Michael Burry’s “Byte-Sized” Warning to the AI Bulls
As Nvidia’s trillion-dollar AI empire expands, Michael Burry — the man who called the 2008 crash — warns that Big Tech’s obsession with short-lived chips and creative accounting could turn the AI boom into the next financial bubble.
The world has seen its share of speculative frenzies. From the dot-com boom to the crypto craze, investors have always been drawn to the promise of endless growth. Now, artificial intelligence has become the latest obsession, and Nvidia Corp. (Nasdaq: NVDA) stands at its epicenter. Its powerful chips are the foundation of modern AI, driving record profits, stock rallies, and trillion-dollar valuations. Yet amid the euphoria, one seasoned skeptic is once again raising a red flag. Michael Burry, the man who famously predicted the 2008 financial crisis, believes the math behind the AI boom doesn’t add up.
The warning cuts to the core of the AI economy: depreciation. The chips powering this revolution have short lives, falling in value fast as new models roll out. The result? A system that depends on perpetual reinvestment and creative accounting to look sustainable.
The Short Life of AI’s Golden Machines
Every revolution needs its tools, and for AI, those tools are Nvidia’s GPUs—graphics-processing units capable of handling the colossal calculations behind models like ChatGPT. Each one costs tens of thousands of dollars, and together they make up a huge portion of the roughly $400 billion that the world’s largest technology companies plan to spend on AI infrastructure this year. Analysts expect that figure to swell to nearly $3 trillion by 2029.
But unlike factories or fiber-optic networks, these chips don’t last. Their typical lifespan is around five years, after which they become technologically obsolete. Nvidia’s Chief Executive Officer Jensen Huang half-joked earlier this year that once its next-generation “Blackwell” chips start shipping, “you couldn’t give Hoppers away.” That may have been said in jest, but it highlights a serious truth—AI hardware loses value almost as quickly as smartphones.
This short shelf life is forcing the biggest hyperscalers, Amazon.com Inc. (NASDAQ: AMZN), Microsoft Corp. (NASDAQ: MSFT), and Alphabet Inc. (NASDAQ: GOOGL), into a spending cycle that never ends. To stay competitive, they must constantly refresh their data centers with the latest silicon. What was once a one-off investment has become a treadmill of recurring costs, and the financial implications are staggering.
Michael Burry’s Growing Concern
Burry isn’t disputing AI’s potential. He’s questioning its economics. His argument centers on the depreciation schedules these companies are using to justify their massive capital spending. By stretching the “useful life” of their servers from four years to six or even seven, companies make their profits appear higher—because they’re spreading out the cost of their hardware over more years.
Meta Platforms Inc. (Nasdaq: META), for example, boosted its net income by nearly $2 billion in the first nine months of 2025 by extending the assumed lifespan of its servers. Meanwhile, Amazon.com Inc. (NASDAQ: AMZN) quietly reversed course this year, cutting its estimate back to five years after acknowledging “an increased pace of technology development.” That subtle move signaled what Burry has been warning: the hardware is aging faster than anyone wants to admit.
He calls it a “depreciation tsunami.” As each new generation of chips arrives, the value of the last one plummets. What looks like smart investment today could become a costly burden tomorrow, especially for firms that rely heavily on debt to finance their expansion.
The Dangerous World of AI Debt
The AI boom isn’t just being fueled by corporate cash. It’s increasingly powered by borrowed money. Private-credit giants like Blackstone Inc. (NYSE: BX), Magnetar Capital, and Macquarie Group Ltd. (ASX: MQG) are providing loans backed by AI hardware. Some of these loans use Nvidia GPUs as collateral—a striking parallel to the mortgage-backed securities that underpinned the housing bubble before 2008.
Even Nvidia itself is reportedly exploring structured financing models, including special-purpose vehicles designed to buy and lease GPUs to major customers like OpenAI, CoreWeave Inc., and xAI, the artificial-intelligence startup founded by Elon Musk.
CoreWeave (NASDAQ: CRWV), in particular, epitomizes the risks of this new AI-finance frontier. Once a small cloud-computing startup, it now sports a valuation north of $50 billion and a debt load approaching $17 billion, including lease liabilities. The company is using that money to fund as much as $14 billion in capital expenditures this year—an amount that rivals the annual spending of oil giants like Shell Plc (LSE: SHEL).
CoreWeave insists its business model is sustainable thanks to long-term contracts with “investment-grade” clients. But if GPU values fall faster than expected—or if AI demand weakens—those contracts could unravel quickly. For Burry and others, the warning signs are familiar: massive leverage, rapid depreciation, and an unshakeable belief that the future will always be profitable.
When the Hype Meets Hard Reality
The entire AI economy hinges on one assumption, that demand for computing power will continue to rise indefinitely. As long as companies believe that, the spending continues, the borrowing expands, and Nvidia’s share price keeps soaring. Yet the underlying assets are decaying fast.
Rental prices for Nvidia’s once-coveted A100 (Ampere) and Hopper chips have plunged, down 30% or more this year according to industry trackers. Meanwhile, electricity constraints and data-center bottlenecks are adding new costs that could make older hardware uneconomical to operate.
Nvidia’s success depends on constant innovation, but that same innovation accelerates obsolescence. It’s a paradox that even its CEO acknowledges. The faster the chips improve, the faster the value of existing infrastructure collapses.
Investors may take comfort in the profits of today, but Burry’s warning suggests that tomorrow’s balance sheets could tell a harsher story. The AI revolution is real—but so are the costs hiding beneath its glossy surface.
Conclusion
The AI era has ushered in a new industrial revolution powered by silicon instead of steel. Nvidia’s brilliance, Big Tech’s ambition, and Wall Street’s appetite have created a trillion-dollar machine that seems unstoppable. But as Michael Burry reminds us, even the most dazzling technologies obey the laws of economics.
Depreciation doesn’t care about hype. Hardware ages, margins shrink, and optimism fades. The question isn’t whether AI will transform the world, it already has. The question is whether investors will recognize the risks before the next great bubble bursts.