China’s tech holding Huawei is rewriting the rules of chip design. A new scaling law unveiled in Shanghai this week proposes replacing half a century of semiconductor orthodoxy with a fundamentally different approach. Instead of racing to shrink the size of chips to ever more microscopic dimensions, it is reworking the engineering to make the chips work more efficiently to increase processing speed.

For more than half a century, the semiconductor industry has run on a single idea: Gordon Moore's 1965 law that transistor counts would double roughly every two years became the organising principle of the entire technology industry — an invisible metronome behind every generation of smartphones, data centres and artificial intelligence systems. And Moore’s law has held sway until now. However, as technological progress starts to bump up against the physical limits of just how small it is possible to reduce electronic components, scientists in China are changing tack.

In a keynote address at the 2026 IEEE International Symposium on Circuits and Systems in Shanghai on May 24, Huawei proposed replacing Moore’s law with a new idea.

He Tingbo, president of Huawei's semiconductor business and director of its Scientist Committee, introduced what the company calls the Tau Scaling Law in a speech titled "New Semiconductor Path in Practice." The principle, already being referred to by some of He's peers as "Her's Law," does not abandon the pursuit of higher performance. It abandons the assumption that higher performance must come from making transistors physically smaller.

The Tau Scaling Law proposes replacing geometric scaling with time scaling — using the Greek letter τ (tau) to denote signal propagation delay — as the new guiding principle for the evolution of both semiconductors and electronic systems. In plain terms: instead of building narrower roads, Huawei is shortening the distances signals need to travel. The results, the company claims, are chips that run faster and more efficiently — even without access to the world's most advanced fabrication equipment, Huawei said in a statement.

That will unsettle those in Washington that have been trying to limit China’s access to the most advanced fab equipment to ensure the US’ dominance in the tech business.

The sanctions context

Huawei's semiconductor ambitions did not emerge in a vacuum. Since 2019, the United States has subjected the company to a series of export controls, the CHIPS laws, that have progressively cut it off from leading-edge chip manufacturing technology — most critically from TSMC, the Taiwanese foundry that produces the most advanced chips in the world, and from ASML, the Dutch company whose extreme ultraviolet lithography machines are essential to manufacturing at the frontier.

The conventional assumption was that these controls would cap Huawei's chip performance at a level several generations behind the global frontier, indefinitely. The Tau Scaling Law is an overt challenge to that assumption and a clever workaround.

By 2031, Huawei says the high-end chips it designs based on the Tau Scaling Law are expected to feature a transistor density equivalent to 1.4 nanometre processes – the chips themselves will be bigger, but they will run as fast as if they had been shrunk to a mere 1.4nm.

That figure matters because 1.4nm is projected to be near the global frontier for advanced chipmaking around the end of the decade — putting Huawei's target in direct competition with what TSMC and Samsung are expected to be producing through conventional lithography at that point.

The Tau emphasis is on density equivalence achieved through architectural and system-level innovation rather than strict lithographic parity.

LogicFolding

The practical implementation of the Tau Scaling Law is a chip architecture Huawei calls LogicFolding. The architecture reduces the resistive and capacitive load of signal propagation, ultimately boosting transistor density — compressing the time signals take to travel between components rather than compressing the components themselves.

Huawei said its Kirin chips scheduled to launch in the autumn of 2026 would be the first to use LogicFolding, which the company said would shorten wiring inside chips and considerably improve performance. The Kirin line powers Huawei's flagship smartphones and has been the most visible symbol of the company's attempt to maintain a consumer electronics presence despite sanctions pressure.

Huawei has already used the Tau Scaling Law to design and mass produce 381 chips over the past six years, for use in industries including smartphones and AI computing. This is not a laboratory experiment. The underlying architectural principle has already been used in commercial production — the 1.4nm equivalence goal is just a destination, not the starting point.

Huawei plans to launch its Ascend 950 series, including the 950PR and 950DT models, in 2026, with the Ascend 960 following in 2027 and the Ascend 970 in 2028 — running parallel to AI chip releases from Nvidia and Advanced Micro Devices. The roadmap is explicit, public and — pointedly — timed to shadow its American competitors.

What the industry will want to know

So far Huawei’s speech is just that. No independent performance data has been provided to support the claims. The gap between a keynote projection and a mass-produced chip performing at such extraordinary specifications is a regular feature of the business and not always bridged.

The 1.4nm equivalence framing has an obviously political tint as China attempts to assert its technological prowess in its rivalry with America and Huawei is seen as much as an agent of the state as a commercial tech company.

Transistor density is one measure of chip capability; it is not the only one, and the relationship between density equivalence achieved through architectural innovation and the raw performance of a chip manufactured at the genuine process node is not straightforward. Samsung, TSMC and Intel each use different metrics and marketing approaches for process naming, and Huawei's framing suggests it may be attempting to compete on effective density and performance scaling rather than strict lithographic parity.

The question the industry — and Western governments — will now be asking is not whether Huawei has solved every problem in advanced semiconductor design. It is whether the assumption that export controls would permanently and decisively cap Chinese chip performance is still a safe one to make.

He Tingbo closed her Shanghai keynote with a line that was as much geopolitical statement as scientific principle. "We believe that openness and collaboration are the way forward," she said — from the stage of a conference from which, five years ago, the trajectory of Chinese semiconductor ambition looked considerably less certain.

Moore's Law took decades to reach its limits. Now that progress is hitting the physical bumpers imposed by physics, Huawei is betting it has found what comes next.