Nvidia Says It Has Stopped Shipping Licensed AI Chips to China. The Quarter Before, It Sold $2.4 Billion of Them.
Wednesday, November 18, 2026. On Nvidia’s first earnings call since Jensen Huang resigned, interim CEO Colette Kress said the company is “no longer shipping licensed Hopper data center products to China,” per Reuters, which reported the remark raises doubts about a licensing arrangement negotiated between Washington and Beijing.
She gave no reason and did not say whether the halt is permanent. Nvidia’s CFO commentary shows it shipped $2.4 billion of those chips to China in the quarter that just ended, which Stock Titan’s review of the filings identifies as its largest China Hopper quarter since the license rule.
- $2.4Blast quarterlicensed Hopper data center products shipped to China in the quarter ended Oct. 25, per Nvidia's CFO commentary as reported by Reuters.
- ~600,000H200 chipsreportedly stored in Hong Kong awaiting Beijing's approval, per a Nov. 5 Reuters exclusive.
- 25%tariffon eligible chips imported into the U.S. before being shipped to China under the licensing arrangement, per Reuters.
- 10Chinese buyersapproved by Washington for H200 sales, including Alibaba, Tencent and ByteDance, per earlier Reuters reporting.
Kress, Nvidia’s chief financial officer who became interim chief executive after Huang’s November 12 resignation, made the statement Tuesday, per Reuters and Bloomberg. She added that Nvidia’s guidance includes no licensed Hopper data center products and that there are no current plans for further shipments, per Reuters. Stock Titan notes the news came on the call, not in the written results, and quotes Nvidia’s CFO commentary as saying the company is “not assuming any Data Center compute revenue from China in our outlook.”
Hopper is the chip generation that includes the H200, which trains and runs advanced AI models. It is not Nvidia’s newest design; Blackwell and Rubin are newer families. What stopped is the older, licensed product that Washington had decided to allow, per Reuters.
After a summit with Xi Jinping, President Trump said he would allow H200 sales to approved Chinese customers, per Reuters. In January the Commerce Department moved those exports to case-by-case review, with shipments to China capped at half the volume sold to U.S. customers, independent U.S. testing before export, and a 25% tariff on the chips as they pass through the U.S., per Reuters and Stock Titan. Washington approved ten Chinese buyers, per Reuters.
The money never became large until recently. By Stock Titan’s tally of Nvidia’s filings, the company shipped no Hopper products to China in February through April, under 1% of data center sales in May through July, and then $2.4 billion in August through October.
Nvidia’s own August filing, as quoted by Stock Titan, said the company had been “unable to sell all the products for which we have licenses” and that Beijing “encouraged customers to purchase from our China-based competitors.” Nvidia took a $0.4 billion charge on unsold H200 chips and orders, citing diminished demand. White House AI adviser David Sacks said in December that China was “rejecting” the H200 in favor of domestic chips, per Bloomberg.
Reuters reported on November 5 that about 600,000 H200 chips were sitting in Hong Kong because Beijing had not approved large-volume imports, per two unnamed people. A handful of Chinese internet giants had each obtained roughly 10,000 chips. Huawei, meanwhile, expects about $12 billion in AI chip revenue this year, up from $7.5 billion in 2025, per Tom’s Hardware, and Huang said in May that Nvidia’s direct share of China’s AI accelerator market had fallen to zero, before the $2.4 billion quarter.
The halt arrives two days after Commerce released a rule on Chinese companies renting Nvidia chips through overseas data centers, which we covered in our November 17 story, and the day before the Trump-Xi meeting planned for Thursday in Shenzhen, per ABC News. No source in the record ties the halt to either event. Nvidia has not said whether the cause is Beijing’s import hold, the U.S. licensing conditions, the new scrutiny after Huang’s exit, or something else.
What the record does show is the cost of a licensed channel that neither government fully opened: a chip Washington allowed, a buyer list Washington approved, a customer Beijing slowed, and a vendor now declining to forecast any of it.
Documented: Kress’s statement on the call as reported by Reuters and Bloomberg; the $2.4 billion quarter in Nvidia’s CFO commentary; Nvidia’s August filing language about Beijing and unsold licensed chips.
Reported by unnamed sources: the 600,000 H200 chips in Hong Kong and the roughly 10,000 chips each at several Chinese firms.
Not shown: why Nvidia stopped, whether the stop is permanent, what happens to the Hong Kong stockpile, or any link to the Shenzhen summit.
Nvidia has stopped shipping the Hopper-generation AI chips Washington licensed for China, straight after its biggest China haul since the license rule, and it has not said why.



