What chip-design breakthrough did Huawei announce in May 2026, how does its LogicFolding approach and proposed Tau Scaling Law—or “Her’s Law”—seek to achieve cutting-edge, 1.4-nanometer-equivalent chips by 2031 without the extreme ultraviolet lithography machines restricted by U.

Huawei’s May 2026 announcement was a design-and-packaging roadmap, not proof that China can fabricate true 1.4-nm transistors. It pairs a three-dimensional circuit architecture, LogicFolding, with a proposed “Tau (τ) Scaling Law” to claim 1.4-nm-process-equivalent transistor density by 2031 while avoiding reliance on EUV lithography. 12

• What was announced. At IEEE ISCAS in Shanghai, Huawei semiconductor chief He Tingbo introduced the Tau law—sometimes informally rendered “He’s law,” a play on her surname—and LogicFolding. Huawei says its initial double-layer Kirin implementation would arrive in 2026 and that a three-layer version could reach the stated 1.4-nm-equivalent density in 2031. 13

• How it is meant to work. Rather than chiefly shrink transistor features as Moore’s-law scaling does, LogicFolding reorganizes and vertically stacks logic so critical signal paths are shorter. The Tau concept treats signal-propagation delay—the relevant time constant—as the principal scaling target: shorter connections should reduce interconnect resistance/capacitance penalties, improve effective density and potentially improve performance per watt. Huawei has claimed roughly a 55% stepwise density gain from the approach. 24

• Why this could reduce dependence on EUV. The proposition is to obtain more compute from older, domestically accessible process technology plus advanced 3D design/packaging, rather than patterning ever-smaller planar features with ASML EUV scanners—the machines China cannot obtain under export controls. “1.4-nm equivalent” therefore means a claimed density comparison, not that Huawei or its manufacturing partners will have a literal 1.4-nm fabrication node. 13

• How far behind—or differently positioned—this is. China’s demonstrated leading-edge domestic production has generally been characterized as 7-nm class, achieved with older deep-ultraviolet lithography rather than EUV. Huawei’s target is thus an attempt to bridge several nominal node generations through architecture. TSMC, by contrast, has true 2-nm technology and has said it aims to begin 1.4-nm production in 2028—about three years before Huawei’s 2031 density-equivalence target. 15

Why the claim remains unproven

• Huawei supplied no independently verified benchmarks, chip measurements, yield data, power figures, reliability results, or cost data for a production LogicFolding implementation. Reuters accordingly described whether this is a true breakthrough as unresolved. 12

• Density is not whole-chip leadership. Transistor density alone does not establish equivalent speed, energy efficiency, memory bandwidth, I/O, packaging quality, manufacturability, or software ecosystem. A 3D structure can shorten some paths while making routing, power delivery, testing and yield more difficult.

• Thermals are central. Stacking active logic concentrates heat and makes heat extraction from inner layers harder. It can also exacerbate power-delivery and clocking problems, especially in sustained high-power AI accelerators. Analysts have highlighted heat dissipation, EDA tooling and yields as major hurdles. 6

• Tools and economics matter. China still needs mature domestic EDA flows capable of co-optimizing 3D placement, routing, timing, verification, thermal behavior and test. Even if technically feasible, advanced stacking can increase process steps, packaging complexity, defect exposure and cost; lower yields could erase a nominal density advantage. The available evidence does not demonstrate that Huawei has solved those system-level problems. 26

Strategic context

• The program is Huawei’s most explicit attempt to turn U.S. restrictions into a different technological path: build domestic alternatives in design, fabrication, packaging and AI compute rather than wait for access to EUV, leading U.S. EDA tools, or top Nvidia products. It aligns with Beijing’s broader semiconductor self-reliance agenda. 12

• Huawei’s 2023 Mate 60 return—using a domestically made 7-nm-class Kirin chip—was the earlier political and commercial proof point that sanctions had not stopped its smartphone-chip recovery. The new announcement extends that logic from a single comeback product to a long-horizon design and AI-chip roadmap. 12

• Nvidia’s constrained China position makes the stakes larger. Export controls have limited access to its most advanced accelerators, and reporting indicates domestic vendors have gained from Nvidia’s retreat, although Huawei’s leadership is not assured and Chinese competitors remain active. 78

• Chinese commentary has largely cast the announcement as evidence that external restrictions can stimulate indigenous innovation and competition. That political reading should be separated from technical validation: enthusiasm does not substitute for independent performance, cost, yield, and reliability evidence. 12

• On the specific question of Chinese official and public calls for U.S.–China AI cooperation, the evidence available here is insufficient to attribute a single, authoritative reaction directly to the LogicFolding announcement. The broader policy tension is clear—China seeks technological self-reliance while both sides face incentives to manage AI risks—but this announcement is principally a semiconductor-competition story, not evidence of a settled cooperation framework.