More than $3 billion in Chinese artificial intelligence infrastructure projects could face uncertainty as Washington tightens restrictions on advanced semiconductor exports and Nvidia confronts renewed scrutiny following the resignation of co-founder and CEO Jensen Huang. Although Nvidia says US export controls have effectively shut it out of China’s AI data-centre market, procurement records and investment plans reveal that demand for its restricted processors remains deeply embedded in the country’s technology expansion.
A review of more than 150 Chinese procurement notices, investment plans, construction contracts and corporate disclosures identified projects and equipment agreements worth more than 23 billion yuan, or approximately $3.2 billion. The findings highlight the continuing importance of Nvidia hardware to China’s AI ambitions, including projects backed by local governments, universities and regional computing operators.
However, the reported figure does not represent spending on Nvidia chips alone. It includes planned investments, signed contracts and completed installations, as well as buildings, power systems and other supporting infrastructure. The scale of the projects therefore illustrates the broader financial exposure of China’s AI infrastructure rather than the value of hardware already delivered.
The developments also raise questions about the effectiveness of US export controls and whether Washington can prevent restricted computing technology from reaching Chinese users through increasingly complex commercial arrangements.
Stay ahead of the geopolitical week.
MD Briefing delivers expert analysis across five global fronts — the Indo-Pacific, energy, geoeconomics, European security, and the Middle East — every Monday morning. Free.
China’s AI Infrastructure Still Depends on Nvidia Technology
The projects identified in the review extend from China’s wealthy coastal technology centres to less-developed western provinces. Developers and institutions are advancing large computing facilities designed around Nvidia’s restricted H100, H200 and H800 processors.
Some facilities are already operational or under construction, while others remain at the planning or procurement stage. One project in northwestern Gansu province, for example, was expected to install thousands of Nvidia processors before the end of the year. It remains unclear whether the equipment was ultimately delivered.
This distinction matters because announced investments and procurement agreements do not necessarily translate into completed installations. Nevertheless, the documents indicate that Nvidia’s technology remains central to a substantial number of Chinese computing plans, even as direct access to advanced American processors becomes increasingly difficult.
The exposure is not limited to China’s largest technology companies. Local government-backed ventures, universities and regional data-centre operators have also incorporated Nvidia processors into their infrastructure plans.
For these organisations, restrictions could create significant operational challenges. Delayed deliveries, disrupted supply arrangements or an inability to obtain replacement hardware could undermine project timelines. Some operators may have to redesign facilities around domestic processors, potentially increasing costs and requiring changes to existing computing systems.
The consequences could extend beyond individual companies. Regional computing centres are part of China’s broader effort to expand domestic AI capacity, making access to advanced processors an important factor in the country’s technological development.
Jensen Huang’s Resignation Adds to Market Uncertainty
The timing of Huang’s resignation has added another layer of uncertainty to an already sensitive technology dispute.
Huang had been one of the most prominent advocates of easing US restrictions on semiconductor exports to China. He previously said Nvidia’s market share in China had fallen from 95% to zero as a result of export controls. The company has also excluded Chinese data-centre computing revenue from its forecasts, citing regulatory uncertainty.
His departure could weaken Nvidia’s influence in Washington at a time when lawmakers are pushing for tighter restrictions on advanced AI chips. The company’s ability to defend its commercial interests in China may become more difficult if political scrutiny intensifies.
The resignation followed allegations involving Alice Huang, a Chinese businesswoman and former executive associated with Singapore-based Nvidia cloud partner Megaspeed, who is suspected by US authorities of having intelligence ties to Beijing.
Megaspeed has denied illegally diverting Nvidia processors to China. Nvidia’s board has launched an independent investigation, stating that it has no reason to believe company secrets or US national security were compromised.
There has been no public finding that Jensen Huang violated export-control laws. The allegations and the company’s investigation should therefore be distinguished from any established finding of wrongdoing.
Nevertheless, the episode places Nvidia under greater political and reputational pressure. For Chinese developers whose projects depend on its processors, uncertainty surrounding the company’s leadership and its relationship with US policymakers could complicate future procurement decisions.
Washington Faces a Difficult Export-Control Challenge
The United States began restricting Nvidia’s advanced AI processors in 2022, subsequently expanding controls to cover modified chips designed to meet earlier restrictions. The H800 was among the processors affected by these measures.
In January, the Trump administration moved towards reviewing export licences for H200 processors on an individual basis for eligible Chinese customers, subject to conditions. This approach reflected the continuing tension between Washington’s national security objectives and the commercial interests of American semiconductor companies.
Nvidia’s August filing stated that the company was effectively excluded from China’s data-centre computing market. However, an October report by research firm IDC estimated that Nvidia still accounted for 34% of Chinese AI-accelerator shipments during the first half of 2026.
These figures do not necessarily contradict one another. Nvidia’s statement concerns its ability to compete and generate revenue directly in China’s data-centre market, while IDC’s estimate measures the company’s share of AI-accelerator shipments. Neither figure establishes how the restricted processors identified in the procurement documents reached Chinese users.
The distinction highlights a central weakness in the export-control debate: restricting direct sales does not automatically eliminate demand or establish where every processor ultimately ends up.
Analyst Sunny Cheung of the Washington-based Jamestown Foundation argues that US enforcement needs to focus not only on the immediate buyers of advanced chips but also on their ultimate users and the purposes for which the computing capacity is deployed.
This raises a broader challenge for American policymakers. Monitoring individual transactions may be insufficient if advanced computing infrastructure involves multiple companies, intermediaries, cloud providers and regional operators.
Chinese Developers Face Costly Alternatives
For Chinese institutions and businesses, the uncertainty creates a difficult investment environment. Projects designed around Nvidia processors may become more expensive or technically complicated if the hardware cannot be obtained.
Domestic semiconductor manufacturers offer an alternative, and Beijing has a strong incentive to accelerate the development and adoption of Chinese AI processors. However, replacing established hardware can require changes to software, infrastructure and operating systems.
The challenge is particularly significant for projects already under construction or based on signed procurement agreements. Developers may have to reconsider equipment specifications, delay installations or absorb additional costs if their original plans become unworkable.
The uncertainty also affects investment decisions. Companies planning large data-centre projects need confidence that the processors they intend to use will remain available throughout construction and operation. Continued changes to US licensing rules could make long-term planning more difficult.
At the same time, the scale of the identified projects demonstrates that China’s demand for advanced computing has not disappeared. Restrictions may limit access to particular American technologies, but they have not removed the economic incentives driving the country’s AI expansion.
What Happens Next?
The future of China’s Nvidia-powered projects will depend on several developments: Washington’s export-control decisions, the outcome of Nvidia’s internal investigation and China’s ability to expand domestic alternatives.
US authorities face the challenge of enforcing restrictions without creating loopholes that allow advanced processors to reach prohibited users through indirect arrangements. Greater scrutiny of cloud computing services, procurement networks and the ultimate use of computing capacity could become increasingly important.
For Nvidia, Huang’s departure comes at a sensitive moment. The company must navigate a more politically difficult environment while protecting its commercial interests and addressing questions surrounding its business relationships.
China, meanwhile, will have to manage the risks facing projects that were designed around access to American technology. A faster transition towards domestic processors could reduce long-term dependence, but replacing existing infrastructure will not be immediate or cost-free.
Ultimately, the more than $3 billion in identified projects illustrates the scale of the challenge facing Washington. Export controls may restrict direct access to advanced Nvidia processors, but China’s AI infrastructure plans reveal how deeply American semiconductor technology remains embedded in its computing ambitions. The central question is no longer simply whether Nvidia can sell chips to China, but whether Washington can effectively control how advanced computing capacity is acquired, distributed and used.
With information from Reuters.
