RPT-BREAKINGVIEWS-Huawei advance will mark new phase in chip war
NVIDIA Corporation NVDA | 0.00 | |
Synopsys SNPS | 0.00 | |
Cadence Design Systems, Inc. CDNS | 0.00 |
The author is a Reuters Breakingviews columnist. The opinions expressed are his own.
By Ka Sing Chan
HONG KONG, June 3 (Reuters Breakingviews) - China is gaining new confidence in its tech capabilities. Its AI champion Huawei says it can make industry-leading semiconductors within five years, effectively side-stepping Washington's sanctions and export controls. That would reduce demand in the world's second-largest economy for products made by Nvidia NVDA.O and other Western suppliers. The threat ups the pressure for U.S. President Donald Trump to tighten restrictions on China.
In a high-profile Shanghai conference last week, He Tingbo, Huawei's top engineer referred to as China's "chip queen", laid out a groundbreaking approach to designing and manufacturing processors dubbed the Tau Scaling Law. Manufacturers led by the $2 trillion Taiwan Semiconductor Manufacturing 2330.TW rely on shrinking microchips to pack in more transistors in order to make ever more powerful processors vital to AI. But that has become prohibitively expensive and technically complex given the near‑atomic precision required.
For Huawei, which has been cut off from American tech since 2019 under crippling U.S. sanctions, the workaround is to design chips with stacked circuitry. This shortens the distance signals travel, boosting performance to a level Huawei's He claims can match TSMC's most advanced chips.
It's not an entirely new concept. Industry giants TSMC and Samsung Electronics 005930.KS have been advancing breakthroughs in so-called stacking technology for years already. Whether Huawei's timeline is achievable is hard to assess but existing US restrictions, plus power and other technical constraints, mean the company will face formidable hurdles to delivering its workaround at scale.
But success would mark a big step forward in Beijing's campaign to break the West's stranglehold in tech supply chains. Nvidia, which outsources chipmaking to TSMC, currently dominates the market for processors used to train the latest AI models, including those from Chinese AI labs DeepSeek and Alibaba 9988.HK. The company led by Jensen Huang posted a whopping $130 billion in revenue in the last fiscal year, with China and Hong Kong accounting for about 13% of sales despite export curbs.

Huang's firm increasingly faces the risk of being locked out of China at a time when frontier technologies, from self-driving cars to robotics, are just taking off. DeepSeek and rivals are already eschewing Nvidia and other foreign chips for processing queries on already-trained AI models, known as inference, for domestic alternatives.
Washington can tighten up sanctions on Huawei, including targeting the group's subsidiaries. It might also expand curbs to cover more design kit and software to ensure other Chinese chipmakers can't achieve similar breakthroughs. That would squeeze U.S. suppliers like Synopsys SNPS.O and the $103 billion Cadence Design Systems CDNS.O. The Tau Scaling Law looks set to herald a supersizing of the U.S.-China chip war.
CONTEXT NEWS
Huawei Technologies on May 25 unveiled a new principle for improving chips. The Tau Scaling Law, as the Chinese company calls it, focuses on cutting the time it takes signals and data to move through chips instead of making the chips smaller.
By applying the Tau Scaling Law, Huawei claimed it could produce high-end chips with transistor density equivalent to 1.4 nanometre processors by 2031.
