Chinese AI Firms Rush to Secure Overseas Nvidia Chips

Chinese Artificial Intelligence companies are scrambling to secure access to Nvidia-powered computing capacity at overseas data centres amid growing concerns that a controversy involving Nvidia chief executive Jensen Huang could push the United States to tighten restrictions on China’s access to advanced AI chips.

Several Chinese technology companies are reportedly exploring ways to reserve computing capacity outside China, allowing them to use powerful Nvidia processors remotely without physically importing the restricted hardware. The move reflects growing uncertainty over whether Washington will close a gap in its export-control framework that currently allows some Chinese developers to access advanced computing resources through overseas cloud infrastructure.

The concern has intensified following allegations involving Huang and Alice Huang, a former executive at Singapore-based cloud-computing company Megaspeed. The controversy has drawn renewed attention to the possibility that overseas data centres and cloud services could provide Chinese companies with access to advanced American technology despite restrictions on direct chip exports.

According to people familiar with the discussions, a major Chinese cloud provider has internally considered securing overseas Nvidia computing capacity. Other companies have approached data-centre operators about reserving capacity, while smaller AI laboratories are exploring similar arrangements with computing providers in Southeast Asia.

Chinese AI Companies Seek to Reserve Overseas Nvidia Computing Capacity

The rush to secure overseas computing resources highlights the importance of Nvidia processors to China’s artificial intelligence industry. Advanced graphics processing units, or GPUs, are widely used to train and operate large AI models that power chatbots, generative AI applications and other intelligent software.

Although US export controls restrict Chinese companies from directly obtaining certain advanced Nvidia processors, access to computing power can also be arranged through data centres located in other countries. Under these arrangements, customers rent processing capacity remotely rather than taking physical possession of the chips.

This distinction has become increasingly important in the US-China technology rivalry. Direct exports of restricted processors can be subject to licensing requirements, but remote access through overseas cloud infrastructure has raised questions about how effectively existing rules cover the use of controlled technology.

Chinese technology companies are concerned that this access could become more difficult if Washington introduces stricter regulations. Reserving overseas capacity in advance could help them maintain access to computing resources while uncertainty over future US policy grows.

People familiar with the discussions said a major Chinese cloud provider had considered locking in overseas Nvidia chip capacity. Other firms had contacted data-centre operators to explore reservations, while smaller AI laboratories were considering approaching major providers of rented computing power in Southeast Asia.

The companies involved have not been publicly identified, and the reported discussions do not establish that any particular provider has completed a new reservation agreement.

Remote Access Security Act Could Close a Gap in US Export Controls

One of the biggest concerns among Chinese AI companies is the possible expansion of US oversight through the Remote Access Security Act, or RASA. The proposed legislation would give the US Commerce Department authority to regulate remote, cloud-based access to restricted technology.

The US House of Representatives passed the bill in January by a vote of 369 to 22. It is awaiting action in the Senate.

The legislation is significant because it addresses a different route to advanced computing power. Existing export restrictions focus heavily on the movement and sale of controlled hardware. However, cloud computing allows users to access processors located overseas without importing the physical devices into their own country.

If the legislation becomes law, the US government could gain additional authority to oversee or restrict certain remote access arrangements involving controlled technology. That could make it more difficult for Chinese developers to use high-performance Nvidia processors hosted in foreign data centres.

For AI companies, the consequences could extend beyond the loss of a particular chip supply. Training large models requires substantial computing resources, while operating popular AI services also demands access to powerful processors. Restrictions on overseas cloud access could therefore affect both the development and deployment of advanced AI systems.

The bill’s future remains uncertain because it still requires Senate action. Until then, Chinese companies face the possibility of tighter controls without knowing precisely how future regulations might affect their overseas computing arrangements.

Jensen Huang Controversy Adds Political Pressure to the Chip Debate

The growing concern over remote access has coincided with a controversy involving Nvidia chief executive Jensen Huang and Alice Huang, a former executive at Megaspeed.

Huang said on Sunday that he had been unfaithful to his wife. The New York Times subsequently reported that the woman involved was Alice Huang and that US authorities had concluded she was a Chinese intelligence asset.

Reuters said it could not independently verify the allegation concerning Alice Huang’s alleged intelligence links. China has denied the accusation, and it should not be treated as an established finding.

The controversy has nevertheless increased political scrutiny of Nvidia’s business relationships and the wider question of whether advanced American AI processors can reach Chinese customers through indirect channels.

The issue is particularly sensitive because Nvidia occupies a central position in the global AI computing industry. Its processors are used by technology companies, research institutions and cloud providers to develop and operate advanced AI systems.

Any perception that restricted chips could be accessed by Chinese organisations through overseas infrastructure may strengthen arguments among US lawmakers for broader controls. The controversy involving Huang has therefore become part of a larger debate about technology security, export enforcement and America’s ability to maintain an advantage in artificial intelligence.

Alice Huang’s Business Connections With Chinese Technology Companies

Alice Huang’s reported business relationships have added another dimension to the controversy. According to people familiar with the matter, she had business dealings and investment connections with several prominent Chinese technology companies, including Tencent and Alibaba.

Neither company immediately responded to requests for comment about the reported connections.

Corporate records reviewed by Reuters also showed that Alibaba subsidiary Apex Enterprise Solutions provided financing to Speedmatrix, a Malaysian company linked to Megaspeed. The records offer information about a business relationship involving an Alibaba subsidiary and a company associated with the Singapore-based cloud provider.

However, the existence of a financing relationship does not independently establish that Alibaba or its subsidiary knowingly participated in any illegal transfer of restricted Nvidia technology.

The distinction matters because the controversy involves several separate questions: Alice Huang’s reported business connections, allegations about possible intelligence links, and suspicions concerning access to advanced Nvidia processors. These issues require separate evidence and should not automatically be treated as proof of wrongdoing by the companies mentioned.

Megaspeed Faces Scrutiny Over Access to Restricted Nvidia Chips

Megaspeed, the Singapore-based cloud-computing company associated with Alice Huang, has faced scrutiny over suspicions that it helped Chinese customers obtain access to advanced Nvidia chips covered by US export controls.

The company has denied illegally diverting chips to China. The allegations have not been established as proven violations through the information available in the reported investigation.

The scrutiny reflects a broader challenge for US export-control authorities. Advanced processors can be installed in data centres outside China and made available to customers through cloud services. Determining who ultimately uses that computing power can be more complicated than tracking a direct shipment of physical chips.

Cloud infrastructure also operates across international markets, with data-centre operators providing computing resources to customers in different countries. This creates questions about customer identification, end-user verification and the monitoring of remote access to restricted hardware.

The US government has increasingly focused on preventing controlled chips and AI servers from reaching prohibited users through intermediaries. The Megaspeed controversy has brought renewed attention to the role that overseas cloud providers could play in this process.

Nvidia, Megaspeed and the US Commerce Department did not immediately respond to requests for comment on the reported developments.

Grey-Market Demand for Restricted Nvidia AI Chips Rises in China

Concerns about tougher US enforcement are also affecting China’s unofficial market for advanced AI processors. Traders and intermediaries have reportedly started stockpiling restricted Nvidia chips in anticipation of a possible crackdown on smuggling and indirect supply channels.

People familiar with the market activity said buyers were concerned that the controversy could increase political pressure on Washington to investigate the movement of advanced processors and AI servers into China.

The grey market refers to transactions taking place outside authorised distribution channels. In the semiconductor industry, such activity can involve intermediaries, indirect shipments and attempts to obtain hardware that is difficult or impossible to purchase through normal commercial routes.

For Chinese AI developers facing restrictions on certain high-performance processors, unofficial supply channels can appear attractive when domestic alternatives do not offer equivalent capabilities. However, acquiring restricted hardware through unauthorised routes can create significant legal and commercial risks.

Stockpiling also reflects uncertainty about future availability. If US authorities introduce stricter enforcement measures, buyers may find it harder to obtain certain processors or servers through existing intermediaries.

The reported increase in grey-market activity does not mean every transaction involving Nvidia hardware in China is illegal. The legal position depends on the specific processor, licensing conditions, parties involved and the route through which the equipment is obtained.

Chinese University With Military Links Bought Nvidia H200 Chips

Questions about the enforcement of US export controls have also emerged following reports that a Chinese university with close ties to the People’s Liberation Army purchased Nvidia H200 AI chips despite being placed on a US export blacklist.

Procurement records showed that the institution bought the powerful processors in October. The reported purchase has raised questions about how restricted technology can reach organisations that Washington has identified as posing national security or military-related concerns.

The H200 is an advanced Nvidia processor designed for demanding AI workloads. Such chips can support the training and operation of large models, making them valuable to research institutions and technology companies seeking substantial computing power.

The university purchase is particularly sensitive because US restrictions are designed to limit access to advanced American technology among certain Chinese institutions. A blacklist designation can impose additional licensing requirements and restrictions on exports involving listed entities.

However, the reported procurement does not by itself establish how the chips were obtained, whether a specific export-control violation occurred or which intermediaries may have been involved. Those questions require examination of the transaction and the applicable licensing rules.

The case nevertheless illustrates the difficulty of enforcing technology restrictions across complex international supply chains. It also strengthens calls for closer monitoring of chip sales, overseas data centres and companies providing access to advanced computing resources.

US Prosecutors Target Alleged Nvidia Server Diversion Network

US authorities have also taken action in a separate case involving the alleged diversion of servers containing Nvidia processors to China.

In March, prosecutors charged Super Micro Computer co-founder Yih-Shyan “Wally” Liaw and two other individuals with conspiring to divert servers equipped with Nvidia chips to China through a company based in Southeast Asia.

According to prosecutors, the company involved purchased approximately $2.5 billion worth of servers from Super Micro between 2024 and 2025. A substantial portion of those servers was allegedly diverted to China.

The case highlights the role of server systems in the international AI hardware trade. Advanced processors are commonly installed in specialised servers that can be deployed in data centres and connected to large-scale computing networks.

Restricting individual chip shipments is therefore only one part of export enforcement. Authorities must also monitor the sale, movement and eventual use of complete server systems containing controlled processors.

The defendants have not been convicted, and the allegations remain subject to the legal process. The prosecution demonstrates that US authorities are examining suspected diversion schemes involving both chips and the larger computing systems built around them.

Why Nvidia Chips Remain Important to China’s AI Industry

Nvidia processors are important to AI development because they provide the computing capacity needed to train and operate increasingly sophisticated models. Training involves processing large datasets and adjusting model parameters, while inference refers to the process of using a trained model to generate responses or perform tasks.

Both activities require substantial computing resources, although their hardware requirements can differ. Large-scale model training often depends on powerful processors connected in clusters, while inference workloads can be distributed across different types of computing systems.

Chinese AI companies have been investing in domestic semiconductor alternatives, but access to Nvidia hardware remains valuable because of its performance and established software ecosystem.

US restrictions have encouraged Chinese technology companies to explore different supply arrangements, including domestic processors, export-compliant products and overseas cloud infrastructure.

However, these alternatives do not necessarily provide identical capabilities or availability. Companies developing advanced models must consider processor performance, software compatibility, energy costs and the amount of computing capacity they can secure.

This helps explain why Chinese AI firms are reportedly trying to reserve overseas Nvidia capacity before further restrictions potentially change the market.

What the Developing Scandal Means for the US-China AI Chip Race

The developments highlight how the competition between the United States and China increasingly extends beyond manufacturing and direct semiconductor exports. Access to computing infrastructure, cloud services and international data centres is becoming an equally important part of the technology rivalry.

Washington has sought to limit China’s access to advanced processors on national security grounds, while Chinese technology companies continue looking for ways to obtain the computing power required to compete in the global AI market.

The Remote Access Security Act could become an important part of this debate if the Senate advances the legislation. Giving the Commerce Department additional authority over remote access could make overseas cloud computing a more closely regulated area of US technology policy.

For Chinese AI companies, the immediate challenge is uncertainty. Reserving computing capacity abroad may help them maintain access in the short term, but it cannot guarantee that future regulatory changes will allow those arrangements to continue.

For Nvidia, the controversy creates additional scrutiny around the international use of its processors and the responsibilities of its commercial partners. The company operates in a market where commercial opportunities, national security concerns and government export policies increasingly overlap.

The allegations involving Alice Huang remain unverified by Reuters, while Megaspeed has denied illegally diverting chips to China and the defendants in the Super Micro-related case have not been convicted. Nevertheless, the combination of political controversy, suspected chip diversion and growing demand for overseas computing capacity has intensified attention on how advanced AI technology moves across borders.

As the US-China AI competition develops, the central question will be whether Washington can extend export controls beyond physical shipments to cover the remote use of restricted processors. The outcome could reshape the international cloud-computing market and determine how easily Chinese AI developers can access some of the world’s most powerful commercial AI hardware.