Nvidia Halts Shipments of Licensed Hopper Chips to China After $2.4 Billion in Q3 Sales
- Dispatched: Nov 18, 2026 · 07:38 ET (12:38 UTC)
One-sentence summary: Nvidia interim CEO Colette Kress told the fiscal Q3 2027 earnings call that the company has stopped shipping licensed Hopper-generation data center products to China and that fourth-quarter guidance no longer includes them, casting doubt on the future of the special licensing arrangement negotiated between Washington and Beijing.
Verification status: ✓ Primary official disclosure (earnings call remarks) / cross-reported by Reuters
Core Facts & Consensus Delta:
Reuters first reported the news on 2026-11-18 04:16 ET / 09:16 UTC, by Eduardo Baptista in Shenzhen. Kress said on the November 17 call that Nvidia sold $2.4 billion worth of licensed Hopper-generation data center products to China in the quarter ended October 25, 2026, but that there were no current plans for further shipments and that guidance no longer includes licensed Hopper data center products. She did not explain why Nvidia had stopped shipping the processors or whether the halt would be permanent. The Hopper family includes the H200, a powerful processor used to train and run advanced AI models.
For scale, the $2.4 billion equals about 2% of the quarter's $115.8 billion total revenue; fourth-quarter guidance of $130.0 billion, plus or minus 3%, already excludes these products, so the halt adds no new drag to guidance but zeroes out that revenue line. Context: the U.S. Commerce Department published an interim final rule on November 16 closing the "cloud loophole" for remotely renting advanced AI compute to China (reported separately), with the rule expected to take effect on 2026-11-19.
Supply-Chain & Market Read-Through: The licensed shipment channel to China is narrowing: the H200 uses HBM3E memory, so the halt directly removes one HBM demand channel — yet Nvidia's simultaneously disclosed $390 billion in supply commitments (reported separately) shows its global memory procurement lock-in is still expanding, so a change in one regional channel does not alter the overall tight memory supply picture. For the licensing arrangement and export controls, the halt puts the future of the special U.S.-China licensing deal in doubt; if the halt becomes permanent, watch whether the U.S. government revises the licensing framework and whether Chinese customers shift to domestic alternatives or third-party compute leasing. For data centers and cloud services, the customers directly affected are data center operators inside China, and remote rental of overseas compute is also constrained by the November 16 interim final rule, so both routes are narrowing at once.
Next Catalysts & Watchpoints:
- 2026-11-19: The Commerce Department interim final rule is expected to take effect; check the final text and licensing thresholds.
- 2026-11-19: President Trump meets Presidents Xi and Putin in Shenzhen on the sidelines of APEC; check post-meeting statements on chips and export controls.
- 2026-12-01: Jensen Huang appears for transcribed interviews with three House committees; check the export-control compliance exchanges.
Sources:
- SRN News (full Reuters text) (Published: 2026-11-18 04:16 ET / 09:16 UTC)
- WNCY (full Reuters text) (Published: 2026-11-18 04:16 ET / 09:16 UTC)
Information only, not investment advice.
Read full 2026-11-18 daily market brief & cross-asset tape → All flash alerts
More flash alerts from 2026-11-18
- House Democrats Open Probe Into Whether Lutnick Used Office to Shield Cantor Fitzgerald From Nvidia-Linked Losses⚠ UNCONFIRMED
- AP Calls Alaska Senate Race for Peltola, Locking In a 56-44 Democratic Majority✓ VERIFIED
- Microsoft and Intel Sign Long-Term Azure Deal; Microsoft Becomes First Announced Intel EMIB Packaging Customer✓ VERIFIED
- Fed October Minutes: Almost All Officials Saw Further Hikes as Likely Appropriate, With Unusually Detailed Debate on AI Financing Risks✓ VERIFIED
- Trump Administration Weighs Forcing US Firms to Disclose Use of Chinese AI Models⚠ UNCONFIRMED
- Waller: Fed Needs to Get More Aggressive on Inflation; Half-Point December Hike May Be Needed✓ VERIFIED
- Samsung Executive Downplays Neocloud Bubble Fears: Segment Is Only About 8% of the Market, AI Investment Boom Is Structural, Not Cyclical✓ VERIFIED
- Nvidia Supply Commitments Jump $111 Billion in One Quarter to $390 Billion, Increase Driven Mainly by Multi-Year Memory Procurement✓ VERIFIED
- Semafor: Minority of Nvidia Directors Still Pushing for Jensen Huang's Return as Arm, Marvell CEOs Enter Succession Picture⚠ UNCONFIRMED
- NVIDIA Expands Supply-Chain Commitments With Six Partners to Scale AI Infrastructure Capacity✓ VERIFIED
- US Federal Judge Denies Samsung, SK Hynix and Micron Motion to Dismiss, DRAM Price-Fixing Class Action Advances to Discovery✓ VERIFIED
- Nvidia Guarantee Exposure Rises to $118 Billion, $105 Billion Tied to OpenAI Ohio Campus Leases✓ VERIFIED