Nvidia is preparing to start the upcoming year with a resumption in shipments of advanced AI processors to China, reports Reuters. The company has told its Chinese customers of plans to begin delivering its H200 chips before the Lunar New Year in mid-February, pending regulatory approval in Beijing.
From the looks of it, initial deliveries would be drawn from the company’s existing inventory, with early shipments expected to total between 5,000 and 10,000 chip modules, equivalent to roughly 40,000 to 80,000 individual H200 processors. Expansion of production capacity for the H200 is reportedly underway as well, with orders for new capacity expected to open in the second quarter of the coming year.
The plan remains uncertain because Chinese authorities have not yet approved any purchases of the H200. “The whole plan is contingent on government approval,” one person familiar with the discussions said to Reuters, adding that timelines could shift depending on policy decisions in Beijing. Nvidia and China’s Ministry of Industry and Information Technology did not respond to requests for comment.
If the shipments proceed, they would mark the first deliveries of H200 chips to China since US President Donald Trump said earlier this month that Washington would allow such sales, subject to a 25% surcharge. The Trump administration has launched an inter-agency review of export license applications for the H200, reversing the previous stance taken under President Joe Biden, whose administration barred exports of advanced AI chips to China on national security grounds.
The decision marks a sharp change in Washington’s approach to tech controls, reopening a channel that had been largely closed to one of Nvidia’s most lucrative markets. China accounted for a meaningful share of Nvidia’s data-center revenue before restrictions intensified, although the company does not disclose country-level figures.
For those who need a recap, the H200 belongs to Nvidia’s Hopper generation and, while no longer the company’s most advanced offering, remains a workhorse for training and running large AI models. Nvidia has since moved its production focus to its newer Blackwell architecture and its upcoming Rubin platform, which has made H200 supply comparatively tight. For Chinese customers, the H200 represents a major step up from the H20, a downgraded processor Nvidia designed specifically to comply with earlier US export limits. People familiar with the specifications say the H200 offers performance several times higher than the H20, making it attractive to cloud providers and AI developers.
Major Chinese tech groups, including Alibaba Group and ByteDance, have expressed interest in acquiring the H200 for large-scale model training and inference. At the same time, Beijing has been pushing aggressively to build a domestic AI chip industry, backing local designers and manufacturers to reduce reliance on foreign suppliers.
Chinese officials have held internal discussions in recent weeks on whether to allow H200 imports, according to people familiar with the matter.
One proposal under consideration would require each purchase of Nvidia chips to be bundled with a specified ratio of domestically produced processors, a measure aimed at supporting local suppliers. Local chipmakers have yet to match the performance of Nvidia’s top-tier products, creating a tension between near-term demand for compute power and long-term industrial policy goals.
For Nvidia, renewed access to China could provide incremental revenue from a product line it is no longer prioritizing, while allowing the company to concentrate cutting-edge capacity on newer architectures for customers elsewhere. Using existing H200 stock to fulfill early Chinese orders would also help Nvidia monetize inventory as it transitions to its next generation of chips.
The Tech Portal is published by Blue Box Media Private Limited. Our investors have no influence over our reporting. Read our full Ownership and Funding Disclosure →
