CUDA turned a hardware lead into a software standard; the ecosystem is the moat.
| Dimension | MOATE | Community |
|---|---|---|
| Switching costsweight 15Leaving CUDA means rewriting and re-validating the stack — engineering-years, not a procurement decision. | 9 | — |
| Network effectsweight 15Every framework, tutorial, and hire assumes CUDA; each reinforces the next. | 8 | — |
| Scale & cost advantageweight 12Volume buys priority at TSMC and amortizes R&D, but the fab is not theirs. | 7 | — |
| Brand & pricing powerweight 12Sells at premium margins in a shortage; customers queue rather than substitute. | 9 | — |
| Cornered resourceweight 12Two decades of CUDA libraries and the talent that wrote them. | 8 | — |
| Distribution controlweight 12Sold through clouds and OEMs who are also building their own silicon — but buying Hugging Face gives Nvidia a direct line to 18 million developers for the first time. | 7 | — |
| Disruption resilienceweight 12Abstraction layers that make hardware interchangeable remain the live threat; owning where open models are distributed lengthens the escape route. | 7 | — |
| Independenceweight 10Single-foundry dependence, export controls and antitrust — and now the credit of financed customers, who will be roughly a quarter of next years business. | 4 | — |
The four biggest tech companies plan to spend about $725 billion on AI this year, and even they can't pay for it out of pocket. Amazon is moving chips off its balance sheet, Alphabet is selling stock and Meta is renting its own data centers back from investors. Who can afford the war says more about their moats than the spending does.
In March, Jensen Huang said Nvidia was probably done writing big checks to AI labs. The money never stopped. It changed form, from equity stakes into loan guarantees and financing deals, and now roughly a quarter of Nvidia's business comes from customers it helps pay for.
Nvidia paid $12.9 billion for the place where the world's open AI models live. Open models are Nvidia's broadest customer base and the most likely path around its software lock-in. Jensen Huang decided to own both.
From AI race to AI pace. The words changed over a weekend. The spending did not move at all.
Question 1 of 8
Saved.
