Huawei’s Her’s Law eyes AI chips as China reduces Nvidia reliance

• Huawei’s Her’s Law focuses on AI chip performance beyond smaller transistors.
• IDC data showed Nvidia still led China’s AI accelerator market in 2025.

Huawei is promoting a new chipmaking principle as Chinese technology companies compete to reduce reliance on Nvidia’s AI accelerators.

The development, reported by Reuterscomes as Huawei, Alibaba, and other Chinese chipmakers outline plans to expand their role in China’s AI hardware market under continued US technology restrictions.

The company this week said it had discovered and applied what it calls the Tau Scaling Law. It is also referred to as “Her’s Law” after He Tingbo, a long-serving Huawei executive involved in the company’s semiconductor work.

A new route for chip performance

Huawei has presented the idea as an alternative to Moore’s Law, the long-standing industry benchmark based on shrinking transistors and placing more of them on a chip. The approach has faced limits as transistor sizes have continued to shrink.

Huawei said its proposal focuses on system-level performance, including how groups of chips are connected and integrated across AI clusters and data centres.

Huawei said Yes Scaling Law is designed to shorten the time constant used in chip operations. The company said the approach is aimed at improving performance and energy efficiency across multiple layers of chip design. It also said LogicFolding is one of the core technologies developed under the approach.

The company said the principle would allow it to reach transistor density equivalent to 1.4 nanometres by 2031. Taiwan Semiconductor Manufacturing Co is expected to begin producing 1.4nm chips in 2028.

Huawei’s announcement comes as the United States continues to restrict China’s access to advanced chipmaking tools, including ASML lithography machines used to manufacture leading-edge semiconductors. Huawei and other Chinese companies have been among the targets of those controls.

The company has repeatedly sought to show that its chip business can continue despite those restrictions. Huawei’s roadmap includes packaging, integration, and cluster efficiency as part of its chip development plans.

The approach is also tied to Huawei’s wider AI chip ambitions. The company said techniques linked to Her’s Law would appear in phone chips later this year, while full integration into its Ascend AI chips is expected in 2030.

Huawei said the method is aimed at reducing the time and energy spent moving data between and within chips.

Limits of the approach

The company has not provided a full explanation of how the approach would address engineering issues such as heat. It also remains unclear how closely transistor density projections would translate into real-world AI chip performance.

Chipmakers and researchers in the United States, China, and elsewhere have also worked on advanced packaging and chiplets to improve performance without relying only on smaller transistors.

China’s AI chip market widens

The announcement comes as China’s domestic AI chip market becomes more competitive. Alibaba’s chip design unit, T-Head, last week unveiled the Zhenwu M890 AI chip. It also outlined a multi-year roadmap for future performance gains.

Alibaba has not been placed under the same US sanctions as Huawei. It has also not framed its AI chip plans as directly around replacing Nvidia in the same way Huawei has.

Alibaba’s roadmap shows another Chinese supplier outlining plans for AI chip development. In China, the company is competing in AI models as well as the chips needed to run them. DeepSeek has drawn attention in the model market.

IDC data reviewed by Reuters showed that about 4 million AI accelerator cards were shipped in China in 2025. Nvidia remained the largest supplier with around 2.2 million units, giving it a 55% market share.

Chinese vendors shipped 1.65 million AI chips in 2025, accounting for about 41% of the market. Huawei led domestic suppliers with 812,000 units, while Alibaba’s T-Head followed with about 265,000.

Baidu’s Kunlunxin and Cambricon each shipped around 116,000 units, according to the same Reuters report. Huawei led Chinese suppliers, followed by Alibaba, Baidu’s Kunlunxin, and Cambricon.

Huawei faces domestic rivals

Other large Chinese technology companies, including ByteDance and Tencent, have the resources to compete in the domestic AI chip market. Startups are trying to capture part of the market as China looks for alternatives to Nvidia hardware.

The activity comes as Beijing continues to push for more domestic semiconductor capability under US restrictions.

Huawei has not shown that the approach can match the manufacturing advantages held by leading global chipmakers.

Chinese chipmakers are outlining AI hardware roadmaps while US controls continue to limit access to advanced manufacturing tools.

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology eventsclick here for more information.