Policy & Regulation

Nvidia stops shipping licensed AI chips to China despite US approval

Nvidia says it is 'no longer shipping licensed Hopper data center products to China' and cut them from its $130 billion Q4 forecast, leaving the cause unclear.

Nvidia halts China AI chip shipments despite US approval
Nvidia halts China AI chip shipments despite US approvalLeslin_Liu / pixabay
By Sophie Lindqvist4 min read

Why it matters

  • Nvidia forecasts Q4 revenue of about $130 billion, plus or minus 3%, with zero China data center sales included
  • Colette Kress: 'we are no longer shipping licensed Hopper data center products to China'
  • Nvidia sold $2.4 billion of licensed Hopper data center products to China in the quarter ended Oct. 25
  • Q3 revenue reached nearly $116 billion, more than double a year earlier; data center revenue topped $107 billion, up 110%
  • Jensen Huang resigned last week; CFO Colette Kress is interim CEO

Nvidia has stopped shipping licensed Hopper AI chips to China — even though the U.S. government has approved those exports — and the company has wiped Chinese data center sales out of its revenue forecast for the first time.

Interim CEO and longtime CFO Colette Kress delivered the news on Tuesday alongside Nvidia's quarterly results. "Our guidance does not have any licensed Hopper data center products included, and we are no longer shipping licensed Hopper data center products to China," Kress said in a statement.

The company forecasts fourth-quarter revenue of approximately $130 billion, plus or minus 3%, with no contribution from AI computing chips and systems sold into China. That is a striking omission from a company whose China business ran into the billions just last quarter.

How big was the China business?

Nvidia sold $2.4 billion worth of licensed Hopper-generation data center products to China during its fiscal third quarter, which ended October 25, according to Kress. That figure covered chips Washington had already cleared for export under revised licensing rules.

Removing that stream from guidance is not a rounding error. It represents real, previously authorized revenue that Nvidia now declines to count — and declines to explain.

The takeaway: Nvidia is treating China data center sales as if they do not exist, at least for planning purposes.

What does the broader financial picture look like?

The rest of the business is not slowing down. Nvidia reported third-quarter revenue of nearly $116 billion — more than double the figure recorded a year earlier. Its data center business generated more than $107 billion in revenue, up 110% from the same period last year.

So the China halt lands on a company still growing at extraordinary speed. The question investors and policymakers now face is whether the fastest-growing AI chip market outside the U.S. is effectively closed to Nvidia regardless of what Washington allows.

Why did the shipments stop?

Here the record goes quiet. Nvidia did not explain why the shipments stopped or say whether they would resume. The company has not publicly clarified whether the halt stems from U.S. licensing requirements, Chinese import restrictions, or a separate business decision.

That ambiguity matters. The U.S. government eased its stance earlier this year. In January, the Commerce Department revised its licensing policy for Nvidia's H200 chips, allowing export applications to be considered individually under specified security conditions. The H200 belongs to the Hopper family of processors, which are used to train and run advanced AI systems.

The policy shift followed discussions between President Donald Trump and Chinese President Xi Jinping, along with efforts by Nvidia's former CEO, Jensen Huang, to expand the company's access to China.

In short: Washington opened a door. Nvidia has now, for reasons unstated, walked away from it.

What is Beijing's role?

U.S. export approval has not eliminated barriers to selling in China. Beijing has also restricted H200 imports as it seeks to reduce its dependence on American semiconductor technology.

That leaves Nvidia squeezed from both directions. The U.S. government decides what the company may ship; the Chinese government decides what its customers may buy. When both governments move at once, even an export license can become commercially worthless.

The mechanism is simple and brutal: a chip sale needs permission at both ends of the transaction, and right now Nvidia lacks confidence in at least one of them.

Who is running Nvidia right now?

The announcement also lands in the middle of a leadership transition. Jensen Huang resigned last week amid reports of multiple extramarital affairs, including one with an alleged Chinese spy. Kress was appointed interim CEO while continuing to serve as the company's chief financial officer.

Huang had been the public face of Nvidia's China push, personally lobbying for expanded access. His departure removes the executive most closely identified with that effort, and Kress — a finance chief doubling as interim CEO — is now the one drawing a hard line around China revenue in guidance.

The signal to the market is caution: until Nvidia explains otherwise, investors should not model China data center sales at all.

Why does this matter beyond Nvidia?

The stakes run wider than one company's forecast. Nvidia's Hopper chips sit underneath much of the world's AI training infrastructure, and China is the largest AI market the U.S. still partially licenses exports to. If those sales stop despite U.S. approval, the effective policy is more restrictive than the written one.

It also accelerates Beijing's stated goal of reducing dependence on American semiconductors. Every quarter Chinese buyers go without Nvidia hardware is a quarter in which domestic alternatives — or non-U.S. suppliers — gain ground.

For now, Nvidia's silence leaves the central question unanswered: the company has the licenses, but it is not shipping, and it will not say why. Whether shipments resume, and on whose terms, will shape both Nvidia's addressable market and the direction of the U.S.-China technology standoff.

Source: moxie.foxbusiness.com

Share this article:

More from Sophie Lindqvist

Sophie Lindqvist

Show full bio

Staff writer covering regulation, policy and their impact on artificial intelligence and machine learning at AI In Context.

777 articles

Related articles

  1. NVIDIA Kills H200 Sales in China as Export Net Widens Overseas
  2. Chinese AI Data Centers Still Specify Restricted Nvidia Chips
  3. Nvidia Chips Reach China's AI Labs Through a Legal Rental Route
  4. US export rule forces Southeast Asian data centers to get licenses for AI chips
  5. Reuters Finds $3.2 Billion in NVIDIA Chip Projects Nvidia Says It Doesn't Have

« Previous articleNext article »