The United States is no longer trying to block only direct chip shipments to China.
It is now trying to close the side doors too.
The Commerce Department’s latest guidance targets a loophole that may have allowed Chinese companies to access advanced Nvidia AI chips through subsidiaries outside China. That matters because the AI chip war has moved beyond simple export controls. It is now a global chase through subsidiaries, data centers, shell structures, overseas affiliates, and supply-chain blind spots.
This is no longer just about what gets shipped to Beijing or Shanghai.
It is about what gets shipped to Malaysia, Singapore, Dubai, or anywhere else a Chinese-linked company can operate beyond China’s borders.
The Loophole Was the Whole Problem
Export controls are only as strong as the weakest route around them.
For years, Washington has tried to keep China away from the most advanced AI chips needed to build frontier artificial intelligence systems. But if Chinese companies can simply buy through offshore subsidiaries, then the policy becomes much weaker in practice than it looks on paper.
That is the danger the U.S. is now trying to address.
A ban that stops direct sales but allows overseas workarounds is not a wall. It is a fence with a gate left open.
AI Chips Are Now Strategic Weapons
The reason Washington cares so much is simple: advanced AI chips are not ordinary commercial goods anymore.
They are the foundation of frontier AI models, military-relevant computing, surveillance systems, cyber capabilities, autonomous weapons research, and national technological power. Whoever controls access to high-end chips controls the pace at which rivals can build advanced AI.
That is why Nvidia’s Blackwell processors matter so much.
They are not just products. They are strategic assets.
China’s Firms Are Global, So Controls Have to Be Global
This is the part export-control policy is finally catching up to.
Chinese technology companies do not exist only inside China. They have subsidiaries, cloud operations, research teams, shell buyers, and business relationships across the world. If U.S. rules apply only to geography and not corporate control, then firms can exploit the gap.
The new Commerce guidance focuses on entities headquartered in China even when they are physically located outside China.
That is the key shift.
Washington is saying the identity of the buyer matters, not just the destination on the shipping label.
Nvidia Is Caught in the Middle Again
Nvidia remains the company at the center of this whole fight.
It wants to sell chips. China wants chips. U.S. officials want to prevent China from gaining access to the most powerful chips. Investors want Nvidia to keep growing. Regulators want to stop leakage. Every side has a different priority.
That is why Nvidia is trapped between commercial logic and national-security logic.
Commercial logic says sell to the largest market possible.
National-security logic says restrict access before a rival gains too much capability.
In the AI age, national security is winning more often.
The Scale Could Be Enormous
One industry source estimated that hundreds of thousands of chips may have reached Chinese subsidiaries during the period when enforcement gaps remained open.
If that estimate is even close, the implications are serious.
It would mean Chinese-linked firms may already have acquired major quantities of high-end AI hardware despite Washington’s broader effort to limit access. That would not just be a technical failure. It would be a strategic failure.
Because once chips are delivered, they cannot easily be pulled back from the global AI race.
The Guidance Does Not Undo What Already Happened
That is an important point.
The new U.S. guidance may stop future shipments or make them harder. But it does not necessarily remove chips already deployed in overseas data centers. It also does not require data centers to stop using or servicing advanced computing systems already in place.
That means Washington is trying to stop the next wave, not fully reverse the last one.
This is the harsh reality of export controls: enforcement usually arrives after clever buyers have already tested the edges.
The AI Diffusion Rule Still Haunts the Story
The loophole appears to be tied to the Trump administration’s decision not to enforce the Biden-era AI Diffusion rule.
That rule was designed to regulate global access to advanced AI chips more broadly. When enforcement was relaxed, the door opened for uncertainty, and uncertainty is exactly what sophisticated buyers exploit.
Now Washington is trying to patch the gap.
But the episode shows how inconsistent policy can create strategic openings. In technology competition, even a year of ambiguity can matter.
Another Loophole May Still Remain
Even this new guidance may not close everything.
One expert warned that the rules still do not fully require foundries like TSMC to conduct extra due diligence to ensure high-end AI chips they produce are not ultimately destined for Chinese front companies.
That matters because the chip supply chain is layered.
A buyer may not appear Chinese at first glance. A customer may use intermediaries. A company may operate through jurisdictions that look neutral. If foundries and suppliers are not required to aggressively investigate end users, the system can still be gamed.
Export controls are becoming a game of paperwork, ownership, routing, and intent.
China understands that.
So does Washington.
The Real Fight Is Over Compute Sovereignty
This entire battle is really about compute sovereignty.
The United States wants to preserve its lead by denying China access to the most advanced AI infrastructure. China wants enough chips to build powerful models without being dependent on American approval. Each side sees compute as a national resource.
That means the chip war will not calm down soon.
Every new U.S. restriction will push China to find alternatives. Every Chinese workaround will push Washington to tighten enforcement. Every tightened rule will push more effort into domestic Chinese chips, smuggling routes, offshore data centers, and third-country partnerships.
This is a cycle, not a one-time policy fix.
The Global South May Become the New Battlefield
Countries like Malaysia and other regional hubs may become increasingly important in this fight.
If Chinese-linked AI firms use overseas subsidiaries to acquire chips or run data centers, then countries outside the U.S.-China rivalry become part of the enforcement map. They may face pressure from Washington, business opportunities from Chinese companies, and difficult choices about how closely to police advanced technology flows.
That is how great-power competition spreads.
It does not stay between Washington and Beijing.
It moves through ports, cloud facilities, chip distributors, regulators, and corporate registration offices around the world.
This Could Accelerate China’s Domestic Chip Push
The more Washington tightens access, the harder China will push to build its own alternatives.
That is the long-term paradox of U.S. chip controls. They may slow China in the short term, but they also give Beijing a powerful reason to pour more money into domestic semiconductors, alternative architectures, and local AI infrastructure.
China may not match Nvidia overnight.
But every restriction makes self-sufficiency more urgent.
The Meaning of the Moment
The Commerce Department’s new guidance is not just a technical clarification.
It is a signal that the United States believes its AI chip controls have been leaking through offshore subsidiaries and that the fight must now extend beyond China’s borders. Washington is no longer only asking where the chip is going. It is asking who ultimately controls the company receiving it.
That is a major escalation in the AI technology war.
The message is clear: if Chinese firms cannot buy the chips directly, they should not be able to buy them indirectly either.
But the bigger lesson is even clearer.
In the AI age, the battle for chips will not be fought only in factories and export offices. It will be fought through subsidiaries, data centers, loopholes, shipping routes, licensing letters, and every shadowy corner of the global supply chain.
The chip war has gone global.
