Huawei presented its Tau Scaling Law and LogicFolding chip architecture at the 2026 IEEE International Symposium on Circuits and Systems, claiming it can achieve 1.4nm-equivalent transistor density by 2031 without restricted Western lithography tools. The announcement comes as U.S. sanctions since 2019 blocked the company from accessing ASML's extreme ultraviolet machines meant to curb China's advanced computing progress. Huawei plans to integrate the design into Kirin smartphone chips later this year and Ascend AI processors before 2030.
Workaround for Sanctions
U.S. restrictions cut Huawei off from ASML's EUV lithography systems five years ago, forcing a redesign of its chip development strategy. The company focused on vertical stacking and signal delay reduction rather than transistor miniaturization to bypass equipment bans. This approach directly addresses the core limitation imposed by Western export controls on semiconductor manufacturing tools.
Vertical Stacking Approach
LogicFolding achieves density gains by stacking components vertically instead of shrinking transistors. The architecture prioritizes reducing signal delays between layers through new routing techniques. Huawei says this method avoids the need for lithography precision typically required for sub-2nm processes. The design will first appear in mobile devices before scaling to AI workloads.
Nvidia's Unshaken Lead
Nvidia maintains its global AI market dominance through the CUDA software ecosystem, TSMC manufacturing partnerships, and hyperscale infrastructure investments. No independent benchmarks exist yet to compare Huawei's architecture against Nvidia's high-end AI chips in large-scale training scenarios. The industry currently lacks real-world performance data for LogicFolding implementations.
Manufacturing Hurdles Remain
Unresolved challenges include manufacturing yields, power efficiency, heat dissipation, and memory integration at scale. Huawei hasn't provided production timelines beyond the Kirin and Ascend integration plans. Analysts note Nvidia's market position remains unchallenged for the foreseeable future despite Huawei's claims.
Huawei aims to embed LogicFolding in Kirin smartphone processors by late 2026, but independent performance validation against Nvidia's chips won't happen before 2027.
