Michael Burry is doing what he does best: looking at a market drunk on optimism and asking the question nobody wants to answer. This time, his crosshairs are on artificial intelligence, and more specifically, on the companies spending hundreds of billions of dollars building the infrastructure to power it.
The man who famously bet against the US housing market before its catastrophic collapse in 2008 is now drawing a striking parallel between the current AI boom and one of history’s most celebrated market manias: the 1920s radio craze centered on Radio Corporation of America, better known as RCA.
The RCA lesson nobody wants to learn
RCA’s story is one of the most instructive cautionary tales in financial history, yet investors rarely seem to learn from it. In the 1920s, radio was genuinely revolutionary. It changed how people consumed news, entertainment, and culture. The technology worked. The demand was real. The future was bright.
And yet, RCA’s stock surged roughly 200-fold before collapsing 98% between 1929 and 1932. Radio itself kept growing for decades. The technology did not fail. The stock failed, because the price had run so far ahead of economic reality that even getting the big picture right could not save investors from financial ruin.
Burry tagged Nvidia directly in his RCA comparison, and the parallel is uncomfortable in its precision. Nvidia‘s stock has surged roughly 12-fold since early 2023, making it briefly the most valuable public company on earth. The chips are real. The demand, for now, is real. But Burry’s question is not whether AI works. It is whether the financial math works, and he believes it does not.
The depreciation problem at the heart of the bubble
Burry’s most technically detailed argument centers on accounting, specifically how companies are depreciating Nvidia’s GPUs. His concern is that the physical hardware powering today’s AI buildout may become economically obsolete in two to three years, yet companies are spreading those costs over much longer depreciation schedules. The practical effect is that expenses look smaller than they actually are, which makes earnings look better than they actually are.
Technology analyst Aakash Gupta, who elaborated on Burry’s argument, estimated this gap could understate depreciation by roughly $176 billion between 2026 and 2028, inflating reported income by more than 20% at companies like Oracle and Meta. If those chips need to be replaced or written down sooner than expected, the earnings hit will be sudden and severe. This is the accounting trick Burry referenced, and it matters because it means investors may be paying premium valuations for earnings that are, in part, a financial illusion.
Spending that has no visible end
Beyond the accounting, Burry’s broader concern is simpler: the spending is staggering, it is accelerating, and nobody seems to know when it stops. The four largest hyperscalers, Amazon, Microsoft, Alphabet, and Meta, have collectively guided for $650 to $700 billion in capital expenditures in 2026 alone, representing a greater than 60% increase from 2025. Amazon has committed $200 billion, a figure so far above analyst expectations that the company lost $450 billion in market capitalization over nine consecutive sessions after announcing it.
Amazon is projected to run negative free cash flow in 2026. Alphabet’s free cash flow is expected to fall roughly 90% from where it stood recently. Major tech companies raised over $100 billion in bonds in 2025. JPMorgan projects $1.5 trillion in tech debt issuance ahead.
Burry estimates Nvidia will sell $400 billion worth of chips this year against less than $100 billion in identifiable application-layer use cases generating real revenue. The gap between what is being spent and what is being earned from it is where bubbles live.
Why he is short Nvidia specifically
Burry has been clear about why he chose Nvidia as his primary short rather than Microsoft, Meta, or Alphabet. Those companies have dominant businesses that exist entirely outside of AI, search engines, social media platforms, enterprise software, and cloud services, that give them staying power even if the AI buildout disappoints. Nvidia has no such cushion. It is, as Burry put it, the purest play on the AI buildout, which means it is also the purest exposure to the downside if that buildout stalls or reverses.
His critics are not entirely wrong when they note that his timing has sometimes lagged his analysis. He acknowledged his own imperfections freely. But the pattern he is identifying, massive capital expenditure cycles driven by competitive pressure rather than proven returns, followed by asset writedowns and earnings restatements, is one that has played out repeatedly across different industries and eras.
The central question Burry is asking is not whether AI will transform the world. It probably will. The question is whether the companies spending the most aggressively will profit from it, or whether they will simply hand the bill to their shareholders while Nvidia collects the revenue. History, from RCA to the fiber optic boom of the late 1990s, suggests the answer is rarely as clean as the optimists believe.
Sonia Boolchandani is a seasoned financial writer She has written for prominent firms like Vested Finance, and Finology, where she has crafted content that simplifies complex financial concepts for diverse audiences.
Disclosure: The writer and her his dependents do not hold the stocks discussed in this article. The website managers, its employee(s), and contributors/writers/authors of articles have or may have an outstanding buy or sell position or holding in the securities, options on securities or other related investments of issuers and/or companies discussed therein. The content of the articles and the interpretation of data are solely the personal views of the contributors/ writers/authors. Investors must make their own investment decisions based on their specific objectives, resources and only after consulting such independent advisors as may be necessary.