What did Huawei announce about AI chips?

Huawei unveiled a new semiconductor approach called Tau Scaling Law and LogicFolding, which aims to improve AI chip performance without relying on advanced Western lithography machines restricted by US sanctions.

What is Tau Scaling Law?

Huawei described Tau Scaling Law as an alternative semiconductor scaling method focused on vertical chip stacking and shorter signal pathways rather than relying only on shrinking transistor sizes through traditional lithography advances.

What does “1.4nm-equivalent” chip density mean?

Huawei claims its architecture could eventually achieve transistor density comparable to future 1.4nm manufacturing processes by 2031. Smaller node sizes generally improve performance and efficiency in advanced processors.

Why are US sanctions central to this story?

Since 2019, the United States has restricted Chinese access to advanced semiconductor tools, including EUV lithography machines produced by ASML. Those restrictions aimed to slow China’s AI and chip development.

How is Huawei trying to bypass EUV lithography restrictions?

Instead of depending entirely on smaller transistors created through EUV manufacturing, Huawei is focusing on architectural changes such as stacking, routing efficiency and denser chip design methods.

Does Huawei currently compete directly with Nvidia’s AI chips?

Not yet at the same global scale. Huawei has not released independent benchmarks proving its chips can match Nvidia’s highest-end AI processors in large-scale AI training workloads.

What advantages does Nvidia still maintain?

Nvidia continues to dominate AI infrastructure through its CUDA software ecosystem, deep partnerships with Taiwan Semiconductor Manufacturing Company and leadership in hyperscale AI systems used outside China.

Image for illustrative purposes only. Not a real photo.

Huawei unveils AI chip scaling challenge to Nvidia

Q: Why is Huawei’s announcement important for Nvidia?

Analysts said the announcement challenges a key assumption supporting Nvidia’s high valuation: that advanced AI computing power will remain scarce, expensive and difficult to manufacture outside a few Western supply chains.

Written by Jon CuthbertPublished May. 26 2026Last Updated May. 26 2026

Huawei unveiled a new semiconductor architecture called Tau Scaling Law and LogicFolding, presenting what it described as an alternative path toward advanced AI chips without relying on restricted Western lithography systems.

The announcement was made during the 2026 IEEE International Symposium on Circuits and Systems in Shanghai, where Huawei said the technology could eventually achieve transistor density equivalent to 1.4nm chip processes by 2031.

Huawei said the architecture focuses on vertical chip stacking and shorter internal signal pathways rather than relying solely on smaller transistor sizes produced with extreme ultraviolet lithography machines.

The development directly addresses restrictions imposed by the United States since 2019 that blocked Chinese firms from accessing advanced semiconductor manufacturing equipment supplied by companies including ASML.

Analysts said the announcement could challenge one of the major assumptions underpinning Nvidia’s massive valuation: that advanced AI computing power will remain scarce, expensive and difficult to manufacture at scale.

“If China can produce advanced computing power cheaply and at massive scale, the scarcity premium that justifies Nvidia’s valuation disappears entirely,” analyst Bull Theory said following the announcement.

Huawei said the first commercial products using the new architecture would appear in future Kirin smartphone chips before expanding into the company’s Ascend AI processor lineup later this decade.

Despite the attention surrounding Huawei’s announcement, analysts noted Nvidia still maintains major advantages through its CUDA software ecosystem, partnerships with Taiwan Semiconductor Manufacturing Company and leadership in hyperscale AI infrastructure outside China.