Nobody, not even the head of NVIDIA himself, appears to be worried about major tech companies developing their own custom chips (ASICs) to wrest market share from NVIDIA. He has gone as far as to conjecture that most of those ambitious in-house projects will ultimately fail to see the light of day.
NVIDIA has slept little while waking up fast on AI, regularly rolling out even newer and more powerful chip architectures. More solidly, the AI world, its thinking goes, depends heavily on NVIDIA's technology, and amid these grand efforts by big tech like Microsoft's Maia chip and Amazon's Graviton custom silicon, NVIDIA CEO at GTC Paris 2025 showed little concern, alluding to the fact that the custom silicon effort would not affect NVIDIA's market pulling power.
He was quoted as saying,
"First of all, a lot of ASICs are being started, but most of them are getting canceled. And the reason for that is, what's the point of building an ASIC if it's not going to be better than the one you can buy"
It is rumored that the likes of Google, Microsoft, Broadcom, and OpenAI are working secretly on their chips to counter NVIDIA's strong market position and offer cost-effective alternatives. While custom-designed ASICs could offer better efficiency for specified jobs, general-purpose hardware from NVIDIA would at least keep the companies insulated from market shifts and delays.
However, the NVIDIA CEO thinks building these custom chips is not the true challenge; the real problem is to scale them. As he pointed out, while developing ASICs can be done, scaling them effectively requires a type of expertise which, for the moment, lies mainly within NVIDIA. This, he argues, is a major reason for NVIDIA's enduring preeminence as an AI hardware supplier on a global scale.
In summary, the argument of the NVIDIA chief implies that custom ASICs will have little real chance of doing away with the offerings of his company if only NVIDIA could continue to provide the best-alternative technology. His comments suggest that while NVIDIA is aware of the growing competition, it appears confident of holding on to its leadership in the AI hardware market.
