The US-China semiconductor export control regime entered a new phase in early 2026 with policy reversals, expired exemptions, and mounting industry frustration — creating what industry analysts describe as a permanent state of regulatory uncertainty where "nobody won."
The most significant change came December 31, 2025, when the Validated End-User (VEU) exemption expired for Taiwan's TSMC, South Korea's Samsung, and SK Hynix. These companies previously enjoyed automatic access to US semiconductor equipment for their China operations. Now they require annual export licenses from the Commerce Department, giving Washington yearly renewal leverage over the world's leading chipmakers.
Nvidia's H200 chip became a case study in policy whiplash. Initially banned for export to China, it was unbanned in December 2025, then subjected to new conditions in January 2026 — including a 25% tariff and case-by-case licensing with volume caps around 1 million units. The zigzag left customers confused and Nvidia scrambling to provide guidance.
"The framework creates contradictions," according to an analysis by Semiconductors Insight. "The policy simultaneously acknowledges security risks while enabling the restricted activity." The result: Chinese firms accelerate domestic alternatives, meaning Nvidia faces lost market share that won't automatically return even if restrictions ease.
On the Chinese side, SMIC continues pursuing 5nm production using older equipment, while Huawei remains dependent on trailing-edge technology. But the pressure has turbocharged Beijing's self-sufficiency push. What was once a 10-year timeline for semiconductor independence has compressed to 3-5 years, according to industry observers.
US equipment makers — Applied Materials, Lam Research, KLA — maintain limited export channels but operate in constant uncertainty. Supply chain planning horizons have collapsed from 3-5 years to 12 months. Congressional pressure signals continued tightening ahead, with lawmakers calling current restrictions insufficient.
