Nvidia begins shipping H200 AI chips to China after U.S. approves limited exports

Nvidia has started shipping its H200 artificial intelligence chips to customers in China, according to a report citing confirmation from a senior U.S. trade official. The update comes after U.S. authorities approved limited exports, a shift that resumes part of a channel Nvidia had previously faced restrictions on for advanced AI hardware.

The H200 is Nvidia’s data-center accelerator designed to help train and run large-scale AI models. For Nvidia, shipments to China are strategically important because the country remains a major demand center for computing capacity used in cloud and enterprise AI deployments. After export controls tightened in recent cycles, Nvidia’s ability to supply certain high-end systems into China has been shaped by what the rules allow.

In the report, the timing is tied to U.S. clearance. The article says the approval was confirmed Tuesday by a top U.S. trade official, and that Nvidia has now “officially begun” shipping H200 chips to China as a result. That implies the company has met the conditions required under the updated export allowance, though the specific licensing framework and exact quantities were not detailed in the available description.

What remains unclear from the information provided is the scope of the restart. The report characterizes the permissions as “limited exports,” but it does not specify whether limitations are based on chip configuration, performance thresholds, end-user eligibility, or geography-wide quotas. It also does not disclose which Chinese customers are receiving the shipments, whether the chips are arriving for new system builds or for replenishing existing deployments, or how quickly supply will ramp.

Nvidia’s China business has been heavily influenced by U.S. export control policy for years, with regulators attempting to balance national-security considerations against the global demand for AI computing. Any easing or narrowing of restrictions can affect sales timing, order visibility, and the composition of Nvidia’s product mix in the affected market.

For the broader technology sector, the resumption of shipments is another sign of how the AI chip supply chain is being managed around policy constraints rather than only commercial demand. Even when hardware remains technically capable of running frontier AI workloads, the ability to move chips across borders can determine which data centers can scale training and inference and how quickly they can refresh infrastructure.

Investors and customers are likely to watch whether the shipments represent a stable, ongoing channel or a temporary authorization. The next indicates to monitor include any further U.S. policy updates, changes in the volume and availability of H200 supply, and public disclosures by Nvidia regarding shipment schedules or customer orders tied to China.