sign that his contrarian playbook, so effective in 2008, may have become less reliable.

The AI-as-Dot-Com-Bust Thesis Doesn’t Hold Up

michael burry’s latest argument rests on a simple comparison: today’s AI fervor mirrors 1999’s dot-com mania, and the ending will be similarly catastrophic. However, this thesis crumbles under three critical scrutinies.

Depreciation Claims Ignore Infrastructure Realities

Burry contends that tech giants like Meta Platforms, Microsoft, and Alphabet are manipulating depreciation schedules to artificially boost earnings. He specifically flags Alphabet for depreciating servers over just four to six years, implying they’re hiding AI infrastructure costs.

Here’s the problem with this argument: while newer GPUs may have shorter viable lives than legacy servers, the actual useful lifespan of AI infrastructure extends 15-20 years. More importantly, older GPU models don’t become worthless the moment newer chips launch. These dated chips power inference—the process of running trained models for end users—a workload that generates substantial ongoing value. This economic reality fundamentally undermines Burry’s depreciation claim.

The Cash Flow Strain Narrative Contradicts Real Numbers

Burry warns that massive capital expenditure on AI infrastructure will drain cash flows, strangling company profitability. Yet 2026 operational results tell a starkly different story.

Alphabet’s cash flow from operations (trailing twelve months) has surged to $164 billion—up from under $100 billion just years earlier. These aren’t signs of cash flow stress; they’re signals of expansion. More compellingly, companies deploying AI at scale report returns exceeding $3 for every $1 invested—a ratio that would make any capital allocation skeptic reconsider. Margins across the tech sector are expanding dramatically, not contracting. The latest wave of innovation—agentic AI, which automates complex workflows—is reportedly delivering cost savings of 25% or more for enterprises. This is the opposite of Burry’s predicted cash crisis.

The NVIDIA Valuation Comparison Is Flawed

Perhaps most strikingly, Burry compares NVIDIA to Cisco, arguing that today’s AI darling faces the same bubble risks as the internet darling that peaked in 2000. The valuation comparison is fundamentally misguided.

When Cisco topped in March 2000, its price-to-earnings multiple had exploded to over 200x—a truly unsustainable multiple for any business. NVIDIA’s current P/E sits at 47—a material discount and substantially more grounded in fundamental reality. This isn’t a bubble waiting to burst; it’s a premium valuation for a company delivering legitimate returns on invested capital.

Market Signals Suggest Continued Strength

The real world offers its own rebuttal to michael burry’s thesis. Since mid-December, rental prices for NVIDIA’s H100 GPU—the workhorse chip driving AI infrastructure—have climbed approximately 17%. GPU scarcity coupled with surging demand signals that the market remains in expansion mode, not contraction.

This sustained demand funnels support through the AI infrastructure ecosystem: companies like Nebius Group, CoreWeave, and IREN benefit directly from continued chip scarcity. Indirectly, enterprises like Bloom Energy—whose energy solutions address the hyperscaler’s greatest operational constraint—see rising tailwinds.

Options Markets Show Conviction

The latest options activity reinforces the market’s bullish convictions. A single trader deployed roughly $9 million on NVIDIA March $205 calls, betting on upside moves. Bloom Energy saw comparable enthusiasm, with a block trade of 400 maximum strike call contracts representing a $1 million conviction bet. These aren’t casual positions; they’re high-conviction capital deployment on AI infrastructure.

The Bottom Line

michael burry’s contrarian genius earned its fame through one spectacular, well-timed call in 2008. That legacy is secure. However, his current bearish AI thesis collapses under the weight of contemporary evidence. Cash flows are expanding, not contracting. Valuations are reasonable, not absurd. GPU demand remains robust, far from softening. The AI infrastructure economy isn’t a repeat of the dot-com era—it’s producing genuine returns and measurable efficiency gains that translate into real corporate profitability. When the data contradicts the narrative this decisively, perhaps it’s time to reconsider the thesis rather than the market.