Huawei scientist warns Nvidia of coming transistor 'avalanche'

Huawei's top semiconductor scientist Liao Heng used a rare four-hour interview aired in China in late July to argue that Nvidia and the rest of the leading-edge chip industry are running out of runway on the trick that built them, as reported by Bloomberg. "There has to be a limit in how they scale up with ever-increasing compute die and more HBM," Liao said. "The industry is still pushing, but once they cross that physical limit, there will be an avalanche."

The pitch that comes with the warning is Huawei's own alternative, which the company brands the Tau Scaling Law and implements through an architecture it calls LogicFolding. Instead of chasing ever-smaller transistors, the framework prioritises how quickly signals move between parts of a computing system. Liao described a co-design push between chips and models, saying Huawei has "to put in more effort around design to trade higher complexity for less compute resource consumption." That framing is a direct read of Huawei's situation, since US export controls cut it off from EUV lithography and shrinking transistors the traditional way is not on the menu.

The near-term test Liao pointed to is Huawei's next smartphone chip, the first product built under this framework. In the version republished by The Star, Liao said the world will have a clear understanding later this year of how Huawei's alternative approach helps it narrow the gap with rivals once the new chip is dissected. He also framed the wider picture in terms of a hardening split, arguing that "to survive, each side has to build its own complete manufacturing and supply capabilities."

The honest caveat is that most of the specifics here are Huawei's own claim delivered by Huawei's own scientist in a Chinese-language interview, and the reporting does not give independent teardown data, an exact process node, or a head-to-head comparison against a specific Nvidia or TSMC product. Take the density gains and the "avalanche" framing as reported, not settled. What is worth watching is whether the actual smartphone chip, when it lands, backs up the story, because if a design-led post-Moore approach can measurably close the gap without EUV, the strategic picture for China's chip roadmap and for the customers who bet on it starts to look different.