Photo: Matheus Bertelli / Pexels
Washington plans to stop Chinese entities from controlling Nvidia hardware located outside China, targeting a loophole that export rules have struggled to close
The US Commerce Department plans to block China from remotely controlling Nvidia chips that sit outside Chinese borders.
For years, export controls have worked on a simple premise: stop the chip from crossing the border. The problem is that in the age of cloud computing, the chip never has to move.
What the Commerce Department is targeting #
Advanced AI chips work much the same way. A company can rent processing power from a data center in another country and run workloads on it from anywhere, without a single crate of hardware ever touching Chinese soil.
That is the gap Washington now wants to close. The plan would extend restrictions beyond physical shipments to cover remote control of Nvidia hardware abroad.
The move builds on licensing guidance the department clarified in May 2026. Under that guidance, China-headquartered firms must obtain licenses for GPU chips, including Nvidia’s Blackwell and H200 models, even when those chips are located outside China.
The guidance was aimed at loopholes involving overseas subsidiaries, with Malaysia specifically flagged. A Chinese parent company with a Malaysian subsidiary could, in theory, have accessed hardware that the parent itself was barred from buying.
In practical terms, the rules are shifting their focus from where a chip is to who is using it.
How the rules got here #
The remote access push did not appear out of nowhere. In January 2026, the House passed the Remote Access Security Act by a 369-22 vote.
That bill expands the definition of an export to include cloud-based access to restricted hardware. Put plainly, logging into a restricted chip from a distance could count the same as having it shipped to you.
The crypto and markets news you actually need—every morning.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
Scrutiny has also intensified around how restricted chips may be reaching Chinese companies through Southeast Asian intermediaries. Investigations into the potential diversion of chips through the region have added pressure for tighter rules.
As of early November 2026, investigations were ongoing into approvals granted to Chinese firms linked to restricted entities.
Congress is also pursuing additional measures, including chip tracking requirements, as part of the FY2027 defense authorization bill.
The Hong Kong wrinkle #
About 600,000 Nvidia H200 chips were reported to be stockpiled in Hong Kong as of November 2026.
Those chips were reportedly waiting on import approval from mainland China, amid ongoing geopolitical tensions. So the bottleneck in that case sits on the Chinese side of the approval process, not the American one.
What this means for Nvidia and the chip trade #
For Nvidia, the direction of travel is clear even if the final details are not. Each new rule narrows the set of customers who can use its most advanced products without government sign-off. Cloud providers face a particular kind of exposure. If remote access counts as an export, data center operators outside China may have to verify their customers far more carefully, including tracing corporate ownership back through subsidiaries.
There are several things worth watching from here. The first is whether the Remote Access Security Act clears the rest of the legislative process and becomes binding law.
The second is what chip tracking provisions survive in the FY2027 defense authorization bill.
The third is the outcome of the ongoing investigations into approvals for firms tied to restricted entities.
And then there are those roughly 600,000 H200 chips in Hong Kong.
The broader takeaway for the industry is that the definition of an export is expanding. Hardware location used to be the whole story. Increasingly, control and ownership are what regulators care about.
Disclosure: This article was edited by John Chen. For more information on how we create and review content, see our Editorial Policy.