Huawei's recent progress in advanced semiconductor design is sharpening an uncomfortable argument in Washington: US export controls may have created the competitor they were designed to suppress.

The Startup Fortune analysis points to Huawei's Ascend AI processor line as the clearest evidence. Despite being cut off from leading-edge foundry access since 2020 — when the Commerce Department blocked the company from fabs using US equipment, effectively severing its tie to TSMC — Huawei has continued to ship AI accelerators aimed at replacing Nvidia hardware in Chinese data centers.

The sanctions regime was built on a straightforward theory. Deny China's most strategically important technology firm access to US chipmaking tools, EDA software, and advanced fabrication, and its product roadmap would stall. That theory assumed Huawei depended on American suppliers and allied foundries for anything it could not build itself.

The record since then has complicat­ed the assumption. Huawei partnered with SMIC, China's largest contract chipmaker, to produce processors at what analysts identify as a 7nm-class node — demonstrated publicly in the Kirin 9000S smartphone chip inside 2023's Mate 60 Pro. The phone launched during a visit by a senior US official to Beijing, a timing widely read as a signal rather than a coincidence.

Since that moment, the argument in the source piece runs, US policy has functioned less as a barrier and more as a forcing function. Each round of restrictions pushed Chinese firms to accelerate import substitution across the semiconductor stack: domestic lithography efforts at SMEE, EDA tool development, advanced packaging, and above all a national drive to standardize on homegrown AI compute.

The commercial consequences are now visible. Chinese cloud and AI customers face US rules that restrict access to Nvidia's most powerful accelerators, and Huawei has positioned the Ascend family as the compliant alternative. Demand for domestic AI silicon has grown into a protected market — one that Western vendors cannot contest under current rules. Export controls, in effect, handed Huawei a captive customer base.

There are real limits to the counterargument. SMIC's 7nm-class output depends on DUV multi-patterning, a costlier and lower-yield method than the EUV lithography TSMC and Samsung use for equivalent nodes. Analysts estimate Huawei's AI accelerator shipments run well behind Nvidia's volumes, and gaps remain in interconnect performance, software maturity, and the HBM memory supply chain, where China still lacks a strong domestic option.

But the strategic picture has shifted. The question in 2020 was whether Huawei could make advanced chips at all. The question in 2025 is how fast its ecosystem can scale, and whether the gap to leading-edge Western hardware narrows from a domestic base that no longer imports the tools it needs to improve.

The Startup Fortune piece frames the outcome as an irony of policy design: controls calibrated to freeze a competitor in place instead financed its motivation. Every new restriction has validated, for Beijing, the case for self-sufficiency spending that now runs into tens of billions of dollars annually across wafer fabrication, equipment, and design.

For US chipmakers, the cost is not yet lost market share globally — Nvidia still dominates AI compute outside China. The cost is the Chinese market itself, once among the largest revenue pools for American semiconductor companies, now structurally walled off with a domestic champion growing behind the wall.

How Washington responds will shape the next phase. Tighter enforcement could slow SMIC's yield improvements and Huawei's accelerator roadmap; it could equally deepen the commitment in Beijing to close the remaining gaps. What is no longer in dispute, on the evidence Huawei has already shipped, is that export controls bought time — and Huawei spent that time building.