The chips would then be shipped to those locations, where they could be accessed by their Chinese parent companies
In a move that extends the logic of technological containment beyond borders and legal entities, the United States has closed a quiet but consequential loophole that allowed Chinese firms to acquire advanced American AI chips through overseas subsidiaries and intermediaries. The Commerce Department's clarification signals that the intent of export law — not merely its letter — will now govern enforcement, reaching into global supply chains wherever Chinese ownership or control can be traced. This is less a single policy shift than another turn in a long reckoning between two powers competing for dominance over the foundational technologies of the coming era.
- Chinese companies had been routing chip purchases through offices in Singapore, the UAE, and other third countries — following the letter of the law while quietly defeating its purpose.
- The Commerce Department has now declared that Chinese ownership or control, wherever it exists in the world, triggers the same restrictions as a direct sale to a China-based entity.
- Both Nvidia and AMD are affected, meaning the two most consequential American AI chip manufacturers are now bound by rules that follow corporate lineage rather than geography.
- The announcement landed as Nvidia's Jensen Huang prepared to headline COMPUTEX in Taiwan, underscoring how visibly the U.S. is willing to wield export controls as a competitive instrument.
- Chinese firms now face a narrowing set of options: accept less capable chips, seek special licenses, or accelerate domestic alternatives that still lag years behind American and Taiwanese manufacturers.
- The policy assumes — and implicitly declares — that U.S. enforcement capacity is now sophisticated enough to track semiconductor flows and identify the beneficial owners behind complex corporate structures.
The United States moved this week to seal a gap in its AI chip export controls — one that Chinese companies had been quietly exploiting for months. Rather than purchasing advanced Nvidia and AMD semiconductors directly, firms with Chinese ownership had been routing orders through subsidiaries in Singapore, the UAE, and other countries beyond China's borders. The chips arrived at those locations legally, then flowed back to their intended beneficiaries. The letter of the law was satisfied; its purpose was not.
The Commerce Department's new clarification ends that arrangement. Export restrictions now follow ownership and control, not geography. A Chinese firm's subsidiary in Southeast Asia, a shell company in a third country, an investment vehicle designed to obscure the ultimate recipient — none of these structures will shield a transaction from enforcement. If Chinese ownership or control can be established, the restrictions apply.
The announcement arrived as Nvidia CEO Jensen Huang was preparing to headline COMPUTEX in Taiwan, one of the semiconductor industry's most significant annual gatherings. The juxtaposition was pointed: even as the industry convened to celebrate its advances, Washington was tightening the perimeter around who may access them.
This action fits a recognizable pattern. Over the past two years, U.S. export controls have been progressively refined — each iteration closing gaps that companies discovered in the version before. The implicit message is that this process will continue for as long as the competition does.
For Chinese firms, the path forward has narrowed considerably. Geographic arbitrage and creative corporate structuring are no longer viable. What remains are older, less capable chips; the uncertain prospect of special government licenses; or a faster push toward domestic alternatives — alternatives that, despite heavy investment, still trail American and Taiwanese manufacturers by years. The U.S., for its part, has made its position clear: there is no indirect route to its most advanced AI technology, no matter how distant the subsidiary or elaborate the structure.
The United States moved this week to close a significant gap in its restrictions on advanced artificial intelligence chips, targeting a workaround that Chinese companies had been using to acquire the technology through overseas operations. The new enforcement action extends export controls beyond the direct sale of chips to China-based entities, instead reaching into the global supply chains where Chinese firms maintain subsidiaries and use intermediaries to procure Nvidia and AMD semiconductors.
For months, American export restrictions had focused narrowly on preventing shipments directly into China. But companies with Chinese ownership or control discovered they could circumvent these rules by placing orders through offices in Singapore, the United Arab Emirates, and other countries outside China's borders. The chips would then be shipped to those locations, where they could be accessed by their Chinese parent companies or transferred onward. It was a straightforward loophole: the letter of the law was being followed even as its intent was being sidestepped.
The Commerce Department's clarification makes clear that this path is now closed. The restrictions now apply to any company with Chinese ownership or control, regardless of where that company is physically located or where it places its orders. This means a Chinese firm operating a subsidiary in Southeast Asia cannot use that subsidiary to buy advanced AI chips from American manufacturers. The same applies to shell companies, investment vehicles, and other structures designed to obscure the ultimate beneficiary of the technology.
The timing of the announcement is notable. Nvidia's chief executive, Jensen Huang, was preparing to headline the COMPUTEX conference in Taiwan, one of the semiconductor industry's most important annual gatherings. The policy shift underscores how seriously the U.S. government views the competition for AI chip dominance and how willing it is to use export controls as a tool in that competition. AMD, which also manufactures advanced processors, faces the same new restrictions.
The move reflects a broader hardening of American policy toward Chinese access to cutting-edge semiconductors. Over the past two years, the U.S. has progressively tightened rules around what chips can be sold to China, what performance thresholds trigger restrictions, and what companies are subject to enforcement. Each iteration has been designed to close gaps that companies discovered in the previous version. This latest action suggests that pattern will continue—that as Chinese firms find new ways to acquire the technology, American regulators will find new ways to block them.
For Chinese companies, the implications are significant. They can no longer rely on geographic arbitrage or corporate structure to access the most advanced chips from American suppliers. They will need to either accept older, less capable technology, negotiate special licenses from the U.S. government, or accelerate their own domestic chip development programs. Several Chinese chipmakers have been investing heavily in alternatives, though they remain years behind American and Taiwanese manufacturers in performance and reliability.
The policy also signals confidence within the U.S. government that it can enforce these rules effectively. Export controls only work if they can be monitored and violations can be detected and punished. The Commerce Department has been building its enforcement capacity, working with customs agencies and private companies to track semiconductor shipments and identify suspicious patterns. The new rule assumes that capability exists and will continue to improve.
What remains unclear is how strictly the new restrictions will be applied in practice, and whether Chinese companies will find yet another workaround. History suggests they will try. But for now, the U.S. has made its position unmistakable: there is no indirect route to American AI chips for Chinese firms, no matter how creative the corporate structure or how distant the subsidiary.
Citações Notáveis
The restrictions now apply to any company with Chinese ownership or control, regardless of where that company is physically located or where it places its orders.— U.S. Commerce Department policy clarification
A Conversa do Hearth Outra perspectiva sobre a história
Why does it matter where a Chinese company is located when it buys a chip? Isn't a sale a sale?
Because location determines which laws apply. A sale to a company in Singapore isn't technically a sale to China, so it wasn't covered by China export controls. The loophole was that Chinese-owned companies could use that fact to their advantage.
So the U.S. is saying ownership matters more than location now?
Exactly. They're saying if you're controlled by Chinese interests, you're treated as a Chinese entity for export purposes, no matter where your office is.
What do Chinese companies do now? Just accept slower chips?
They have a few paths. They can try to get special licenses, which is slow and uncertain. They can use older technology that's already approved for export. Or they can pour more money into building their own chips, which is expensive and takes years.
Is this the end of the loophole, or will companies find another way?
History suggests they'll try. But each time they do, the U.S. tightens the rules further. It becomes an escalating game of cat and mouse.