Nvidia has reportedly told its data center and cloud computing partners to reduce their dependence on Chinese customers. The move could make it harder for Chinese technology companies to access advanced artificial intelligence (AI) computing power through overseas services as competition between the United States and China continues to grow.
The development was detailed in a November 3, 2026, Financial Times report citing five people familiar with the matter. According to the report, Nvidia has been giving cloud computing partners new instructions as competition between the United States and China over advanced AI technology continues. The instructions represent a reported business requirement rather than a newly announced U.S. government regulation.
Under the reported instructions, Nvidia has told cloud computing partners to sell at least 51% of their services to non-Chinese companies. This would mean that businesses classified as non-Chinese must account for a majority of the services sold under the arrangement. However, the report does not clearly establish whether the percentage would be measured by revenue, computing capacity or another method. The instructions were reportedly delivered verbally through Nvidia’s sales representatives and regional teams rather than through a publicly available policy document.
The reported instructions would apply to future cloud computing contracts rather than existing leasing agreements. This means companies with current agreements are not expected to lose access automatically because of the reported change. The instructions also should not be confused with a complete ban on Chinese businesses using overseas AI cloud services. However, companies negotiating new contracts could face additional conditions depending on how Nvidia and its partners apply the reported requirements.
Nvidia is one of the world’s leading suppliers of graphics processing units (GPUs) used in artificial intelligence. These powerful processors help companies train large AI models and operate AI applications such as chatbots. Training and running AI models require different types of computing resources, as explained in our guide to AI training and inference data centers. Demand for advanced GPUs has increased as businesses continue investing in AI infrastructure.
Chinese technology companies face U.S. export restrictions on several advanced Nvidia AI processors, although certain chips can still be considered for export licenses under specific conditions. These restrictions have changed over time, including adjustments to the U.S. licensing rules for Nvidia H200 chip exports to China. Some Chinese businesses have also turned to overseas cloud computing providers, which allow customers to rent access to powerful processors without directly purchasing or importing the physical chips.
The reported Nvidia instructions could make this arrangement more difficult for some businesses. Cloud providers that depend heavily on Chinese companies may need to attract more customers from non-Chinese businesses to meet the reported 51% requirement. This could affect how providers negotiate new contracts, plan computing capacity and select customers. However, the actual impact will depend on how Nvidia defines and applies the requirement.
The potential impact could be particularly significant for data center operators in Southeast Asia, where Chinese technology companies represent an important source of demand for advanced computing services. Finding enough non-Chinese customers may take time, especially in markets where demand for expensive AI infrastructure is still developing.
The reported changes could also affect companies planning new AI data center investments. Building these facilities requires substantial spending on advanced processors, electricity, cooling systems and networking equipment. The scale of such investment is highlighted by Oracle’s $20 billion financing plan for AI and cloud expansion. Cloud providers that cannot attract enough eligible customers under the reported Nvidia requirements may need to reconsider expansion plans or adjust how they sell computing services.
The development comes as the United States continues to regulate China’s access to advanced semiconductor technology. The country’s export licensing rules for advanced computing chips place restrictions on certain high-performance processors and related technologies. U.S. officials have raised national security concerns about advanced AI hardware potentially supporting military, intelligence and other sensitive applications. However, these government export controls are separate from the newly reported Nvidia instructions concerning cloud computing partners.
Nvidia has also faced increasing pressure to monitor where its processors are sold and how they are used. In its August 2026 SEC filing, the company discussed the impact of U.S. export controls on its data center business in China and described its relationships with AI cloud partners. These official disclosures provide important background, although they do not establish the newly reported 51% requirement as a publicly documented Nvidia policy.
Several important questions remain unanswered. Nvidia has not publicly explained which cloud partners would be covered by the reported 51% requirement, how the percentage would be calculated or what would happen if a provider failed to meet it. It is also unclear whether the instructions are being enforced as a strict contractual condition or treated as business guidance. These details will be important in determining how significantly the reported changes affect the global AI cloud industry.
For Chinese AI developers, tighter access to overseas computing services could increase costs and make it more difficult to obtain the processing power required for large AI projects. It could also encourage greater investment in locally developed chips, domestic cloud infrastructure and alternative computing technologies.
For international cloud providers, the reported changes highlight how geopolitical tensions are influencing the global AI industry. Companies must now consider not only the cost and performance of advanced processors but also customer eligibility, regulatory requirements and long-term access to technology.
If Nvidia applies the reported requirements broadly, the decision could reshape how cloud providers distribute AI computing services and how Chinese technology companies secure the resources needed for future development. Further clarification will be important in determining whether the instructions become a lasting restriction on global AI infrastructure.
