【Wen/Observer Net Zhang Jingjuan】Reuters on the 22nd cited three anonymous sources who reported that NVIDIA has informed its Chinese customers that it plans to deliver H200 artificial intelligence (AI) chips to China before the Lunar New Year holiday in mid-February 2026.

If this move is implemented, it will be the first substantive delivery following the Trump administration's announcement earlier this month allowing NVIDIA to export H200 chips to China, but with a 25% fee.

Two sources revealed that NVIDIA will use existing inventory for this delivery, with an estimated shipment of 5,000 to 10,000 chip modules, equivalent to approximately 40,000 to 80,000 H200 chips.

A third source added that NVIDIA also informed its Chinese customers that it plans to increase the production capacity of this chip, and the order application channel for related capacity will open in the second quarter of 2026.

However, all delivery plans are subject to significant uncertainty. The Chinese side has not yet approved any H200 chip purchase orders, and the final delivery schedule may be adjusted according to Chinese decisions.

"Whether the entire plan can be implemented depends entirely on the approval result from the other side," said the third source. Before obtaining Chinese permission, everything remains uncertain.

According to the report, this proposed export to China marks the first substantive action after the Trump administration's shift in chip policy toward China. From the Biden administration to the Trump administration, the United States has continuously increased export restrictions under the pretext of "national security" to suppress and contain China's technological development. Earlier this month, Trump announced that he would allow NVIDIA to export its H200 chips to Chinese and other country customers, but with conditions including the U.S. government receiving a 25% share.

Reuters reported last week that the Trump administration had initiated an interdepartmental review process to assess the license applications for exporting H200 chips to China, fulfilling its commitment to allow such chip sales to China.

NVIDIA H200 Chip Photo NVIDIA Official Website

The H200 chip belongs to NVIDIA's previous Hopper series, although it has been replaced by the next-generation Blackwell series, it is still a mainstream product in the field of artificial intelligence. Currently, NVIDIA's production capacity is mainly focused on the Blackwell series and the upcoming Rubin series, leading to a tightening supply of H200 chips.

The report states that the Trump administration made this decision at a critical time when China is vigorously promoting the development of its domestic AI chip industry. At present, the chip products of Chinese domestic companies have not yet reached the level of H200, so some views worry that allowing the import of this chip may slow down the development process of the domestic industry.

According to a report by the U.S. think tank Progressive Policy Institute, the performance of H200 is six times that of the "customized and cut" version of H20 designed specifically for the Chinese market. Exporting this chip will enable Chinese AI laboratories to build supercomputers with performance close to the top AI supercomputers in the United States, although at a higher cost.

In response to the U.S. decision to allow the release of H200 chips, Chinese Foreign Ministry Spokesperson Guo Jiakun stated during a regular press conference that China has always advocated mutual benefit and win-win cooperation between China and the United States.

This article is exclusive to Observer Net. Reproduction without permission is prohibited.

Original: toutiao.com/article/7586858660949738011/

Statement: This article represents the personal views of the author.