wasl.tech

Nvidia H200 China Sales End as Kress Closes the Door After Huang’s Exit

Nvidia H200 China sales are finished. For months, the chips reached Beijing through a strange detour: they travelled through the United States first, so Washington could collect a 25 percent tariff on each one. On Tuesday, that detour closed, when interim chief executive Colette Kress told investors the sales will stop.

It was a quiet end to a loud argument. For years, Nvidia’s co-founder Jensen Huang had fought to keep the Chinese market open, warning that cutting it off would hand the future of artificial intelligence to rivals. His successor ended the debate in a single sentence on her first earnings call.

The decision lands at a strange moment for the world’s most valuable chipmaker. Nvidia has just reported the biggest quarter in its history, yet its leadership is brand new, its founder is gone, and Washington is tightening the rules around it from every side.

What the Nvidia H200 China deal actually was

The arrangement was never a normal export deal. Advanced AI chips were effectively banned from sale to China, so the H200 travelled through a loophole: the chips were shipped via the United States, the government took its 25 percent cut, and the hardware continued on to Chinese buyers.

Huang brokered the deal with President Donald Trump, who approved H200 exports in January 2026. The chips are about six times more powerful than the H20 models previously allowed into China, and Chinese companies reportedly placed orders for more than two million of them, potentially worth $54 billion.

Those were orders, not completed sales, and the distinction matters. According to reports, China had already largely refused H200 deliveries before Kress made her announcement. The market was cooling before the door was shut.

The earnings call that changed the strategy

Kress made the announcement on Nvidia’s first earnings call since Huang resigned on November 12, according to TechCrunch. She also serves as the company’s chief financial officer, a role she has held since September 2013.

The numbers she reported were extraordinary. Nvidia’s third-quarter revenue reached $115.8 billion, up 103 percent from a year earlier, with data-centre sales of $107.6 billion. The company guided to about $130 billion for the current quarter, assuming no data-centre revenue from China at all.

Yet the call was not really about the numbers. It was the first time the new leadership had to explain, in public, what kind of company Nvidia intends to be without the man who built it.

The $44 billion question

Walking away from China is not free. Investment bank Jefferies has forecast that the new export rules could trim between $44 billion and $62 billion from Nvidia’s data-centre revenue next year.

Kress pushed back on that fear. She told investors that the backlog of demand for advanced chips such as Vera Rubin is so large that the rules would have no impact on the company’s top-line revenue. In other words, every chip China will not buy, someone else already wants.

That argument has a limit, though. The company’s own finance team put Hopper-product shipments to China at $2.4 billion for the quarter just ended. A market that large does not vanish without leaving a mark, even for a company growing this fast.

A new committee, and a new era of oversight

On the same day as the call, Nvidia announced a new standing board committee: the National Security and Export Control Oversight Committee. It will oversee national-security policy, export-control compliance and global trade risk.

The committee is a direct response to the scandal that ended Huang’s 33-year run. His resignation followed reports of an affair with the chief executive of Megaspeed, a company under investigation for smuggling chips into China.

US intelligence reportedly believes the executive helped move 240,000 banned Nvidia chips into China, mostly H100 and H200 models, a story wasl.tech covered when records showed blacklisted Chinese universities buying H200 chips anyway.

Board chair Harvey Jones has said the board will not reinstate Huang, who offered to return. Minutes before the earnings report, news also surfaced of a Securities and Exchange Commission investigation into Nvidia’s investor disclosures.

The search for a permanent chief executive is now under way, with the board considering internal and external candidates. Names floated in reports include Arm chief Rene Haas, while Treasury Secretary Scott Bessent has urged an outside successor search. Huang himself is due to testify before Congress on December 1.

Why Beijing may not miss the chips

The uncomfortable truth for Washington is that the controls may already be leaking. TechCrunch reports that all of China’s leading AI models show evidence of having been trained on Nvidia Grace Blackwell chips that are prohibited from being exported to China.

That is why the rules issued on Monday, November 16, matter as much as Kress’s announcement. The new regulations bar advanced American chips from foreign data centres that serve Chinese customers remotely, closing the workaround that let Chinese firms rent computing power abroad.

The US Commerce Department moved to close that loophole the same week wasl.tech reported on the interim final rule targeting remote access to AI chips. Together, the rule and the H200 decision seal off both the front door and the side entrance.

What Nvidia keeps, and what it gives up

Not everything is changing. Kress reaffirmed Nvidia’s open-source strategy, including the company’s $12.9 billion agreement to acquire Hugging Face, the open model-hosting platform. Hugging Face chief executive Clement Delangue has also publicly recommitted to open-source AI since Huang’s departure.

That matters because Nvidia’s power has never been chips alone. Its software, its developer tools and its open ecosystem are what keep researchers building on its hardware. Keeping that ecosystem open is the one part of Huang’s strategy the new leadership has chosen to protect.

What happens next

The immediate question is whether the market believes Kress. Nvidia’s shares have already fallen by roughly a quarter since the start of November, and the company now has to prove that demand elsewhere can replace a market it spent years fighting to keep.

The longer question is about China itself. If Chinese labs are already training leading models on smuggled or remotely rented hardware, cutting off official sales may change less than Washington hopes. And if domestic Chinese chips keep improving, the door Kress closed may not matter much in a few years.

Chinese buyers, meanwhile, face their own reckoning. With more than two million H200 chips reportedly on order and no legal route left, attention will turn to domestic alternatives and to whatever stock is already inside the country. The smuggling networks that moved 240,000 banned chips will not disappear overnight.

For now, the era of the tariff detour is over. The chips that travelled through America to reach Beijing will travel no further. What replaces that trade — for Nvidia, for China, and for the rest of the AI industry — is the question the next few quarters will answer.

Leave a Comment

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

Scroll to Top