According to Business Insider, Michael Burry continues his crusade against Nvidia and the AI boom despite the chipmaker’s blowout earnings. Nvidia reported record revenue and profit last quarter with a bullish Q4 forecast, sending shares up 5% in premarket trading. CEO Jensen Huang dismissed AI bubble talk while CFO Colette Kress revealed the company has “visibility to $0.5 trillion in Blackwell and Rubin revenue” through 2025-2026. Kress also estimated “$3 trillion to $4 trillion in annual AI infrastructure build” by 2030 and noted older A100 GPUs from six years ago remain fully utilized thanks to CUDA software. Burry responded with multiple X posts questioning Nvidia’s accounting and suggesting true end demand is “ridiculously small.”
Burry’s Accounting Concerns
Here’s where things get interesting. Burry isn’t buying Nvidia‘s narrative about older chips maintaining value. He argues that just because something is physically used doesn’t mean it’s profitable from an accounting standpoint. His airline analogy hits hard – yeah, you might drag out old planes during holiday season, but they’re barely making money and burning through fuel. That’s basically what he’s saying about Nvidia’s older GPUs running at “full utilization.” They’re far less energy efficient than newer chips, meaning customers are probably eating massive electricity costs. So the question becomes: are these older systems actually creating value, or just being used because companies feel they have to?
The Customer Funding Question
Burry’s second major point cuts even deeper. He calls out the “give-and-take deals” between Nvidia and AI companies like OpenAI, Microsoft, and Oracle. His claim that “true end demand is ridiculously small” and “almost all customers are funded by their dealers” suggests something pretty concerning. Basically, he’s implying that the AI ecosystem is a circular economy where Nvidia is both supplier and indirect funder of its own demand. If that’s true, it completely changes how we should view Nvidia’s growth numbers. Are we looking at genuine market demand or something more artificial?
Nvidia’s Stock Buyback Paradox
Now this one’s a head-scratcher. Burry points out that despite Nvidia buying back nearly $113 billion worth of stock since 2018, the company actually has 47 million more shares outstanding. Wait, what? How does that math work? He calculates the “true cost” of stock-based compensation dilution at $112.5 billion, reducing owner’s earnings by 50%. That’s a massive number that doesn’t get much attention. It raises serious questions about whether the buybacks are actually benefiting shareholders or just offsetting dilution from executive compensation. For companies relying on industrial computing infrastructure, understanding these financial dynamics is crucial when evaluating long-term partners.
The Bigger Picture
So where does this leave us? We’ve got Nvidia executives painting a picture of endless growth and Burry warning of dot-com bubble parallels. Both can’t be right. The fact that Burry’s fund owned bearish put options on 1 million Nvidia shares worth $187 million certainly adds weight to his arguments – he’s putting money behind his convictions. But here’s the thing: Nvidia keeps delivering incredible numbers quarter after quarter. The demand for AI infrastructure appears real, and companies building out these systems need reliable hardware partners. For enterprises deploying industrial computing solutions, working with established suppliers who understand manufacturing environments becomes critical. The debate isn‘t going away anytime soon, and Burry’s return to X after two years suggests he’s just getting started.
