
Reuters tally... Check projects using H100, H200, H800, etc.
AI-generated summary
Following the resignation of former Nvidia CEO Jensen Huang, attention is focused on the future of semiconductor export regulations to China. Reuters reported that the total size of projects and equipment contracts was more than 23 billion yuan after examining about 150 data including procurement announcements, investment plans, construction contracts, and corporate disclosures in China. Former CEO Hwang has advocated for easing export controls to China, and has said that Nvidia's market share in China fell from 95% to 0% last year.
Reuters' own tally... "Confirm the project 'using H100, H200, H800, etc.'"
(Shanghai = Yonhap News) Correspondent Cha Byeong-seop = With the resignation of former Nvidia CEO Jensen Huang, attention is focused on the direction of China's semiconductor export regulations, and foreign media reports show that the scale of artificial intelligence (AI) projects in China that rely on Nvidia chips subject to export controls by the United States is worth more than 23 billion yuan (about 4.6 trillion won).
Reuters reported on the 13th (local time) that the total size of projects and equipment contracts was more than 23 billion yuan as a result of examining about 150 data including procurement announcements, investment plans, construction contracts, and corporate disclosures in China.
However, the tally includes not only NVIDIA chips, but also the costs of buildings, power systems, and infrastructure facilities, and includes not only completed projects but also those in which investments are planned or contracts have been signed.
Reuters pointed out that various parts of China are pursuing AI data center projects related to NVIDIA's H100, H200, and H800, whose exports to China are restricted. Among these, H200 has been available for export to China since early this year if it passes individual screening by the U.S. authorities.
Reuters reported that among the projects are not only Chinese technology giants, but also local government-related ventures, universities, and local government computing centers.
For example, an AI data center project in Fujian Province worth 6.2 billion yuan (about 1.2 trillion won) explained that it specified that it would use the H100 chip while looking for a partner.
Reuters has revealed that NVIDIA has been virtually blocked from selling in the Chinese AI data center market due to the US government's export controls, but evaluated that this investigation shows the connection between China's AI expansion and NVIDIA technology.
Former CEO Hwang has advocated easing export controls to China, and last year said Nvidia's market share in China fell from 95% to 0%. Nvidia also excluded data center computing sales in China from its earnings forecast due to regulatory uncertainty.
Reuters pointed out that as the U.S. Congress is pressuring to strengthen semiconductor export controls to China, ongoing projects are exposed to risks such as delays in the supply of Nvidia chips or supply chain disruptions.
Sunny Cheong, an analyst at the Jamestown Foundation, a think tank, said, “The U.S. authorities should not simply focus on who purchases semiconductors on the surface, but ultimately investigate who uses computing power and for what purpose.”
However, China's Ministry of Commerce and Nvidia did not comment on Reuters' request for comment.
AI outlook — possibilities, not facts
If controls on semiconductor exports to China are strengthened due to pressure from the U.S. Congress, AI projects relying on Nvidia chips in China may be delayed or reduced.
Likely · Within months

Space The second stage plans to place 26 Starlink satellites into orbit and return to Earth about a day later to attempt a landing.

Meta is internally testing the prediction market app ‘Takes’ with its employees. It uses virtual points ‘Cred’ instead of real money, and is being developed under the direct instructions of CEO Mark Zuckerberg. It is interpreted as a strategy to prevent the inflow and outflow of Generation Z users.

Personal information was leaked due to hacking at an AI solution company participating in the ‘Everyone’s Startup’ project of the Ministry of SMEs and Startups. It recurred five months after the first accident, and was confirmed to contain the names, phone numbers, and email addresses of 818 project participants. Participants are demanding disclosure of security responsibility, results of implementation of measures to prevent recurrence, and procedures for compensation for damages.

A letter was circulated threatening Nvidia's key executives to resign en masse and move to Space
![[Tech Talk Note] From KTX reservations to phone calls... ‘Pickle’, an AI assistant that takes over my work](/api/img?u=https%3A%2F%2Fimg.yna.co.kr%2Fetc%2Finner%2FKR%2F2026%2F11%2F13%2FAKR20261113158900017_02_i_P2.jpg&w=320&q=72&f=webp)
Personal AI assistant ‘Pickle’ provides KTX ticket reservation, restaurant reservation, and customer service call service. It was developed by a startup of the same name, founded in San Francisco, USA in 2024 by a Korean startup team, and participated in Y Combinator's 2025 Winter Program. Passwords and card numbers are encrypted and stored on the user's device, and major operations are performed after user approval.

Open AI, in collaboration with Visa, Master, and American Express, launched 'ChatGPT Wallet' to support AI agent proxy payments. It is provided first to Pro subscribers over the age of 18 in the United States, and final approval must be obtained through fingerprint or facial authentication at the time of purchase.