The Short Seller's Equation: Michael Burry, Nvidia, and the Myth of Permanent Monopoly
Maxtoshi
The trade is public. Michael Burry, the man who shorted subprime mortgages and called it a debt bomb, has taken a position against Nvidia. The filing shows puts. The filing also shows call options purchased as a hedge. The market reads this as a bearish signal. I read it as a structural audit. Burry's thesis, as reported, hinges on a single word: temporary. He calls Nvidia's monopoly power 'temporary.' That word carries more weight than any price target. It is an admission that the current state of AI infrastructure is a snapshot, not a permanent condition. The question is whether his equation accounts for all variables.
Nvidia's dominance is not a secret. The company controls roughly 90% of the AI data center GPU market. Its gross margins hover above 70%. Its data center revenue accounts for over 85% of total sales. These are the numbers bulls cite. They are real. But they describe a point in time. Burry's bet is that this point is the peak of a cycle, not the beginning of a plateau. He is betting on mean reversion. The market is betting on secular growth. Both cannot be right. The data will decide.
Let me dissect the core of the bull case, because it is not the hardware. It is the software. CUDA, Nvidia's parallel computing platform, has over four million developers. This is the moat. AMD's MI300 series approaches H100 performance in some benchmarks. Google's TPU v5p competes in training. But neither has a software ecosystem that rivals CUDA. Migration costs are not measured in dollars. They are measured in engineering hours. A team that has spent two years optimizing for CUDA does not switch to ROCm because of a spec sheet. This is the lock-in effect. It is real. It is also not permanent.
The contrarian angle, the one the market ignores, is that Burry's hedge reveals his own uncertainty. Buying calls against a short position is not a conviction trade. It is a risk management tool. It caps the downside if the stock rips higher. This is the behavior of an analyst who knows his thesis has a failure rate. Burry's track record with this exact strategy is mixed. He has admitted as much. The market treats him as an oracle. He treats himself as a gambler with better information. That distinction matters.
What the bulls get right is the system-level strategy. Nvidia does not sell chips. It sells infrastructure. DGX systems, NVLink interconnects, InfiniBand networking, and the CUDA software stack form an integrated solution. This is harder to displace than a single GPU. A customer replacing an H100 with a MI300 must also reconsider the entire data center architecture. That is a massive friction point. It is why the customer concentration risk, with Microsoft, Meta, Amazon, and Google representing roughly 40% of revenue, has not yet translated into market share loss. These companies are building custom silicon. Trainium, Maia, TPU. But they are also still buying Nvidia. The coexistence is not a contradiction. It is a transition period.
The variable Burry may be underestimating is the software subscription model. CUDA-X, priced at roughly $4,500 per GPU per year, is an attempt to convert one-time hardware sales into recurring revenue. This is early stage. But if it scales, it changes the earnings equation. It smooths the cyclicality of hardware sales. It increases customer stickiness. It is a hedge against the very competition Burry is betting on. The market has not priced this in. It is still valuing Nvidia as a chip company. That may be the error.
I have audited projects where the pitch was perfect and the code was broken. I have seen teams raise millions on the strength of a whitepaper that could not survive contact with a compiler. Nvidia is not that. The technology is real. The question is not whether Nvidia works. It is whether the market has priced in a future where Nvidia works less well. Burry is betting on that future. The bulls are betting on a future where Nvidia's ecosystem becomes the standard for all AI computation. Both are making a bet on the same variable: the durability of a software moat in a hardware market.
Emotion is a variable I exclude from the equation. The fear of missing out is not a financial model. The fear of being wrong is not a risk assessment. What remains is the structure. Nvidia has a temporary monopoly. That is not a contradiction. It is a definition. The question is how long 'temporary' lasts. Burry says the clock is ticking. The market says the clock has not started. The data will tell us who is right. The only certainty is that the trade is not a simple short. It is a complex bet on the speed of technological displacement. That is a variable worth watching.