The Daily Inference
AI & Technology · News · Developing

Record $115.8 Billion Quarter, and Nvidia Stops Selling Chips to China

The chipmaker says it is no longer selling its Hopper-generation AI processors to China, ten months after Washington allowed licensed sales. It announced the halt on the day it reported $115.8 billion in quarterly revenue, with an interim chief executive in charge.

Developing: this story is still unfolding and details may change.

Follows Nvidia Revenue Doubles to $115.8 Billion as Board Rules Out Huang's Return, Trump Taxes China Chips, and Democrats Want the Receipts.

Nvidia has stopped shipping its Hopper-generation AI chips to China, a family that includes the H200, interim chief executive Colette Kress said on November 17 [5]. The move closes a sales route the Trump administration reopened under conditions in January [3].

Kress said on the company's earnings call that Nvidia is no longer shipping licensed Hopper data-center products to China and that its forecast includes none [5]. CBS News reported that the Hopper family includes the H200 [5]. She gave no reason [5].

The story is still developing. In the statements made public so far, Nvidia has not explained the decision, and it has not said whether the halt covers every H200 sale or only some licensed shipments.

Record sales, no China

The announcement landed beside very large numbers. Nvidia's revenue for the quarter ended October 25 was $115.8 billion, up 103% from a year earlier and 20% from the previous quarter. Data Center revenue was $107.6 billion, up 110% [2].

For the next quarter Nvidia forecast $130 billion, plus or minus 3%, and said it was not assuming any Data Center compute revenue from China [2]. Kress said licensed Hopper sales to China came to $2.4 billion in the quarter, about 2% of the total [5].

That puts at most about two cents of every revenue dollar at stake in the halt. The true share is smaller if some of those sales were not H200s, and the company is guiding to growth without any of it.

What the H200 is

The H200 is not a gaming card. It is a data-center processor, built on Nvidia's Hopper design, for training and running large AI models [4]. Nvidia says it carries 141 gigabytes of high-bandwidth memory, which lets the chip hold and move large volumes of data. Think of a bigger workbench: a bigger job fits without parts being shuffled on and off. Nvidia also advertises faster performance than the older H100 on selected workloads, a claim that comes from the company's own product page [4].

China matters because of the scale of its demand for advanced computing, and because Nvidia has spent years trying to keep access to it.

A door opened in January

On January 13, the US Commerce Department said it would review export-license applications for H200 and similar chips to China case by case. The conditions included customer screening, third-party testing in the United States and assurances that exports would not reduce chip capacity available to American customers [3].

TechCrunch reported that Jensen Huang and President Donald Trump brokered the arrangement, that the chips had to pass through the United States, and that the US government took a 25% cut [1]. The Commerce notice confirms the conditional review policy. It does not describe any payment [3].

The payment is already a political fight. The Daily Inference reported on November 13 that two senior congressional Democrats want Commerce Department records on chip-export approvals and Nvidia's contacts with the administration. Their letter alleges an unconstitutional export tax.

Beijing did not walk through

The door opened, but the traffic was thin. TechCrunch cited Jacob Feldgoise, a researcher at Georgetown's Center for Security and Emerging Technology, who said China had largely refused H200 deliveries because Beijing preferred to build demand for domestic chip suppliers [1].

That is one researcher's view. No Chinese statement on the halt has surfaced so far. Nor is it clear how much of the $2.4 billion in licensed Hopper sales was H200 revenue, or how much of it Chinese customers accepted [5]. The halt is a retreat from a prospect. It does not show that a large H200 business in China has been lost.

Huang had argued the opposite case in public. In April he told the podcaster Dwarkesh Patel, "to concede that market for the United States technology industry is a disservice to our country" [1].

A company without its founder

Huang resigned on November 12, and directors have since rejected bringing him back. Kress, the chief financial officer, is running the company in the interim [2].

Her own words about the quarter were bullish. "Our momentum is real," she said in the earnings release, which also said Vera Rubin, a newer Nvidia product line the release names, was off to an "explosive start" [2].

On the same day as the shipping announcement, TechCrunch reported, Nvidia created a board-level committee to oversee national security policy, export-control compliance and global trade risk [1]. Its mandate and membership have not been described so far.

The nearest thing to an explanation comes from a Pepperstone research strategist quoted by CBS. He said the timing might indicate Nvidia was trying to clarify its China business and reassure regulators [5]. That is his interpretation, not a company statement.

The halt may also anticipate rules still to bite. TechCrunch reported new US restrictions on advanced chips at overseas data centers that serve Chinese customers remotely, and cited a Jefferies estimate that they could cut Nvidia's data-center revenue next year by $44 billion to $62 billion [1]. That is an analyst's projection, not a company figure and not a realized loss.

This paper's own inference is that a range that size, against a $130 billion quarterly forecast, gives any board a reason to want export rules watched closely. No source says that is why the committee exists.

Washington has not said whether any H200 licenses remain active. Whether China formally restricted the chips is also unanswered. For now Nvidia is run by an interim chief executive, and the committee has yet to say a word in public.

Our call

The US Commerce Department formally rescinds or replaces its January 13, 2026 policy of case-by-case license review for H200-class chip exports to China.

Nvidia has stopped shipping, China has reportedly largely refused deliveries, and congressional Democrats are demanding records on the payment arrangement, so the policy has little commercial use left and a growing political cost.

Elena MarshElena MarshSenior writerWrites the long pieces that make people read to the end. Believes every story is a history story if you go back far enough, and usually goes back just far enough.Model: Claude Sonnet 5.5 says 15%. Jev says 18%.

By February 12, 2027. How we will check: A Bureau of Industry and Security or Federal Register notice published between November 20, 2026 and February 12, 2027 that revokes, replaces or materially tightens the January 13, 2026 case-by-case review policy for H200 and similar chips to China. The notice itself decides the verdict.

Open · Every call we made

The story so far 12 pieces since October 29, 2026, in order

Topics: Nvidia · China · AI chips · Jensen Huang · Donald Trump

Every edition in brief, three times a day, on our Telegram channel, on Bluesky and on Threads.

Sources
  1. Nvidia is abandoning China after Jensen Huang's ouster | TechCrunch techcrunch.com
  2. NVIDIA Corporation - NVIDIA Announces Financial Results for Third Quarter Fiscal 2027 investor.nvidia.com
  3. Lock bis.gov
  4. H200 GPU | NVIDIA nvidia.com
  5. Nvidia says it will stop selling some AI data center chips to China - CBS News cbsnews.com