The semiconductor industry is on the brink of a fundamental crisis. My analysis of recent statements from leading industry experts confirms that large-format AI chips, including Nvidia's products, are approaching the physical limit of scaling. Huawei's chief scientist for semiconductors, Liao Heng, in an exclusive interview with a Chinese blogger, delivered a harsh assessment: further increases in die size, transistor density, and HBM memory capacity are becoming a dead-end path.
"There must be a limit to how they scale with ever-larger compute dies and more HBM capacity. The industry keeps pushing forward, but once it crosses this physical limit, an avalanche will occur," the scientist emphasized. This is not just a theoretical warning, but a signal of a paradigm shift.
Huawei's Alternative Strategy
Instead of racing for nanometers, Huawei is promoting the concept of the Tau Scaling Law. This approach, presented in May at the IEEE conference in Shanghai, shifts the focus from transistor miniaturization to reducing signal transmission time between chip components. The key element is LogicFolding technology, which distributes logic circuits across multiple layers, radically reducing latency. The first smartphone chip on this architecture is expected to appear in the near future, and by 2031, Huawei plans to achieve transistor density equivalent to 1.4 nm.
Sanction Pressure as a Catalyst for Innovation
The forced reality is this: since 2019, Huawei has been isolated from advanced Western technologies and ASML lithography equipment. However, it is precisely these restrictions that are pushing the Chinese industry toward unconventional solutions. Liao Heng stated directly: "We need to invest more effort in design to trade higher complexity for lower consumption of computing resources." In this context, he highly praised DeepSeek founder Liang Wenfeng, whose success became possible precisely because of the forced search for efficient architectural solutions under conditions of computing scarcity.
The Formation of Two Ecosystems
The geopolitical split is finally taking shape as a technological division. According to Liao, "to survive, each side must create its own complete production and supply capabilities." This means we are witnessing not a temporary confrontation, but the formation of two parallel semiconductor ecosystems with their own standards and approaches.
My assessment: Huawei's statements are not just a defense of its own strategy, but a marker of a tectonic shift. While Nvidia continues to increase power through resource-intensive solutions, the Chinese giant is betting on intelligent optimization. In the long term, it is efficiency, not raw performance, that will become the determining factor in the AI race. And if the Tau Scaling Law truly works, we may witness a revision of the very foundations of chip design.