JupiteX Get the app
International Relations17 Nov 2026 · about 7 min

Trump administration moves to block China from indirect access to American AI tech

The brief

The United States has added restrictions aimed at closing a gap in its controls on advanced computing technology. The new rule covers remote access to advanced U.S. chips and servers, which can let companies use computing power without importing the hardware themselves. The goal is to stop restrictions being bypassed through data centers in other countries. Overseas data centers serving companies in countries including China, Russia, Iran and Cuba must identify their end users. Separately, chip sales for Indonesia, Malaysia, Singapore, Thailand and Brazil will face additional licensing requirements on a case-by-case basis. The rule describes those five countries as having a higher-than-normal risk of diversion to unauthorized destinations. The Commerce Department’s Bureau of Industry and Security issued the rule Monday, and it took effect immediately. The United States already bans sales of its most advanced AI chips to China. The new measures add oversight of remote computing and certain chip sales, while the article does not detail how individual applications will be decided.

01

What new restrictions did the U.S. introduce on remote access to advanced chips and servers?

The United States has added restrictions aimed at closing a gap in its controls on advanced computing technology. The new rule covers remote access to advanced U.S. chips and servers, which can let companies use computing power without importing the hardware themselves. The goal is to stop restrictions being bypassed through data centers in other countries.

Overseas data centers serving companies in countries including China, Russia, Iran and Cuba must identify their end users. Separately, chip sales for Indonesia, Malaysia, Singapore, Thailand and Brazil will face additional licensing requirements on a case-by-case basis. The rule describes those five countries as having a higher-than-normal risk of diversion to unauthorized destinations.

The Commerce Department’s Bureau of Industry and Security issued the rule Monday, and it took effect immediately. The United States already bans sales of its most advanced AI chips to China. The new measures add oversight of remote computing and certain chip sales, while the article does not detail how individual applications will be decided.

02

What is cloud computing, and how can it let a company use chips located in another country?

Cloud computing is a way to use computing resources over a network, often through a service provider, rather than buying and operating the hardware yourself. A company can request processing power and storage from servers in a data center. This general explanation draws on established knowledge; the article does not define cloud computing.

The key point is that the user and the chips do not have to be in the same place. A company in China, for example, might remotely use computing power from chips installed in a data center in Southeast Asia. The article describes providers serving Chinese companies with chips housed in Southeast Asian facilities, including in Malaysia.

That arrangement matters because the United States bans sales of its most advanced AI chips to China, while some policymakers say cloud access can leave a loophole. The new U.S. rule requires certain overseas data centers to identify end users. It is intended to make remote access harder to use to get around restrictions, though the article does not spell out every compliance step.

03

Which countries' data centers must identify their end users, and which five countries face extra licensing checks for chip sales?

The rule sets out two related but different requirements. Overseas data centers serving companies in certain countries must identify who their end users are. The named countries are China, Russia, Iran and Cuba. This focuses on who is using the computing service, rather than only where the data center is located.

A separate provision adds case-by-case licensing requirements for sales of chips for Indonesia, Malaysia, Singapore, Thailand and Brazil. The rule calls these countries higher-than-normal risks for diversion to unauthorized destinations. These are additional checks on chip sales; they are not described as a blanket ban on sales to those countries.

The distinction matters because remote computing can give companies access to chips located abroad, while chips can also be sold into countries that pose diversion risks. The Commerce Department’s Bureau of Industry and Security said the rule is meant to prevent restrictions from being circumvented through third-country data centers. The rule took effect immediately, but the article does not detail how end-user identification or licensing decisions will work in practice.

04

How could the new rule affect Chinese companies that rent computing power from data centers in Southeast Asia?

The rule could make it harder for Chinese companies to use U.S. computing technology through data centers outside China. U.S. policymakers have worried that cloud computing lets companies rent processing power built on American chips, even when direct chip sales to China are restricted. The new requirements target that remote-access route.

For example, the article says providers use chips housed in Southeast Asian data centers, including facilities in Malaysia, to provide computing power to Chinese companies. Under the rule, overseas data centers serving companies in China must identify their end users. This could bring greater scrutiny to who is actually receiving the computing service.

The rule does not say that every Chinese company’s access will be cut off, or describe a specific outcome for any provider. It took effect immediately and is intended to stop restrictions being circumvented through third-country data centers. The article also notes that Malaysia is among five countries whose chip sales face additional case-by-case licensing checks, due to diversion concerns.

05

Why do U.S. export controls restrict sales of the most advanced AI chips to China?

The article gives a strategic reason for the U.S. restrictions: the United States wants to maintain an edge over China in the competition for predominance in artificial intelligence. The most advanced AI chips provide computing power used to train AI models, so access to them is a focus of export controls.

The United States bans sales of its most advanced AI chips to China. But the article says many lawmakers and policymakers believe existing controls are not tough enough. They point to possible loopholes, including cloud computing, which can let Chinese companies rent processing power built on American technology rather than import chips directly.

The new rule extends oversight to remote access and certain chip sales through third countries. The article reports that a bill introduced by U.S. Rep. Michael Lawler to include remote access in export controls has passed the House. It does not state whether the bill became law or describe what next steps Congress might take. The immediate rule is presented as another effort to limit China’s ability to use U.S. chips for advanced AI.

06

How can chips in a third country be diverted to an unauthorized user, and why can that be hard to detect?

Diversion means technology reaches a destination or user that was not authorized to receive it. In general, physical goods can be diverted through resale or rerouting; the article does not specify the exact route suspected in the case it describes. It reports that Megaspeed International was suspected of diverting advanced Nvidia chips to China in violation of U.S. export controls.

Remote computing creates a different challenge. A company can use processing power from chips housed in a data center in another country, so the hardware stays put while the user accesses it over a network. The article says some lawmakers see this as a way to sidestep controls that apply to physical exports.

These arrangements can make it harder to see who ultimately benefits from the technology: a chip’s location may not reveal who is using its computing power. That is a general explanation, not a specific detection method described in the article. The new rule requires certain overseas data centers to identify end users, aiming to reduce the risk of restrictions being circumvented.

07

What is AI model training, and why does it require so much computing power?

AI model training is the process of using examples and computing to adjust a model so it can produce useful outputs. This is general background knowledge; the article does not explain the training process. It does say the U.S. rules are meant to limit China’s ability to use American chips to train its best AI models.

Training can require a lot of computing because the system must process data and repeatedly update its internal settings. More advanced work may need substantial processing power. The article’s concrete example is that companies can rent computing power from data centers, including facilities in Southeast Asia, where advanced chips are housed.

The United States bans sales of its most advanced AI chips to China, aiming to preserve an edge in the AI rivalry. Some policymakers say cloud computing can let Chinese firms access processing power built on U.S. technology despite export controls. The new rule requires certain overseas data centers to identify end users, seeking to close that remote-access gap. It does not set out specific limits on AI training itself.

This brief was written by AI from the original reporting and checked by other models. Names, figures and quotes come from the source; read it for full context.

Read more in the JupiteX app

Pulse is free. New stories every 4 hours, each one broken into the questions that explain it.

Or read more news on the web