SAN FRANCISCO, March 6, 2026, 03:58 PST
- Nvidia has stopped producing H200 AI chips meant for China, a report said, and shifted capacity to its next-generation Vera Rubin platform.
- The move lands as U.S. officials debate fresh controls on how AI chips can be shipped abroad, including to allies.
- CEO Jensen Huang also signaled Nvidia’s recent investments in OpenAI and Anthropic may be its last as both firms eye IPOs.
Nvidia (NVDA.O) has stopped production of its H200 artificial intelligence chips intended for the Chinese market and shifted manufacturing capacity at Taiwan Semiconductor Manufacturing Co to its next-generation Vera Rubin hardware, the Financial Times reported, citing two people with knowledge of the matter. Reuters could not immediately verify the report. TSMC declined to comment and Nvidia did not respond to a request for comment. 1
The reported shift matters because U.S. export curbs have turned what used to be a straightforward sales channel into a permit-and-compliance exercise, forcing Nvidia to keep reshuffling product plans for China. That uncertainty is now spilling into production decisions, not just customer contracts.
It also comes at an awkward moment for buyers racing to build new AI data centers. When Nvidia moves leading-edge capacity from one chip to another, the ripples can hit delivery schedules across regions, even if the China-bound units were already constrained.
The Financial Times report follows Nvidia’s recent disclosure that it had received U.S. government licenses to ship only “small amounts” of H200 chips to customers in China, suggesting a market that may not scale soon. The H200 is Nvidia’s second-most advanced AI processor, behind its newest flagships.
In Washington, officials are debating a new framework that would put more conditions on large AI chip exports, including requirements that foreign countries invest in U.S. AI data centers or provide security guarantees, according to a document seen by Reuters. The draft would also tighten compliance obligations on chipmakers such as Nvidia and Advanced Micro Devices. 2
Under the proposal, exporters would have to monitor shipments, and recipients would have to use software that would not allow the chips to be linked into a “cluster” — the industry term for large groups of chips used to train or run big AI models.
“The rule could help the U.S. government address chip diversion to China and ensure a more secure buildout of the most powerful AI supercomputers,” Saif Khan, a former Biden administration national security official now at the Institute for Progress, said. He added that the licensing requirements were “overly broad,” raising concerns about whether controls would be used as leverage with allies.
The Commerce Department confirmed on X that it was debating new rules, while arguing it would not copy what it called a “burdensome” framework proposed under the prior administration, according to the Reuters report. The rules are not final and could change.
Nvidia is also trying to manage investor expectations around how far it will go in backing its biggest customers. Speaking at a Morgan Stanley conference on Wednesday, CEO Jensen Huang said “the opportunity to invest $100 billion in OpenAI is probably not in the cards” with the company preparing to go public, and he said Nvidia’s investments in OpenAI and Anthropic might be its last in those firms. 3
The competitive backdrop is tightening, too. Any broad new U.S. licensing regime would apply to multiple AI chip exporters, including AMD, and could reshape how customers split orders across suppliers depending on which products can move, and under what strings.
Still, the big unknown is policy. The export proposal is unfinished, and past chip rules have shifted quickly as national security concerns collide with industry pressure. If Washington tightens further — or if compliance terms discourage buyers from signing up — Nvidia could see more abrupt demand shifts by country, even as it tries to ramp its next wave of chips.
