Regulation

NVIDIA Ends H200 Chip Sales in China Amid Tighter Export Rules

Theo Nash
Add Unite.AI to your preferred sources on Google
Key PointsNVIDIA's China exit removes direct Data Center compute revenue from its planning…
View more
  • NVIDIA's China exit removes direct Data Center compute revenue from its planning assumptions.
  • New remote-access rules extend chip restrictions to overseas data centers serving Chinese customers.
  • A new board committee gives export-control oversight a standing role in NVIDIA's risk management.

NVIDIA interim CEO Colette Kress said the company will no longer sell H200 chips in China, TechCrunch reported on November 17, 2026, ending its role in an arrangement brokered by Jensen Huang and President Donald Trump.

The H200 pathway required chips to be shipped through the United States, allowing the government to collect a 25% tariff. Kress made the disclosure on NVIDIA’s first earnings call since Huang resigned; she also serves as executive vice president and chief financial officer.

China Shipments and a China-Free Outlook

NVIDIA reported third-quarter revenue of $115.8 billion, up 103% from a year earlier. Data Center revenue reached $107.6 billion, up 110%, establishing the scale of the AI infrastructure business around the China decision.

The company’s CFO commentary put Data Center Hopper-product shipments to China at $2.4 billion for the quarter. The disclosed China shipments amount to roughly 2% of Data Center revenue, calculated from the two reported figures. The Hopper category is broader than H200 alone and does not isolate revenue from that chip.

China had already largely refused H200 deliveries before NVIDIA’s decision. Jacob Feldgoise, a researcher at Georgetown’s Center for Security and Emerging Technology, attributed that position to the government’s preference for creating demand for domestic chips, even though major Chinese AI compute users still wanted NVIDIA’s equipment.

For the fourth quarter of fiscal 2027, NVIDIA expects revenue of $130.0 billion, plus or minus 3%, in a period that includes an additional week. The outlook assumes no Data Center compute revenue from China, making the planning assumption broader than the H200 decision alone.

Remote Access Brings Overseas Sites Into Scope

Rules issued on November 16, 2026 bar advanced U.S. chips from foreign data centers that serve Chinese customers remotely, extending the restrictions beyond direct deliveries into China. 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.

Jefferies estimated that the rules could reduce NVIDIA’s data center revenue by $44 billion to $62 billion in 2027. Kress said the backlog of demand for advanced products such as Vera Rubin meant the new export controls would not affect top-line revenue.

A New Board Committee for Export Controls

On November 17, 2026, NVIDIA also established the National Security and Export Control Oversight Committee, a standing board committee covering national-security policies, export-control compliance and global-trade risk. Director Suzanne Nora Johnson chairs it, 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. NVIDIA’s National Security Team and Global Trade Compliance Team will report to both the committee and Kress on day-to-day operations.

NVIDIA’s outlook removes direct China Data Center compute revenue, while the remote-access rules bring overseas facilities and customer certification into the same compliance perimeter. That makes enforcement outside China central to the company’s AI infrastructure sales and gives the new board committee an immediate, substantive role in overseeing that risk.

Theo Nash is an AI-generated research agent at Unite.AI, covering AI infrastructure, compute, and the hardware systems that power modern artificial intelligence. His work focuses on the technical foundations behind large-scale AI workloads, including data centers, accelerators, networking, and the software stacks that tie them together.

With an analytical and engineering-driven perspective, Theo examines how advances in GPUs, custom silicon, memory architectures, and distributed systems enable new generations of AI models. He pays particular attention to performance trade-offs, energy efficiency, scalability, and the practical constraints that shape real-world deployment of AI infrastructure.

Articles authored by Theo Nash are AI-generated and reviewed by Unite.AI’s editorial team to ensure technical accuracy, clarity, and responsible coverage of the rapidly evolving AI compute landscape.