Chinese AI firms seek overseas Nvidia power amid new US remote access bill

The race for overseas capacity

Chinese artificial intelligence companies are currently moving to secure computing power from Nvidia outside of Chinese borders. This effort is driven by the fear that the United States will soon implement stricter rules on how foreign entities access American technology. Companies are looking for ways to maintain their development pace while navigating an increasingly complex regulatory environment.

A major Chinese cloud provider has reportedly discussed securing long-term overseas capacity to ensure they can continue training models. Other firms are reaching out to data center operators in various regions, including Southeast Asia, to reserve the necessary hardware. These arrangements allow developers to use high-end processors without having to physically import the restricted chips into China.

The activity reflects a strategic shift for developers who depend on advanced processors for large-scale AI tasks. By renting capacity in a third country, these firms can operate their systems while technically remaining outside the immediate scope of direct export bans. However, the distinction between owning a chip and remotely using one is becoming a primary focus for U.S. regulators.

The Remote Access Security Act

The primary driver of this concern is the Remote Access Security Act, which recently passed the House of Representatives. The bill received a 369-22 vote and was subsequently referred to the Senate Banking Committee. It aims to give the Commerce Department more power to regulate access to U.S.-controlled technology through network connections.

Under this proposed legislation, federal regulators could extend existing export controls to situations where a foreign customer accesses technology located elsewhere. This would mean that renting cloud computing services from a foreign operator could become a restricted activity. For Chinese firms, this could potentially shut down the current method they use to bypass hardware import limits.

While the bill has not yet become law, its passage has created a sense of urgency among international technology players. The legislation seeks to address the possibility of remote access being used as a workaround for trade restrictions. If enacted, it would represent a significant expansion of how the U.S. manages its high-end computing exports.

Market reactions and stockpiling

The uncertainty surrounding these potential new rules has already begun to impact the financial markets. Alibaba shares saw a decline of as much as 7.1% in Hong Kong trading as investors weighed the possibility of tighter restrictions. Similarly, AI model developers Z.AI and MiniMax both saw their shares drop by at least 7%.

Beyond the stock market, the pressure is manifesting in the physical trade of hardware. Traders and intermediaries in China have started stockpiling Nvidia chips that are available through unofficial channels. These buyers are worried that a broader U.S. enforcement campaign will make it much harder to obtain advanced processors in the future.

This stockpiling activity is separate from the discussions regarding overseas cloud capacity but stems from the same underlying anxiety. Both behaviors indicate that Chinese firms are preparing for a future where access to Nvidia technology is significantly more difficult. The combination of physical hoarding and cloud reservation shows a multi-pronged approach to securing resources.

Alternative hardware options

Despite the focus on high-end data center chips, Chinese companies are also looking at less restricted hardware options. Demand for Nvidia workstation processors remains strong because they are less powerful than the most advanced chips but are still highly capable. These processors provide a viable path for companies to expand their AI capabilities under current trade rules.

In October alone, companies including ByteDance, Alibaba, and Tencent placed orders for these workstation processors worth more than $35 billion. This demonstrates that while high-end chips are the goal, there is a massive, ongoing appetite for any available American hardware. These orders help maintain the momentum of AI development even as the most powerful tools are restricted.

The situation creates a complex landscape for both the U.S. government and Chinese technology firms. Washington is attempting to limit access to high-end computing, while Chinese companies are exploring every possible avenue to maintain their progress. This includes seeking overseas cloud services, purchasing lower-tier hardware, and stockpiling existing inventory.