China is expanding its technology substitution strategy into one of the most sensitive areas of the digital economy: AI chips. The pressure from U.S. restrictions on advanced semiconductors has propelled Beijing and its major manufacturers to seek alternative avenues to reduce dependence on foreign technologies, especially in extreme ultraviolet lithography, AI accelerators, and next-generation manufacturing processes.
Huawei’s latest move has brought this debate to the forefront. The company presented in Shanghai the so-called Tau Scaling Law, a technical proposal that, according to Huawei, would enable progress toward high-end chips with a density comparable to 1.4-nanometer processes by 2031 without necessarily relying on the most advanced EUV tools. While ambitious and still to be proven in actual production, it demonstrates how China is attempting to turn sanctions into an engineering challenge, not just an access issue to machinery.
The core idea is not to immediately manufacture 1.4 nm chips with a conventional node similar to TSMC or Samsung. Huawei proposes a shift in approach: reducing internal signal transmission times and improving design efficiency instead of solely depending on geometric transistor miniaturization. This nuanced perspective is important because it prevents interpreting the announcement as an immediate overthrow of Taiwanese or South Korean industry dominance.
Huawei aims to bypass the EUV lithography bottleneck
U.S. restrictions have prevented China from normal access to the most advanced chip manufacturing equipment, especially the EUV machines from ASML, necessary to produce competitively at 3 nm nodes and below. This blockade has limited China’s ability to compete at the semiconductor frontier, though it hasn’t halted progress on mature nodes, packaging, design, and domestic manufacturing.
Huawei’s proposal seeks to circumvent this bottleneck. He Tingbo, Huawei’s scientific committee chair and a key figure in its semiconductor division, argued that the industry’s future cannot rely solely on continued shrinking of transistor physical sizes. Instead, Huawei advocates a strategy based on architecture, internal connectivity, and temporal efficiency.
Specialized media have linked this strategy to LogicFolding, an architecture aimed at increasing effective density and reducing internal delays through new chip organization methods. If successful at scale, this approach could help Huawei approximate the performance of more advanced nodes without access to the same manufacturing tools as its competitors.
However, there’s a significant gap between a technical proposal and a competitive industrial platform. China still faces challenges in lithography, materials, process control, wafer performance, advanced packaging, EDA software, and high-bandwidth memory supply. Huawei’s announcement alleviates some of the narrative pressure from sanctions but does not close the technological gap instantly.
AI forces China to accelerate
The landscape has shifted dramatically with the rise of Artificial Intelligence. Previously, China’s technological substitution mainly focused on servers, operating systems, databases, telecommunications equipment, and strategic electronics. Now, the focus is also on AI accelerators and all the infrastructure needed to train and run models.
The reason is clear. If Chinese companies can’t freely access NVIDIA’s most advanced GPUs or other U.S. suppliers, they need domestic alternatives to sustain their AI industry. Huawei, with its Ascend family, has already become one of the most relevant domestic players in this field, although it still faces disadvantages compared to NVIDIA’s CUDA ecosystem, mature software, and advanced manufacturing capacities outside China.
Replacing AI chips isn’t just about silicon. It requires compilers, libraries, development tools, networks, complete servers, cooling systems, data center integration, and a community of developers capable of handling real workloads. China is working to advance this systematically, combining government support, public procurement, pressure on local companies, and a narrative of technological self-sufficiency.
Less power for sanctions, but not immediately
If Huawei succeeds in turning its approach into competitive products, Washington would have less room to use export controls as a pressure tool. This is the most strategically relevant interpretation. Sanctions work best when the affected country has no viable alternatives; their efficacy diminishes when they motivate the development of a local supply chain, even if initially more costly or less efficient.
Still, it’s important not to overestimate the immediate impact. Huawei is talking about a roadmap toward 2031, not the current widespread availability of 1.4 nm chips. Moreover, reaching a comparable density doesn’t necessarily mean matching consumption, cost, performance, volume, reliability, or manufacturing ease.
The real test will be whether the architecture can be produced reliably, whether it can scale to large AI chips, whether thermal issues are managed, whether volume manufacturing is feasible, and whether software support is adequate. In semiconductors, a promising idea only truly alters the market when it can be delivered at scale—millions of devices or thousands of data centers at controlled costs.
What seems clear is that the technological rivalry between China and the U.S. is entering a more complex phase. Restrictions have slowed China’s access to critical tools but also accelerated efforts to develop indigenous solutions. Huawei aims to demonstrate that the future of chips depends not only on lithography machines but also on architecture, integration, and system-level design. This technological battle now features a new concept: Tau Scaling Law, although it will take years to see its actual results.
Frequently Asked Questions
What has Huawei announced?
Huawei has introduced the Tau Scaling Law, a proposal to improve chip efficiency and effective density through architecture and internal timing reduction, rather than relying solely on traditional miniaturization.
Does this mean China can now produce 1.4 nm chips?
No. Huawei is referring to an effective density comparable to 1.4 nm by 2031, not the immediate production of a real 1.4 nm node comparable to current industry leaders.
Why is this important for Artificial Intelligence?
Because China needs domestic AI chips to reduce reliance on foreign suppliers affected by export restrictions, especially for data centers and model training.
via: SCMP