Billions in Chinese AI projects remain tied to restricted Nvidia chips

Hidden infrastructure in China

A review by Reuters identified more than $3.2 billion in Chinese computing projects that are built around restricted Nvidia processors. These projects are spread across the country, reaching from wealthy coastal technology hubs to remote western provinces. The findings suggest that Nvidia technology remains deeply embedded in China's AI expansion despite heavy US restrictions.

The investigation looked at more than 150 different documents, including investment plans, construction contracts, and corporate disclosures. These records revealed agreements for equipment and infrastructure with a combined reported value exceeding 23 billion yuan. This estimate includes planned investments and signed contracts as well as projects that are already completed.

The scope of these projects is broad and goes beyond the country's largest technology firms. The data shows that Nvidia's technology reaches into local government-backed ventures, universities, and regional computing centers. It is important to note that these figures cover buildings and power systems, rather than just the cost of the chips themselves.

The limits of export controls

The United States began restricting Nvidia's advanced AI processors in 2022 and later tightened rules on modified chips like the H800. Nvidia has stated that these controls have effectively shut it out of the Chinese AI data-centre market. Because of this uncertainty, the company has removed China data-centre computing revenue from its financial forecasts.

However, the Reuters review shows that Chinese customers are still planning purchases and, in some cases, operating restricted processors. For example, a facility in northwestern Gansu province was expected to install thousands of Nvidia chips before the end of the year. It remains unclear if those specific processors ever actually arrived at the site.

These findings create a complex picture of how hardware moves across borders. While Nvidia's official statements focus on its inability to sell directly, the documents show a persistent demand for the hardware. This discrepancy highlights the gap between official trade policy and the actual infrastructure being built on the ground.

Leadership shifts and scrutiny

The situation has become more complicated following the resignation of Nvidia co-founder Jensen Huang as chief executive. His departure came amid a scandal involving a businesswoman suspected by US authorities of having intelligence ties to Beijing. While no public finding exists that Huang violated export-control laws, his exit may weaken Nvidia's ability to lobby for easier trade rules.

Nvidia's board launched an independent investigation into the matter but stated they had no reason to believe company secrets were compromised. The company has been a vocal advocate for easing export restrictions, but the current political climate is increasingly hostile. The scrutiny from Washington has intensified as lawmakers press for even tighter controls on the technology powering the AI boom.

The loss of a prominent leader who advocated for open trade could change how the company navigates these regulations. As the US government moves to review export licenses on an individual basis, Nvidia faces a more difficult path. The company must now balance its market position with a rapidly evolving and restrictive regulatory environment.

The path forward for developers

Operators of the identified Chinese projects face significant risks as export controls continue to tighten. They are exposed to potential delays, supply disruptions, or the need for costly redesigns to accommodate domestic chips. This creates a precarious environment for developers who have already committed billions to infrastructure built around American hardware.

Analysts suggest that the focus of enforcement may need to shift from who buys the chips to who ultimately uses the computing power. This perspective emphasizes the difficulty of policing the end-use of technology once it has entered a specific region. The ongoing tension between trade security and technological growth remains a central theme for the industry.

Moving forward, the industry will likely see a push toward domestic alternatives in China as the risk of using Nvidia hardware grows. However, the sheer scale of the existing $3.2 billion in projects suggests a long transition period. The reliance on established American hardware remains a significant hurdle for those trying to decouple the two economies.