US export controls on advanced semiconductors bound for China escalated for three straight years before finally cooling off in 2026. The Bureau of Industry and Security’s initial controls, rolled out in October 2022, targeted advanced logic chips, integrated circuits, and semiconductor manufacturing equipment. They were tightened again in October 2023 and December 2024, and the Trump administration added further restrictions in March 2025, blacklisting dozens of additional Chinese entities from trading in semiconductors and other strategic technologies.
That multi-year tightening cycle has now partially reversed. Following a December 8, 2025 announcement from the Trump administration, BIS revised its licensing policy to review export applications for advanced AI chips, including Nvidia’s H200 and AMD’s MI325X, on a case-by-case basis rather than blocking them outright, provided certain security requirements are met. It’s a meaningful shift from blanket restriction toward a more negotiated, transactional posture.
The loosening comes with a deliberate catch. Even as it opens the door to case-by-case chip sales, the US has affirmed that its licensing requirements extend to subsidiaries of Chinese companies operating anywhere in the world, not just companies physically based in China. That closes an obvious loophole, using an offshore subsidiary to route around restrictions, that would have otherwise undercut the entire framework. The Commerce Department is reportedly unlikely to introduce fresh restrictions in the near term, with the current posture shaped heavily by active US-China trade negotiations rather than by any change in the underlying technology-security calculus.
The bigger structural trend sits underneath both the tightening and the current cooling: a bifurcated global chip ecosystem has already taken shape regardless of where US policy lands month to month. Both the US and China are now building parallel supply chains, chip architectures, and AI infrastructure increasingly designed not to interoperate with the other side’s stack. At this point, marginal easing of export controls functions more as a negotiating chip in broader trade diplomacy than as a serious attempt to slow China’s push toward chip self-sufficiency, which has been underway regardless of the restrictions for years.
The Philippines has real stakes in how this plays out beyond spectator interest. The reported $10 billion AI infrastructure hub taking shape near Clark Freeport Zone is being pitched partly on the premise that the Philippines can serve as a trusted, US-aligned node in chip and AI supply chains, distinct from China. A durable US-China truce on chips would reduce the urgency of that positioning for American investors; renewed escalation would likely accelerate it. Philippine policymakers courting this investment would do well to track Washington’s chip diplomacy as closely as they track the investment pledges themselves, since the two are more tightly linked than the headline dollar figures suggest.
Analysts studying the export control regime’s track record remain split on whether it ever worked as intended in the first place. Research from the Center for Strategic and International Studies has pointed out that export controls alone have consistently struggled to meaningfully slow China’s underlying semiconductor ambitions, since restricted access to cutting-edge chips like Nvidia’s H200 and AMD’s MI325X has mostly accelerated China’s own domestic chip investment rather than stalled its AI progress outright. That critique is part of why the current case-by-case licensing approach reads less like a policy reversal and more like an acknowledgment that blanket restriction was never going to be a permanent solution on its own.
Share this article