AI & Computing NewsNews

Nvidia Stops H200 Shipments to China as US Expands Overseas Chip Controls

The halted China shipments were about 2% of quarterly revenue by BigGo's figures, while a Commerce rule issued a day earlier reaches chips in other countries' data centers.

Key Takeaways

  • On its November 17 earnings call, interim CEO Colette Kress said Nvidia is no longer shipping licensed Hopper data center products to China. That covers the H200.
  • Bloomberg, via Gizmodo, puts those sales at $2.4 billion last quarter. 
  • A Commerce rule issued November 16 requires licenses for chips going to data centers that serve China-headquartered customers remotely. 
  • The rule’s effective date is reported only as expected.

Nvidia‘s China headline is the smaller story. The company has stopped shipping H200s, but TechCrunch reports Chinese buyers had largely turned them away already, and the revenue was small next to Nvidia’s size. 

The Commerce Department regulation that arrived the day before has more reach. It targets chips in data centers outside China that Chinese companies rent. 

The rule addresses a major loophole that analysts at Jefferies project could cost Nvidia tens of billions in data center revenue next year. 

What Nvidia actually stopped

Nvidia’s interim CEO Colette Kress chose her words carefully on the earnings call. As Gimodo quotes her, “we are no longer shipping licensed Hopper data center products to China”.

Hopper is the architecture, and the H200 is the version Washington allowed in a deal that sent chips through the US so the government could collect a 25% tariff, TechCrunch reports. The sales run reportedly ended last week, when Hunag resigned.

TechCrunch cites Georgetown’s Jacob Feldgoise saying Beijing had largely turned the chips away, preferring to build demand for domestic ones. 

Because of that shift, Nvidia had already excluded those China Hopper sales from its soaring fourth-quarter guidance of $130 billion.

The Rule That Reaches Further

Shipments of advanced semiconductors to Indonesia, Malaysia, Singapore, Thailand, and Brazil now require case-by-case export licenses. 

Furthermore, a license is mandatory whenever an exporter knows chips are heading to an overseas data center offering more than 12 million in aggregate processing performance to firms headquartered in China or similar restricted destinations. 

The threshold counts computing capacity across linked facilities. 

If a data center refuses or fails to share customer due diligence data, exporters are legally presumed to know of a restricted end use. 

The interim final rule takes effect upon publication in the Federal Register on November 19, 2026, giving existing operators a 90-day grace period to obtain required licenses.

What Changes For Those Who Run and Rent Compute

This rule moves the question from who bought a chip to who is using it. A data-center operator in Singapore or Malaysia now has to know where its tenants are headquartered, and customers with no connection to China may still get asked.

Whether it can be enforced is open. 

Kress said demand for products like Vera Rubin means top-line revenue won’t suffer, though Nvidia may lose Southeast Asian customers over time. 

TechCrunch notes critics’ claim that up to 20% of revenue ties to China isn’t verified, though recent reporting highlights that more than $3 billion in Chinese AI infrastructure projects still hinge on restricted Nvidia chips. 

Source: Nvidia is abandoning China after Jensen Huang’s ouster

NogenTech News Desk

NogenTech News Desk covers the latest developments in technology, AI, software, SaaS, and emerging digital trends. The team reports on product launches, company updates, and industry developments, with each story reviewed for accuracy, clarity, and relevance before publication.

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button