NVIDIA Kills H200 Sales in China as Export Net Widens Overseas
NVIDIA interim CEO Colette Kress says the company will stop selling H200 chips in China, as new remote-access rules extend U.S. export controls to overseas data centers serving Chinese customers.

Why it matters
- NVIDIA reported Q3 revenue of $115.8 billion, up 103%, with Data Center revenue of $107.6 billion, up 110%.
- Interim CEO Colette Kress confirmed NVIDIA will no longer sell H200 chips in China, per TechCrunch on November 17, 2026.
- Rules issued November 16, 2026 bar advanced U.S. chips from foreign data centers serving Chinese customers remotely; Jefferies estimates a $44–62 billion hit to 2027 data center revenue.
- Q4 fiscal 2027 guidance of $130.0 billion (±3%) assumes no Data Center compute revenue from China.
- NVIDIA formed a National Security and Export Control Oversight Committee chaired by Suzanne Nora Johnson on November 17, 2026.
NVIDIA will no longer sell its H200 chips in China, interim CEO Colette Kress announced on the company's first earnings call since Jensen Huang resigned, ending a sales arrangement that Huang had personally brokered with President Donald Trump. The disclosure, reported by TechCrunch on November 17, 2026, closes a chapter in one of the strangest deals of the AI trade war — and it lands alongside a new set of export rules that reach far beyond China's borders.
The stakes are enormous. NVIDIA just reported third-quarter revenue of $115.8 billion, up 103% year over year, with Data Center revenue at $107.6 billion, up 110%. That's the business now being re-planned around a China-free future.
Why the H200 deal existed — and why it died
The H200 pathway had a built-in condition: chips had to ship through the United States, which let the U.S. government collect a 25% tariff on each sale. It was a compromise that kept NVIDIA in the Chinese market while giving Washington a cut and a checkpoint. Kress, who also serves as executive vice president and chief financial officer, put an end to it on the call.
The numbers behind the decision are smaller than you might expect. NVIDIA's CFO commentary put Data Center Hopper-product shipments to China at $2.4 billion for the quarter. Measured against the reported Data Center revenue, that's roughly 2% of the segment. One caveat matters here: the Hopper category is broader than the H200 alone, so the figure doesn't isolate what that specific chip earned in China.
China, in any case, had largely stopped taking the deliveries before NVIDIA pulled the plug. Jacob Feldgoise, a researcher at Georgetown's Center for Security and Emerging Technology, attributed Beijing's refusal to a deliberate policy choice: the government wants to build demand for domestic chips, even though major Chinese AI compute users still wanted NVIDIA's hardware.
In plain terms: NVIDIA walked away from a channel its customer's own government had already blocked.
What does the new guidance assume?
The fourth-quarter outlook for fiscal 2027 points to revenue of $130.0 billion, plus or minus 3%, in a period that includes an extra week. The critical detail is what's missing from that number: NVIDIA is assuming zero Data Center compute revenue from China.
That planning assumption is broader than the H200 decision alone. It writes China out of the forecast entirely, not just one product line. For a company that once counted China among its largest markets, this is a structural change to how the business models its future — not a one-off write-down.
The takeaway: NVIDIA's baseline no longer includes Chinese compute revenue in any form.
How do the new rules reach outside China?
The same week brought a second, arguably bigger change. Rules issued on November 16, 2026 bar advanced U.S. chips from foreign data centers that serve Chinese customers remotely. That extends export restrictions beyond direct deliveries into China and into the global cloud market.
The mechanism works through certification. Export-control attorney Dan Fisher-Owens said large data centers seeking NVIDIA chips for AI training would have to certify that they do not serve Chinese customers. In practice, that means a data center in Singapore, Amsterdam or São Paulo can't buy restricted NVIDIA silicon and then rent that capacity to Chinese clients over the network.
The financial exposure is significant. Jefferies estimated that the rules could reduce NVIDIA's data center revenue by $44 billion to $62 billion in 2027. That's a wide range, and it reflects genuine uncertainty about how many Chinese customers currently reach U.S. chips through overseas infrastructure rather than direct imports.
Kress pushed back on the bearish read. She said the backlog of demand for advanced products — the next-generation Vera Rubin platform among them — means the new export controls would not affect top-line revenue. In other words, NVIDIA believes there are enough non-Chinese buyers waiting in line to absorb any lost sales.
That claim and the Jefferies estimate can't both be fully right, and the next few quarters of guidance will show which reading holds.
Why NVIDIA created a standing export-control board committee
NVIDIA didn't stop at sales decisions. On November 17, 2026, it established the National Security and Export Control Oversight Committee, a standing board committee covering national-security policies, export-control compliance and global-trade risk.
The composition signals seriousness. Director Suzanne Nora Johnson chairs the committee, joined by Stephen C. Neal and John O. Dabiri. The committee will meet at least quarterly and work with independent legal, national-security and trade-compliance experts.
The structure also rewires internal reporting. NVIDIA's National Security Team and Global Trade Compliance Team will now report on day-to-day operations to both the committee and Kress. Compliance no longer lives purely inside the executive suite — it has a permanent seat at the board table.
This matters because the remote-access rules make enforcement a global, ongoing problem rather than a customs question at the Chinese border. Certifying that overseas data centers don't serve Chinese customers requires continuous oversight, not a one-time review.
The takeaway: NVIDIA has given export-control risk a standing governance structure, which tells you how central it now is to the business.
What does this mean for the AI chip market?
Three threads tie together here. NVIDIA's outlook removes direct China Data Center compute revenue. The remote-access rules pull overseas facilities and customer certification into the same compliance perimeter. And the new board committee gives that risk a permanent institutional owner.
For the AI industry, the consequence is a bifurcating market. Chinese AI developers lose their last official route to NVIDIA's Hopper hardware — H200 included — and face stronger incentives, on top of Beijing's own preferences, to buy domestic chips. Non-Chinese cloud and AI operators face a new paperwork burden: certify their customer base or lose access to NVIDIA's silicon.
For NVIDIA, the near-term risk is the gap between Kress's confidence and Jefferies's $44–62 billion estimate. Her argument rests on demand for Vera Rubin-class products outstripping supply, and the company's 103% revenue growth suggests that demand is real. But if enforcement of the remote-access rules tightens faster than NVIDIA's non-Chinese backlog grows, the certification regime could bite into sales in ways a backlog can't fully offset.
The new board committee's quarterly meetings will be one place to watch how that tension gets managed — and whether NVIDIA's China-free forecast holds once the remote-access rules are tested in practice.
Source: unite.ai
More from Sophie Lindqvist
Show full bio
Staff writer covering regulation, policy and their impact on artificial intelligence and machine learning at AI In Context.
749 articles
Related articles
- Reuters Finds $3.2 Billion in NVIDIA Chip Projects Nvidia Says It Doesn't Have
- Nvidia scandal collides with Trump-Xi summit as Congress demands answers
- Nvidia Chips Reach China's AI Labs Through a Legal Rental Route
- Nvidia Loses $430 Billion as CEO Affair and China Chip Claims Escalate
- Nvidia waits as pressure grows to restore Jensen Huang