Remember to star the public account ⭐️ to receive notifications promptly.
Source: Content from wccftech.
NVIDIA’s CEO has once again commented on the competition with ASIC manufacturers like Google and Amazon, claiming that not many teams can achieve what the ‘green team’ has done.
Since companies like Google released their latest solutions, the debate surrounding NVIDIA and ASIC chips has intensified, with the core argument being that as the world shifts from training workloads to inference workloads, NVIDIA’s technology stack is replaceable. In the recent third-quarter earnings call, CEO Jensen Huang discussed the internal ASIC chip developments at major tech companies. When asked whether these projects would lead to actual large-scale deployments, NVIDIA’s CEO responded:
An analyst asked: “Jensen, this question is for you. Considering your announced deal with Anthropologie and the overall scale of your customers, I’m curious about your views on the role of AI ASICs or dedicated XPUs in these architecture builds? Have you noticed that you’ve been quite firm in the past that some of these projects may never truly deploy?”
Huang replied: “Yes. Thank you very much, and I appreciate your question. First, you are not competing with the teams—sorry, you are competing with the companies, but you are competing with the teams. And there are not many teams in the world that are truly good at building these extremely complex systems.”
If you still don’t understand what Jensen means here, he refers to the recent agreement reached by Anthropic, which includes infrastructure built around the Blackwell and Rubin systems. Meanwhile, Anthropic has also signed an agreement for Google’s latest Ironwood TPU, raising new questions about whether ASIC chips can truly compete with NVIDIA. Jensen commented on the analyst’s inference, pointing out that when companies develop custom chips, the real competition does not come from NVIDIA but from their respective engineering teams.
Jensen stated that there are very few teams in the market that can invest as much effort as NVIDIA, which is one of the reasons the company denies there is fierce competition with ASIC chips. I recently discussed that Google’s TPU is also a competitive option in the inference space, but Jensen claims that NVIDIA excels in all AI segments, indicating that the company is committed to maintaining an ‘irreplaceable’ position in the three main areas of the AI industry—pre-training, post-training, and inference.
Interestingly, Jensen pointed out that for cloud service providers (CSPs), deploying ‘random ASICs’ in data centers is far from ideal compared to choosing NVIDIA’s technology stack, as NVIDIA’s products have a broader application range. Therefore, Jensen ultimately believes that the custom chips offered by large tech companies still cannot match NVIDIA on an ‘engineering’ level. Moreover, even if they could replicate NVIDIA’s computing power, NVIDIA has a powerful software stack called CUDA, which is the real key to attracting industry attention.
*Disclaimer: This article is original by the author. The content reflects the author’s personal views, and Semiconductor Industry Observation reprints it only to convey a different perspective, not representing Semiconductor Industry Observation’s endorsement or support of this view. If there are any objections, please contact Semiconductor Industry Observation.
END
This is the 4232nd issue shared by Semiconductor Industry Observation, welcome to follow.
Recommended Reading
★A Chip That Changed the World
★U.S. Secretary of Commerce: Huawei’s Chips Are Not That Advanced
★“ASML’s New Lithography Machine, Too Expensive!”
★NVIDIA’s New Competitors Rising Quietly
★Chip Prices Plummet, All Blame Trump
★New Solutions Announced to Replace EUV Lithography!
★Semi Equipment Giants, Salaries Soar by 40%
★Foreign Media: The U.S. Will Propose Banning Software and Hardware for Cars Made in China
Star the public account ⭐️ to receive notifications promptly, small account to prevent loss
Like request
Share request
Recommendation request
