JAKARTA - Huawei Technologies is preparing a new path to pursue premium chips with a transistor density equivalent to a 1.4 nanometer process by 2031. The target was announced when US technology restrictions were still pressing China's access to advanced chip equipment.
Quoted from China Daily, Tuesday, May 26, Huawei introduced a semiconductor development framework called Tau Scaling Law at a conference in Shanghai on Monday.
The framework was introduced by He Tingbo, a member of Huawei's board and President of Huawei's Semiconductor Business, in a keynote speech titled New Semiconductor Path in Practice.
Unlike the old way of chasing more advanced chips by constantly shrinking transistors, this approach focuses on reducing the propagation time of signals in devices, circuits, chips, and systems. In other words, Huawei is trying to improve chip performance by setting up faster and more efficient signal flows.
He said Huawei has designed and mass-produced 381 chips based on the Tau Scaling Law in the past six years. The chips are used for various sectors, from smart phones to artificial intelligence computing.
This fall, Huawei will launch a new Kirin chip for smartphones. The chip uses a multi-layer circuit architecture that is claimed to shorten important cable paths, increase transistor density, and save energy.
Huawei's move is important because China is still restricted in its access to advanced lithography tools. Lithography tools are the main machines for printing very small patterns on chips. Without these tools, making the most advanced chips is much more difficult.
Experts say China is unlikely to achieve 1.4 nanometer capability just through conventional fabrication. However, if the Tau Scaling Law is successfully applied, this approach could help improve chip performance and density even with limited equipment.
Huawei's target puts the company on a par with TSMC, the Taiwanese chip manufacturing giant. TSMC is currently producing 2-nanometer chips and plans to start mass production of the A14 or 1.4-nanometer process in 2028.
In the artificial intelligence sector, Chinese developers are also increasingly eyeing local chips. In the DeepSeek-V4 technical report, DeepSeek lists Huawei's Ascend NPU along with Nvidia's GPU in the same hardware validation framework. This is the first time DeepSeek has placed Chinese AI chips on par with Nvidia in its official documents.
The model has also completed inference adaptation on Huawei's Ascend platform. Inference is the process when an AI model executes commands or generates answers after being trained.
Kimi's latest research on cross-center data inference architecture also hints at the use of domestic chips to reduce token costs. Tokens are text units that are calculated in AI processing.
He said the future of the semiconductor industry depends on open collaboration. According to him, no one company can find all the answers on its own.
"With the Tau Scaling Law, we hope to work closely with scientists, engineers, and industry partners around the world to promote the sustainable development of the semiconductor industry," said He.
China Daily wrote that Huawei's progress came as a number of Chinese semiconductor companies began to enter deeper into the chip supply chain, from materials, packaging, manufacturing processes, to chip-making equipment.
Gartner Research Vice President Roger Sheng said Chinese chip companies showed resilience and continued innovation capabilities amid major challenges.
Head of China Economics at Morgan Stanley Xing Ziqiang said China's technological breakthrough was driven by three main forces: industrial clusters, a large number of science and engineering talents, and a very large domestic market.
According to Xing, China is expected to reach a localization rate of 50 percent in GPUs by 2027 or 2028.
The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)
