An easy-to-miss detail: when designing its own chips, NVIDIA uses its own GPU to run simulations and EDA, while AMD has to rely on solutions from external competitors. This isn’t a joke—it’s a real difference in moat. An in-house toolchain means faster iteration and more controllable costs, effectively using your own products to power your own R&D feedback loop. AMD just saw an 8% drop after the close; it’s already struggling to catch up on AI accelerator cards, and its toolchain is also constrained by others. Looking long-term, the cost of catching up with each new generation will be higher. Of course, AMD still has opportunities—there’s room on the inference side to push a value-for-money route—but in the short term, the fundamentals are definitely under pressure. This kind of divergence in underlying capability often doesn’t show up in quarterly results figures, but it will determine who runs more steadily over the next two to three years. $NVDAB $AMDB
#AI芯片 #AMD stock dropped 8% after the close