Huawei unveils 1.4nm-equivalent chip target despite US sanctions

Huawei Technologies Co expects its premium chips to achieve transistor density equivalent to 1.4-nanometer processes by 2031 under a new semiconductor development framework, the company announced on Monday at a conference in Shanghai.

He Tingbo, Huawei board member and president of its semiconductor business, formally introduced the "Tau Scaling Law" in a keynote speech titled "New Semiconductor Path in Practice."

Unlike traditional approaches that rely on miniaturizing transistors — an industry paradigm increasingly challenged by physical limits and diminishing cost benefits — the new methodology prioritizes reducing signal propagation time across devices, circuits, chips and systems, He said.

Huawei has already designed and mass-produced 381 chips based on the Tau Scaling Law over the past six years, covering industries from smartphones to AI computing, He said.

This autumn, the company will launch a new Kirin smartphone chip that adopts a multi‑layer circuit architecture that shortens critical‑path wiring and improves transistor density and energy efficiency for the first time.

The announcement underscores China's push to circumvent US export restrictions that have limited domestic access to advanced lithography tools and cutting‑edge semiconductor manufacturing.

China is widely seen as unlikely to reach 1.4nm capabilities through conventional fabrication alone, but a successful application of the Tau Scaling Law could offer a way to improve performance and chip density despite equipment constraints, experts said.

Huawei's projection places it on a competitive timeline with TSMC, the world's largest advanced chip foundry. TSMC currently produces chips on 2nm technology and plans to begin mass production of its A14 (1.4nm) process in 2028.

Beyond Huawei's internal roadmap, Chinese AI developers are increasingly turning to domestic chips. In its DeepSeek‑V4 technical report, DeepSeek listed Huawei's Ascend NPU alongside Nvidia GPUs within the same hardware validation framework — the first time the company has positioned a Chinese AI chip on equal footing with Nvidia in an official document. The model has already completed inference adaptation on Huawei's Ascend platform.

Meanwhile, Kimi's recent research paper on cross‑datacenter inference architecture has also hinted at exploring domestic‑chip‑based token cost reduction.

As overseas advanced computing procurement tightens, the shift toward domestic AI chips is accelerating. He from Huawei noted that the semiconductor industry's future depends on open collaboration, adding that no single company can independently find all the answers.

"With the Tau Scaling Law, we look forward to working closely with scientists, engineers and industry partners around the world to drive the sustainable development of the semiconductor industry," He said.