Elon Musk says he is building a chip that will be "2-3x better than Nvidia" at just 10 percent of the cost — a claim that landed across financial outlets including The Motley Fool, The Globe and Mail, Yahoo Finance, and AOL, each asking the same pointed question: should Nvidia investors be worried?
The claim, attributed directly to Musk, is striking in its specificity. A chip with two to three times the performance at a tenth of the price would represent a dramatic shift in the economics of AI infrastructure, where Nvidia's H100 and successor GPUs currently sell for tens of thousands of dollars apiece and dominate data center buildouts worldwide.
Musk has not been shy about competing with the companies he once relied on. His AI venture xAI built the Colossus supercluster — one of the largest in the world — using Nvidia hardware, but the new chip claim signals an intention to eventually cut that dependency. Custom silicon has become a strategic priority across big tech: Google has its TPUs, Amazon its Trainium and Inferentia chips, and Microsoft is developing its own AI accelerators.
For now, the claim remains unverified. Chip development timelines are notoriously long, and the gap between a founder's announcement and a shipping product that outperforms entrenched hardware at scale is wide. Nvidia, for its part, has decades of software ecosystem investment — the CUDA platform — that rivals have found difficult to replicate regardless of raw hardware specs.
If Musk's chip delivers even a fraction of what he promises, it would reshape how companies think about the cost of training and running AI models — and that makes this claim worth watching closely.