Nvidia isn’t selling any more AI chips to China
Find out what the shipment stop means for AI hardware supply and how Nvidia’s policy shift could affect your projects.
Nvidia told investors on Tuesday that it has ended all shipments of licensed Hopper H200 AI chips to China, a move that reverses a December White House green light. Interim CEO Colette Kress said the company’s guidance now excludes any Hopper data‑center products for the Chinese market. The decision marks a sharp policy turn after former CEO Jensen Huang publicly defended the right to sell the chips abroad.
The H200 is Nvidia’s flagship Hopper architecture processor, designed for large‑scale generative‑AI workloads and priced at the high end of the data‑center market. Shipping it to China would have opened a sizable new revenue stream, given the country’s aggressive AI compute expansion. Nvidia’s earlier stance relied on a licensing model that allowed foreign customers to use the chips while keeping the underlying design under U.S. control.
Kress did not provide a reason for the halt, but the wording of the earnings call suggests the company is now excluding the product from its official guidance entirely. By removing the H200 from its sales forecast, Nvidia signals that the chips will not count toward its projected revenue for the quarter. The lack of a public explanation leaves analysts to infer that regulatory pressure or export‑control compliance may be influencing the change.
For engineers building AI services, the immediate effect is the loss of a high‑performance, Nvidia‑backed option for Chinese data‑center deployments. Teams will need to look at alternative GPUs, such as AMD’s MI300 series, or consider on‑premise CPU‑accelerator combos. Procurement timelines may stretch as suppliers renegotiate contracts and certify compliance with evolving export rules.
TakeawayNvidia’s withdrawal of H200 chips from China removes a top-tier AI GPU option for the market, forcing a shift to other hardware providers.