Jensen Huang responds to Huawei's 'Tao Ding' law after 'trillion yuan banquet': It's a breakthrough for Huawei, but TSMC has used similar technology for a decade

NVIDIA CEO Jensen Huang hosted a dinner in Taipei on the evening of May 28 for senior executives of supply chain partners. TSMC Chairman C.C. Wei, Hon Hai Chairman Young Liu, Quanta Chairman Barry Lin, and other leaders of Taiwan’s semiconductor and electronics industry were all in attendance. The event was described by outsiders as the “trillion-dollar banquet.”

After the dinner, Huang accepted media interviews outside the restaurant, responding to Huawei’s recently announced “Tao Law (τ),” saying, “This is a breakthrough for Huawei, but it is not a threat to TSMC.” He explained that Huawei uses chip stacking and 3D packaging technology to double or even triple or quadruple the number of transistors without shrinking process line widths. This is an excellent technical path, but TSMC and Taiwan have been laying groundwork in this area for nearly a decade, with deep technical accumulation.

Huawei’s “Tao Law” was officially announced on May 25 by company director and president of the semiconductor business unit, He Tingbo, at the 2026 International Symposium on Circuits and Systems. It establishes a multi-level collaborative optimization system spanning devices, circuits, chips, and systems, and predicts that by 2031, high-end chips based on this law will achieve transistor density equivalent to that of a 1.4nm process. It is seen as the first time a Chinese company has proposed a new principle to lead the global semiconductor industry.

Regarding the competitive landscape of cloud service providers developing their own ASIC chips, Huang emphasized that NVIDIA is “the only platform and architecture that can be used in every cloud service,” and welcomed competition, saying, “We just need to keep running ahead.” He also publicly reiterated a call for Taiwan to address its energy supply issues, and encouraged all sectors of Taiwan to proactively use AI, rather than just manufacturing AI infrastructure for others.

CNA | CNA