What happened
Semafor’s investigation, published November 13, 2026, says that in January Nvidia general counsel Tim Teter delivered the company’s position on proposed export-license conditions to more than a dozen government officials. According to people with access to a detailed account of the call, Teter said he had been with President Trump when Trump and Nvidia co-founder Jensen Huang discussed the issue, and that Trump wanted the process to move quickly. Semafor reports that officials on the call fell silent.
The proposed conditions concerned licenses to export Nvidia H200 chips to China. They included requiring sellers to know which cloud customers would receive chips before shipping and conducting know-your-customer checks on buyers purchasing more than a few thousand chips. Semafor says Nvidia objected to the cloud-customer requirement, which Teter called a “poison pill,” and wanted the threshold for buyer checks raised to tens of thousands of chips—an amount the article describes as worth billions of dollars. Nvidia also sought to limit end-of-life conditions to chips already subject to export controls.
The story places those license discussions within a wider account of Nvidia’s work in Washington. Semafor says Teter and Nvidia lobbyists often invoked Trump’s name while pushing the company’s agenda. Current and former congressional staffers told the outlet that Nvidia argued some restrictions were technically infeasible and that its claims could be difficult for staff without technical expertise to assess. One former staffer said Nvidia had described location verification as requiring new technology and being easy to spoof. The article says Reuters later reported on a location-verification feature that Nvidia had demonstrated privately for months and launched in May. Nvidia declined to comment on that technology.
Semafor also reports that Commerce Secretary Howard Lutnick was receptive to allowing H200 sales, citing people familiar with the matter and a former Commerce official. The Bureau of Industry and Security, within the Commerce Department, said company consultation on license conditions was routine and that conditions can change in response to legitimate commercial considerations. Its spokesperson said final license conditions are reviewed by the State, Defense, Energy, and Commerce departments, and that Nvidia went through the same process as other license applicants. Nvidia declined to comment on the investigation; White House spokespeople did not immediately respond to Semafor’s requests for comment.
What the archive adds
New Runtime’s earlier coverage argued that scrutiny of Nvidia’s China access could continue without proof of information transfer. This investigation adds reported detail about the company’s lobbying over H200 license conditions and chip-location rules, making it a new fact on a known story: Congressional and White House scrutiny of Nvidia’s China access.
Trend signal
This episode supports the view that scrutiny of Nvidia’s access to China will persist even absent proof of information transfer. It adds reported examples of the policy dispute, but does not resolve the underlying questions about enforcement or diversion: the earlier analysis. This is a political and regulatory signal, not evidence of a change in AI development practice.
Why now
Semafor’s reporting comes amid renewed scrutiny of Nvidia’s Washington relationships and China-related chip sales. The article says the license discussions gained significance after reports about Huang’s relationships with subordinates and a customer suspected of being a Chinese spy, and notes reports about sales and possible chip diversion. Those claims form context for the investigation; they do not establish wrongdoing in the license discussions.
Connections to other events
New Runtime’s November 12 coverage of Huawei’s data-center plans described China’s effort to expand domestic AI compute capacity and the infrastructure constraints that may limit it. Semafor’s investigation concerns US export policy rather than Chinese data-center construction, but both stories show how access to compute is shaped by forces beyond chip performance: Huawei and China’s data-center ambitions.
Why it matters
For teams building AI products, compute availability is a strategic dependency with a geopolitical dimension. Export conditions can affect which accelerators reach which markets, how cloud providers document customers, and whether hardware can be tracked after shipment. The reported dispute also illustrates an information asymmetry: policymakers may need to evaluate technical claims made by companies with deep knowledge of their own products, while those decisions carry security and commercial consequences across the industry.
The article does not establish that Nvidia’s lobbying caused a particular policy outcome, nor that any specific chip was diverted. Its value is as a reported account of the channels through which industry influence may enter rulemaking. Builders should treat export access and compliance as operating risks, not assume that today’s hardware availability or rules will remain fixed.
Limits and uncertainty
The account relies on people familiar with discussions and staffers speaking to Semafor; the article does not publish the detailed call account or the proposed license text. Nvidia declined to comment, and White House representatives had not responded at publication. BIS characterized consultation as routine and described interagency review of final conditions. The investigation does not independently establish that Nvidia’s lobbying determined the final license terms, that officials acted improperly, or that particular chips were diverted to China. The claims about Megaspeed sales and location verification are attributed to other reporting and sources quoted by Semafor, not independently demonstrated in the article.
Reader takeaway
Semafor’s reporting adds detail to the debate over how Nvidia’s Washington access intersected with proposed China chip-export controls. For AI builders, the practical point is not to infer misconduct from reported lobbying, but to recognize that supply and compliance conditions for advanced compute can be shaped through policy negotiations as well as technical and market forces.
Open question
What conditions ultimately governed H200 exports to China, and how did they differ from the proposals Nvidia challenged?