Nvidia chief Jensen Huang is praising Huawei’s chip work while arguing it is still playing catch-up to the industry’s most experienced players, especially TSMC. His comments land at a sensitive moment for the semiconductor race: Chinese firms are pushing harder on advanced packaging and new design methods, while the established supply-chain heavyweights keep extending their lead through years of accumulated know-how.
Speaking at a banquet for Taiwan’s semiconductor industry, Huang said Nvidia expects a strong second-half performance as it deepens ties with Taiwanese suppliers on next-generation technology. That is classic Nvidia: thank the ecosystem, hype the pipeline, and make sure everyone remembers where the bottlenecks still lie.
What Huang said about Huawei’s chip technology
Asked about Huawei’s Tau law chip technology, Huang described it as impressive but far from a threat to companies such as TSMC. He said Huawei is using the approach to double or even triple transistor counts without shrinking line width in the semiconductor process, then added that TSMC has had similar technology for about 10 years.
That comparison matters because advanced packaging and 3D stacking are no longer exotic tricks; they are the real battleground as Moore’s Law slows. Whoever can squeeze more performance out of packaging, interconnects, and chiplet design gets a growing advantage, and TSMC has spent a decade turning that into a business moat.
Huawei’s folding-chip push
Nvidia’s assessment does not dismiss Huawei outright. Huang praised the company’s innovation in chip technology and its new Tau scaling approach, while still saying the Chinese maker remains behind other global competitors. That is a polite way of saying ”interesting” without handing out a trophy.
Huawei’s recent work also includes the LogicFolding architecture, which researchers at Peking University said is based on their 3D design tools. Separately, Huawei has said the new Tau scaling law points to a new direction for the global chip industry and will soon appear in international chip technologies.
Nvidia’s own pitch is scale, not novelty
Huang closed by turning the spotlight back on Nvidia, saying the company is the only platform, chip, and computing architecture that can be used in any cloud service. The subtext is obvious: in a market obsessed with the next clever trick, Nvidia still wants to be judged on reach, not just research.
Huawei and Chinese researchers have also developed what they describe as the world’s first chip for two-dimensional parallel computing. The likely next phase is less about who invents the flashiest technique and more about who can turn it into volume production, which is where TSMC usually starts smiling and everyone else starts writing checks.