How China Gets the Nvidia Chips It Needs to Develop Artificial Intelligence Despite US Export Restrictions
China continues to obtain advanced Nvidia artificial intelligence chips despite strict US export restrictions designed to limit Beijing’s access to high-performance computing technology. Through an expanding network of international intermediaries, overseas data centres, chip brokers and alleged smuggling operations, Chinese technology companies are finding alternative ways to secure the computing power needed to develop increasingly sophisticated AI models.
The issue has become a major concern for Washington as China accelerates its competition with the United States in artificial intelligence. Nvidia’s advanced graphics processing units (GPUs) are particularly valuable because they can perform the enormous calculations required to train and operate powerful AI systems.
Why Nvidia Chips Are So Important
Nvidia has established itself as one of the world’s most important suppliers of AI computing hardware. Its GPUs, including the H100 and newer Blackwell-generation processors, are widely used by leading technology companies to train large language models, develop generative AI applications and operate advanced computing infrastructure.
China’s AI developers, including DeepSeek, Moonshot AI and Z.ai, require substantial computing resources to compete with American technology companies. Although China has made progress in developing domestic semiconductor alternatives, Nvidia’s hardware remains attractive because of its performance, software ecosystem and established developer infrastructure.
Smuggling Through Third Countries
One of the principal ways restricted Nvidia chips reportedly reach Chinese customers is through international smuggling networks operating across Asia.
Investigative reporting has identified brokers in countries such as Malaysia and other Southeast Asian markets who arrange purchases through intermediary companies. These operations can involve shell companies, false corporate identities and complicated supply chains designed to conceal the ultimate Chinese buyers.
In one reported transaction, a Chinese electronics company allegedly arranged a $120 million purchase involving 300 servers equipped with Nvidia H100 processors through a Malaysian intermediary. The servers were initially presented as purchases by an overseas company before being transferred towards China.
Some brokers reportedly use temporary warehouse facilities, misleading documentation and repackaging arrangements to make restricted hardware harder to identify during supply-chain inspections.
The growing sophistication of these operations has transformed chip smuggling from isolated transactions into a potentially lucrative international business.
Using Overseas Data Centres to Access Nvidia GPUs
Another important route involves accessing Nvidia-powered computing infrastructure located outside mainland China.
Instead of physically importing restricted processors, Chinese companies can rent computing capacity from overseas data centres. These facilities may contain thousands of advanced Nvidia GPUs installed in countries such as Malaysia and Thailand.
Chinese AI developers can then connect remotely to the servers and use their computing resources to train and operate AI models.
This arrangement creates a significant enforcement challenge because the physical chips remain outside China, while the computing services can be accessed by customers located inside the country.
However, Washington has increasingly focused on this potential loophole and is considering or implementing tighter restrictions on remote access to advanced computing resources.
The Role of International Chip Brokers
A growing network of independent semiconductor brokers and intermediary businesses has also emerged around the demand for restricted Nvidia hardware.
These intermediaries reportedly connect Chinese buyers with overseas suppliers, arrange server purchases, establish overseas corporate entities and coordinate transportation.
Some transactions are believed to involve companies with connections to larger technology businesses and government-backed organisations.
The high demand for advanced AI processors has created substantial financial incentives for these networks. Restricted Nvidia chips can command significant premiums on unofficial markets because Chinese buyers have fewer legitimate purchasing options.
What Washington Is Doing to Stop the Supply
The United States has introduced multiple rounds of semiconductor export restrictions to prevent China from obtaining the most advanced American AI processors.
The restrictions cover powerful Nvidia products, including its Blackwell and newer Rubin generations. Washington has also tightened scrutiny of intermediary companies, overseas data centres and suspicious semiconductor transactions.
In 2026, the US eased certain restrictions on Nvidia’s H200 processors, subject to government approval and additional conditions. However, the arrangement has faced obstacles, including regulatory requirements and reluctance among some Chinese buyers to depend on American technology.
US authorities are also examining whether export controls should cover remote access to computing infrastructure located in third countries, rather than focusing exclusively on the physical movement of chips.
China Is Also Building Its Own AI Chip Industry
Despite its continued interest in Nvidia processors, China is investing heavily in domestic semiconductor development to reduce its dependence on American technology.
Huawei has emerged as one of the country’s leading competitors in the AI chip market. Chinese manufacturers are working to improve processor performance, manufacturing capabilities and software compatibility.
Companies such as DeepSeek are also exploring ways to develop competitive AI models with more efficient computing requirements.
However, domestic alternatives still face challenges involving manufacturing technology, processing efficiency, software ecosystems and the engineering resources needed to operate large computing clusters.
China’s relatively inexpensive electricity and government-backed investment programmes could help offset some of these disadvantages by allowing companies to deploy larger numbers of domestically developed processors.
A Growing Technology Battle Between Washington and Beijing
The continuing movement of Nvidia chips into China highlights the difficulties of enforcing technology restrictions in a highly interconnected global economy.
While Washington aims to slow China’s access to advanced AI computing, Beijing is pursuing several parallel strategies: securing hardware through international supply chains, accessing overseas computing facilities and strengthening domestic semiconductor production.
The competition is no longer limited to who manufactures the most advanced chips. It increasingly involves control over data centres, semiconductor supply chains, computing infrastructure, software ecosystems and the global rules governing access to artificial intelligence technology.
For the United States, preventing restricted chips from reaching Chinese AI developers remains a significant national security challenge. For China, securing enough computing power to compete with American AI companies has become a strategic technological priority.
