The Big Short Takes On Big AI
Michael Burry, the investor immortalized in “The Big Short” for predicting the 2008 financial crisis, has turned his sights on a new target: the artificial intelligence boom. He’s now leveling accusations of systemic fraud and financial engineering at the AI sector’s biggest names, arguing their soaring valuations are built on a house of cards. His claims have grown so loud that they have provoked public responses from industry titans like Nvidia and Palantir. This article breaks down Burry’s three core arguments, examines the corporate rebuttals, and clarifies the real warning he is issuing about the overheated AI market.
Burry’s Thesis Part 1: The Circular Money Machine
Burry’s first argument centers on what he views as a massive circular financing scheme among major AI players. The concept is best explained with a simple analogy: imagine a group of people who all owe each other money. To settle their debts, they simply pass the same $10 bill around in a circle until everyone is “even.” Burry argues the AI industry is doing the same, but instead of settling debts, they use the money to invest in the next company, which then signs a deal to spend that money with another, creating the illusion of a booming ecosystem.
Burry shared a Bloomberg chart illustrating a complex web of spending and investment linking Nvidia, OpenAI, Microsoft, and Oracle. As he bluntly stated, “Almost all customers are funded by their dealers.” This is a sophisticated version of the classic “picks and shovels” analogy. In the AI gold rush, Nvidia is selling the essential hardware. But it has taken things a step further by also investing in the “gold miners,” effectively giving them money to buy more of its own picks and shovels. This creates a feedback loop where investment dollars are counted as new revenue, a practice Burry believes manufactures growth rather than reflects it. For more details on these investment structures, see recent analysis of Nvidia’s venture investments in startups like Coreweave.
Burry’s Thesis Part 2: The Real Growth Is Slowing
This manufactured revenue becomes particularly alarming when, according to Burry, you peel back the layers and discover that the underlying engine of organic growth is sputtering. This directly contradicts the prevailing narrative of a once in a generation technology boom. He points to the slowing growth rates in cloud computing, the very infrastructure that should be exploding if genuine AI demand were as high as claimed.
The data supports his skepticism. A few years ago, cloud growth was exploding: Amazon Web Services (AWS) was growing at nearly 40% year over year and Google Cloud at 45%. Today, that growth has slowed into the teens for AWS, while both Google Cloud and Microsoft Azure have seen their once explosive growth rates decelerate significantly. According to Burry, this is the smoking gun. If a true AI revolution were underway, cloud revenues would be rising exponentially. Instead, the slowdown suggests that circular revenue is masking a concerning weakness in real, underlying customer demand.
Burry’s Thesis Part 3: The Accounting Trick Hiding In Plain Sight
Burry’s third and most directly manipulative accusation involves an accounting change: extending hardware depreciation schedules.
First, the concept of depreciation is straightforward. When a company buys expensive hardware like servers, it spreads the cost of that asset over its estimated “useful life.” This annual cost is an expense on the income statement.
Second, by extending that useful life, a company can perform a kind of financial magic. If a server’s useful life is changed from three years to six years, the annual depreciation expense is cut in half. This reduction in expenses directly boosts reported earnings, making the company appear more profitable without any actual improvement in its business operations.
Burry notes that a who’s who of tech has made this change, including Meta, Google, Microsoft, and Oracle. He highlights Baidu as a perfect example. The company extended its server life from four to five years, a move that instantly added the equivalent of $1.5 billion to its profit. They then extended it again from five to six years, and their earnings jumped again. All of this happened while Baidu admitted in the same filing that a large portion of its old hardware was already obsolete and had to be written down.
This creates the central contradiction Burry points out. Companies are claiming their hardware has a longer useful life even as Nvidia releases a new, more powerful generation of chips every 12 to 18 months. These new chips make older models economically obsolete much faster, meaning their true useful life is shrinking, not growing. To Burry, this is “one of the more common ones [frauds] in the modern era.”
The Clap Back: How Palantir And Nvidia Responded
Burry’s pointed critiques have gotten right up the nose of these big tech companies, prompting unusually direct and defensive responses.
Palantir’s “AI Tantrum”
On CNBC, Palantir CEO Alex Karp was asked about Burry’s comments and delivered an indignant and evasive response. Instead of addressing the arguments, he attacked the messenger, dismissing Burry as a short seller engaged in “market manipulation.” Regarding the idea of shorting chip companies, Karp declared it was “batshit crazy.” This deflection, focusing on the critic rather than the criticism, suggests Burry’s thesis may have landed uncomfortably close to home.
Nvidia’s Unusual Memo
Nvidia, the largest company in the world right now, took the highly unusual step of issuing a seven page memo to Wall Street analysts that mentioned Burry by name. The move signals that his arguments are gaining traction, yet the memo failed to meaningfully refute his core points.
• On Depreciation: Nvidia’s memo stated that customers depreciate GPUs over four to six years and that older chips still have “high utilization.” This completely sidesteps the argument. Burry’s claim is not about physical usage but about economic value. High utilization does not disprove the fact that extending the accounting life of rapidly aging hardware inflates current earnings.
• On Circular Financing: Nvidia countered that its investments are small and that its startup customers earn most of their revenue from third parties. This defense misses the point entirely. The issue is the existence of the investment to spending loop that artificially boosts Nvidia’s reported demand. For a deeper look at Nvidia’s strategic position, consider reading about The Unstoppable Rise of Nvidia’s Market Dominance.
The Real Warning: It’s Not Enron, It’s Cisco
In a follow up post, Burry clarified his ultimate warning. He emphasized that he is not calling Nvidia the next Enron, a company destroyed by outright accounting fraud. Instead, he is comparing the current situation to Cisco during the dot com bubble.
Cisco was a legitimate, highly profitable company that became the “poster child of the overbuild.” Investors mistook a massive, temporary boom in companies buying networking equipment for genuine, sustainable, long term demand. When the bubble burst, that demand vanished, and Cisco’s stock cratered. Burry’s warning is that the AI industry is making the same mistake, confusing a capital expenditure boom fueled by hype and circular funding for a durable economic revolution.
Final Takeaway
The question for investors is no longer if an AI revolution is happening, but whether the astronomical figures reported by its leaders represent genuine economic transformation or merely the echo of their own venture capital. Are Burry’s claims of manufactured revenue and accounting tricks a vital reality check on a sector consumed by hype? Or are they simply the calculated skepticism of a famous short seller? As capital continues to pour into AI, Burry’s warnings suggest investors should look past the soaring revenue figures and question whether the foundation of the boom is as solid as it appears.
