JupiteX Get the app
Economy & Business19 Nov 2026 · about 7 min

Nvidia isn’t selling any more AI chips to China

The brief

Nvidia has stopped shipping licensed Hopper data-center products to China, including H200 chips, even though the White House approved sales in December last year. The announcement marks a change from the company’s earlier effort, led by former CEO Jensen Huang, to secure permission to sell the chips. It matters because the decision halts a planned route for Nvidia to sell these advanced products in China. Interim CEO Colette Kress announced the change during an earnings call. She said Nvidia’s sales guidance no longer included licensed Hopper data-center products and that the company was no longer shipping them to China. She did not explain the reason. Separately, Reuters reported that about 600,000 H200 chips were stuck in Hong Kong after Beijing withheld approval for large-scale imports. The company has not clarified whether it will pursue other business in China. The article also describes concerns about properly vetting Chinese companies approved to handle the chips. So the immediate reality is that licensed Hopper shipments have stopped, while the company’s future China business remains unclear. The article does not say what Nvidia might sell there instead.

01

What did Nvidia announce about shipping Hopper data-center chips to China?

Nvidia has stopped shipping licensed Hopper data-center products to China, including H200 chips, even though the White House approved sales in December last year. The announcement marks a change from the company’s earlier effort, led by former CEO Jensen Huang, to secure permission to sell the chips. It matters because the decision halts a planned route for Nvidia to sell these advanced products in China.

Interim CEO Colette Kress announced the change during an earnings call. She said Nvidia’s sales guidance no longer included licensed Hopper data-center products and that the company was no longer shipping them to China. She did not explain the reason. Separately, Reuters reported that about 600,000 H200 chips were stuck in Hong Kong after Beijing withheld approval for large-scale imports.

The company has not clarified whether it will pursue other business in China. The article also describes concerns about properly vetting Chinese companies approved to handle the chips. So the immediate reality is that licensed Hopper shipments have stopped, while the company’s future China business remains unclear. The article does not say what Nvidia might sell there instead.

02

What are H200 chips, and what do AI data-center processors do?

The article identifies the H200 as a Hopper architecture data-center chip, but does not define its technical design. In general computing terms, an H200 is a graphics processing unit designed to accelerate demanding workloads. AI data-center processors are specialized hardware used to perform large numbers of calculations for machine-learning systems.

During training, processors calculate how a model’s internal settings should change as it learns from data. When a trained system answers a prompt or generates content, processors also carry out the computations needed to produce that result. The article does not detail these tasks; they are standard functions of AI computing hardware. A data center brings many such processors together to handle large workloads.

The H200 matters here because Nvidia had sought permission to sell it in China, but now says it has stopped shipping licensed Hopper data-center products there. Reuters reported that around 600,000 H200 chips were stuck in Hong Kong after Beijing withheld approval for large-scale imports. The article does not say what processors Chinese companies may use instead.

03

How many H200 chips were reportedly stuck in Hong Kong?

Reuters reported that around 600,000 Nvidia H200 chips were stuck in Hong Kong. That is the article’s reported figure, not a statement that the chips had reached Chinese buyers. The distinction matters: the report describes chips held up before large-scale imports into China received approval.

The article connects the delay to Beijing withholding approval for those imports. It does not give details about the shipment’s ownership, how long the chips had been there, or whether any were later released. Separately, Nvidia’s interim CEO Colette Kress said the company was no longer shipping licensed Hopper data-center products to China.

The reported stockpile helps explain why the shipments were already facing difficulties, even before Nvidia’s announcement. China was seeking to reduce its dependence on US technology, according to the article. The figure does not establish how many chips Chinese companies ultimately received or how much computing capacity they had available. Nvidia did not clarify whether it would pursue other business with China.

04

What could stopping these shipments mean for Nvidia and Chinese companies building AI systems?

H200 chips are advanced data-center products that Nvidia had sought approval to sell in China. Stopping shipments could mean Chinese companies have less access to that particular source of computing hardware for AI systems. It could also mean Nvidia gives up sales it had hoped to make there. These are possible consequences; the article does not measure their size or say what companies will do next.

The immediate mechanism is the end of licensed Hopper data-center shipments. Interim CEO Colette Kress said Nvidia’s sales guidance no longer included those products and that the company was no longer shipping them to China. Separately, Reuters reported that around 600,000 H200 chips were stuck in Hong Kong after Beijing withheld approval for large-scale imports.

The article gives no estimate of lost revenue, the chips’ share of Chinese computing capacity, or the effect on AI development. It also says Nvidia has not clarified whether it will pursue other business with China. China was seeking to reduce its dependence on US technology, a factor that could shape what happens next, but no specific alternative is identified.

05

Why did shipping the chips involve both US export approval and Chinese import approval?

A cross-border chip sale can be subject to rules in both the country shipping the product and the country receiving it. In general, an exporting government may control whether a company can send certain technology abroad, while the destination government may separately control whether that technology can be imported. One government’s approval does not automatically grant the other’s.

That distinction appears in this case. The article says Nvidia received White House approval in December last year to sell H200 chips in China. Yet Reuters reported that around 600,000 H200 chips were stuck in Hong Kong after Beijing withheld approval for large-scale imports. The two approvals address different sides of the transaction.

The article does not explain the detailed rules or give a full account of the approval process. It does report that China was seeking to reduce its dependence on US technology. Whatever the broader policy context, the practical result described is that US approval did not ensure the chips could enter China at scale. Nvidia has now stopped shipping licensed Hopper data-center products there.

06

What other ways could China get the computing power it needs for AI?

The article does not describe alternatives China could use to obtain AI computing power. In general, possible routes include developing domestic processors, buying suitable chips from suppliers outside the United States, or using older hardware. These are potential approaches, not claims about what Chinese companies currently have available or are doing.

Companies can also make better use of existing computing capacity by sharing resources or adjusting how they train and run models. Such steps may help manage a shortage, but they do not necessarily replace the performance or scale of advanced data-center processors. The article provides no details about Chinese firms’ infrastructure, chip supplies, or technical plans.

The immediate facts are that Nvidia says it has stopped shipping licensed Hopper data-center products to China, and Reuters reported about 600,000 H200 chips stuck in Hong Kong after Beijing withheld approval for large-scale imports. China was seeking to reduce its dependence on US technology. That context points to a possible interest in alternatives, but the article does not identify any specific substitute or say what comes next.

07

Why are advanced semiconductors essential for training and running modern AI systems?

Modern AI systems rely on large amounts of computation. In general, advanced semiconductors are important because they can perform the many calculations involved in building and using these systems. The source identifies Nvidia’s H200 as a Hopper data-center product, though it does not explain the chip’s technical specifications or describe AI computation in detail.

During training, a system processes examples and adjusts its internal settings; running a trained system also requires calculations to respond to inputs. Specialized processors can accelerate such workloads. Data centers use processors to provide computing capacity at scale. These are general facts about AI hardware, not details supplied by the article.

The article’s example is Nvidia’s effort to sell H200 chips in China and its later decision to stop shipping licensed Hopper data-center products there. Reuters reported around 600,000 H200 chips stuck in Hong Kong after Beijing withheld approval for large-scale imports. The article does not quantify the chips’ performance or say how their absence will affect AI projects. It does show how access to advanced processors can become a point of friction in cross-border technology trade.

This brief was written by AI from the original reporting and checked by other models. Names, figures and quotes come from the source; read it for full context.

Read more in the JupiteX app

Pulse is free. New stories every 4 hours, each one broken into the questions that explain it.

Or read more news on the web