The US Department of Commerce is preparing to introduce new export controls that would prevent Chinese artificial intelligence companies from remotely accessing advanced AI chips through data centres located in other countries. The proposed regulation aims to close a loophole that allows Chinese companies to rent computing capacity powered by US-designed processors without physically importing the restricted hardware into China. The move could significantly change how international data centre operators provide AI computing services to customers in restricted markets.
According to a November 11 report by The Information, the Commerce Department is finalising the long-awaited rule and could announce it as early as November 2026. The proposed restrictions would expand existing semiconductor export controls to seven additional countries, including Singapore, Malaysia, Thailand, Indonesia, Brazil, Norway and South Africa. The measures are expected to cover Nvidia’s Blackwell processors and more advanced chips, while introducing stricter customer disclosure, licensing and monitoring requirements.
US Plans New AI Remote Access Restrictions
The proposed regulation is being developed by the Commerce Department’s Bureau of Industry and Security (BIS), which oversees US export controls on sensitive technologies.
Under existing restrictions, certain advanced AI chips cannot be freely exported to China. However, Chinese technology companies have found alternative ways to access computing resources by renting processing capacity from data centres located in countries with fewer restrictions.
This arrangement allows companies to use powerful AI processors remotely through cloud computing services without physically bringing the chips into China.
The proposed rule aims to address this practice by extending licensing requirements to additional countries and placing greater responsibility on data centre operators to verify their customers.
The objective is to prevent Chinese AI developers from accessing advanced American computing technology through overseas infrastructure when direct exports would be restricted.
The regulation is still undergoing interagency consultation and has not yet been formally published. Its final provisions could therefore change before implementation.
Seven Countries Could Face Expanded AI Chip Licensing Requirements
One of the most significant aspects of the proposed regulation is the expansion of existing semiconductor-related controls to seven additional countries.
The countries identified in The Information’s report are:
- Singapore
- Malaysia
- Thailand
- Indonesia
- Brazil
- Norway
- South Africa
These countries have growing data centre industries and are important locations for international cloud computing infrastructure.
Singapore and Malaysia, in particular, have become important destinations for data centre investment in Southeast Asia.
Under the proposed framework, companies seeking to export covered advanced AI chips to data centres in these countries could face additional licensing requirements.
The new requirements would also extend to customers purchasing the hardware, who may need to disclose information about the intended end users and how the processors will be used.
The measures are intended to make it more difficult for restricted customers to obtain computing capacity indirectly through overseas operators.
Nvidia Blackwell Chips and Next-Generation Processors in Focus
The reported restrictions would cover Nvidia’s Blackwell architecture and more advanced AI processors.
Blackwell is Nvidia’s advanced GPU platform designed for demanding artificial intelligence workloads, including model training and inference.
These processors are particularly important for companies developing large language models, generative AI applications and advanced AI agents.
Under the proposed controls, overseas data centre operators seeking to obtain covered processors could face additional scrutiny regarding their customers and intended use.
The licensing framework would require relevant customers to disclose:
- The identity of the end users receiving access to computing resources.
- The intended purpose of the advanced AI processors.
- The location and use of the computing infrastructure.
- Information required for ongoing compliance monitoring.
The requirements could make it harder for Chinese AI companies to secure overseas computing capacity through third-party cloud providers.
For Nvidia, the development could create additional compliance requirements for customers and data centre operators purchasing its advanced processors.
How Chinese AI Companies Have Used Overseas Data Centres
US export restrictions have limited China’s ability to directly purchase certain advanced American AI processors.
However, access to computing power is not necessarily dependent on owning physical hardware.
Cloud computing allows businesses to rent processing capacity from data centres located in other countries.
A company in China can potentially connect remotely to servers hosted in Singapore or Malaysia and use their computing resources for AI development.
This arrangement has become an important issue for US policymakers because advanced AI chips can support the development of increasingly capable artificial intelligence systems.
The proposed rule seeks to address this indirect access by focusing on who can remotely use the processors, rather than only where the physical hardware is shipped.
If implemented, the regulation could require overseas data centre operators to introduce stronger customer verification systems and technical restrictions preventing prohibited users from accessing covered computing resources.
US-Operated Data Centres Could Receive Different Treatment
The reported proposal also includes a distinction between foreign-operated and US-operated data centres.
According to The Information, data centres operated by American companies in the seven additional countries could receive an exemption from individual export licensing requirements, provided they meet specific conditions.
These operators would need to prevent remote access to covered AI chips from countries subject to US arms embargoes, including China, Russia and Venezuela.
However, the exemption would not mean that American operators could avoid all regulatory obligations.
They would reportedly face additional reporting requirements, including disclosing:
- Whether they have customers from arms-embargoed countries.
- The identities of those customers.
- The amount of computing capacity provided to those customers.
The requirements would allow US authorities to monitor how advanced AI computing resources are being used, even when the physical infrastructure is located outside the United States.
The distinction could become important for American cloud providers operating large international data centre networks.
Commerce Department Also Plans to Rescind Biden-Era AI Diffusion Rule
The proposed remote access regulation is also connected to a broader effort to reshape US artificial intelligence export policy.
The Commerce Department is reportedly planning to formally rescind the Biden-era AI Diffusion Rule, which established a framework for controlling the international distribution of advanced AI technology and computing resources.
The Trump administration announced in May 2025 that it intended to withdraw the framework and replace it with a different approach.
However, the formal rescission had not been completed at the time of the November 2026 report.
The new remote access regulation would represent a more targeted approach focused on preventing restricted countries from accessing advanced processors through overseas data centres.
The distinction is important because the proposed rule would concentrate on specific technology, customer access and international computing infrastructure rather than applying a broad global framework in the same way as the earlier policy.
The final regulatory structure will become clearer once the Commerce Department publishes its official documents.
Nvidia and Global Cloud Providers Could Face New Compliance Challenges
The proposed restrictions could affect Nvidia and other companies involved in supplying advanced AI computing infrastructure.
Nvidia sells processors to data centre operators, cloud computing companies and technology businesses around the world.
If additional countries are brought under stricter licensing requirements, customers purchasing covered processors could face longer approval processes and additional disclosure obligations.
International cloud providers may also need to strengthen their customer verification systems to demonstrate that restricted entities cannot access controlled computing resources.
For data centre operators, this could involve:
- Introducing stronger customer identity verification.
- Monitoring the geographical origin of remote computing requests.
- Restricting access based on regulatory requirements.
- Maintaining detailed records of computing usage.
- Reporting relevant customer and capacity information to US authorities.
These measures could increase compliance costs and create additional operational complexity for international cloud businesses.
The impact will depend on the final scope of the regulation, including the technical thresholds, exemptions and enforcement requirements.
Why the Proposed Rule Matters for the Global AI Industry
Access to advanced computing infrastructure has become a major factor in the global AI competition.
Developing sophisticated AI models requires large quantities of processing power, specialised chips and high-performance data centres.
The United States has increasingly used semiconductor export controls to limit China’s access to certain advanced technologies.
However, the growth of international cloud infrastructure has created new challenges for regulators because computing resources can be accessed remotely across national borders.
The proposed rule signals that US authorities want to extend export-control enforcement beyond physical chip shipments and examine how advanced processors are used after reaching overseas destinations.
For countries such as Singapore and Malaysia, the regulation could also create questions about how international data centre investment, cloud services and technology exports are affected by US national security policies.
Governments and technology companies in the affected markets may seek clarity on the licensing requirements before the rules are implemented.
The Bigger Picture
The Commerce Department’s proposed AI remote access rule highlights how the global competition over artificial intelligence is increasingly centred on access to computing infrastructure. Restricting direct chip exports may not be sufficient if companies can rent equivalent processing capacity through overseas data centres. By expanding licensing requirements and introducing customer monitoring, the US government is attempting to close this indirect access route.
For Nvidia and international cloud providers, the proposal could create additional regulatory obligations across several important data centre markets. For Chinese AI companies, it could make access to advanced American processors through overseas infrastructure more difficult. The final impact will depend on the official rule, its technical definitions and the exemptions available to international operators.
Looking Ahead
The Commerce Department is expected to complete consultations with other US government agencies before formally announcing the regulation. If the reported proposal moves forward, data centre operators in the seven additional countries could face new licensing, customer disclosure and monitoring requirements. The final announcement will determine the precise scope of the restrictions and how quickly companies must comply.
The development could also influence future investment decisions across the global data centre industry. Cloud providers may need to reassess customer access policies, while chip suppliers could face additional scrutiny over where their advanced processors are deployed and who ultimately uses them. As governments increasingly treat AI computing capacity as a strategic resource, the regulation could become an important development in the international semiconductor and artificial intelligence industries.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
