Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseInsight · On Radar3DNews AI→ original

Reuters: China's $3.2B AI Data Center Buildout Relies on Nvidia Chips as Huang Steps Down

Reuters reviewed more than 150 Chinese documents and found projects worth 23 billion yuan ($3.2 billion). The data centers are being built for Nvidia H100, H200, and H800 accelerators, AI chips that fall under US export restrictions. The total includes buildings and electrical infrastructure, not just chips. Huang, who had pushed for easing the rules, stepped down as Nvidia CEO on November 12. If controls tighten, the buildouts could stall or switch to Chinese chips.

AI-processed from 3DNews AI; edited by Hamidun News
Reuters: China's $3.2B AI Data Center Buildout Relies on Nvidia Chips as Huang Steps Down
Source: 3DNews AI. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary source3DNews AI ↗
Editor-in-Chief
Key metric #1In Gansu, one facility was promised several thousand accelerators by the end of the year
Key metric #2IDC estimates Nvidia held 34% of AI accelerator shipments in China in the first half of 2026
Key metric #3A university on the US blacklist received two H200s through a local supplier

Chinese companies are building AI data centers around Nvidia graphics chips that the US is trying to keep out of the country. Now the story has numbers. Reuters reporters reviewed more than 150 Chinese documents, including procurement notices, investment plans, construction contracts, and corporate reports. They identified projects and agreements worth over 23 billion yuan, roughly $3.2 billion. All of them assume the use of Nvidia H100, H200, and H800 accelerators, server chips used to train and run large models. On November 12, Jensen Huang stepped down as Nvidia's CEO. He was one of the main advocates for easing export restrictions, so uncertainty around these projects has grown.

What the $3.2 Billion Actually Means

The figure is easy to misread. It is not $3.2 billion spent on Nvidia processors themselves. The total covers construction of buildings, electrical hookups, other infrastructure, and equipment. Think of a crypto-mining or model-training farm: the cards account for only part of the budget, while the rest goes to walls, power, and cooling. It works the same way here.

The projects are at different stages. Some facilities are already operating or under construction, while others remain at the planning stage or awaiting contract signings. The participants include not only large tech companies but also regional data centers, universities, and state-linked enterprises. In Gansu province in northwest China, one facility was supposed to receive several thousand Nvidia accelerators by the end of the year. Reporters could not establish whether the hardware reached the customer.

How Chips Still Reach China Under Restrictions

The US began restricting shipments of advanced accelerators to China in 2022. The bans were later extended to models Nvidia had designed specifically for the Chinese market. In January 2026, Washington moved to case-by-case review of H200 export license applications from buyers that meet the established requirements.

Access to American hardware has not disappeared, however. Northwestern Polytechnical University is on the US export blacklist, yet it managed to obtain two H200 accelerators through a local supplier. It also signed contracts to rent computing capacity built on Nvidia products.

The market picture is contradictory. Nvidia itself has said that US restrictions effectively cut off its ability to sell data center hardware in China. But according to IDC, the company accounted for 34% of AI accelerator shipments in China in the first half of 2026. These metrics describe different sides of the market. Taken together, they do not reveal exactly how export-restricted chips reach Chinese buyers.

Why Huang's Exit Matters

Huang advocated easing export restrictions, while US lawmakers are pushing for tighter controls on shipments of advanced chips. His departure came amid an investigation into alleged ties between Huang and businesswoman Alice Huang, whom US authorities suspect of contacts with Chinese intelligence.

An important caveat: there is no public evidence that Huang himself violated export law. The projects Reuters identified are not part of that investigation either. But if the rules tighten, some Chinese projects could face supply delays or be forced to switch to domestic processors.

What This Means for the Industry

Three conclusions stand out.

  • Demand for Nvidia in China has not disappeared, but it has become less transparent, ranging from purchases through local suppliers to rented capacity.
  • Some projects exist only on paper for now, so the actual number of chips delivered under these documents cannot be estimated.
  • If controls get stricter, domestic accelerator makers stand to gain. Regional buyers who built data centers around a single architecture stand to lose.

Those who build or rent compute in the region should check in advance whether their workloads can move to other hardware. Tying all infrastructure to one supplier while sanctions keep shifting remains the main risk behind these $3.2 billion.

Story

Key Questions Answered

What the $3.2 Billion Actually Means?

The figure is easy to misread. It is not $3.2 billion spent on Nvidia processors themselves. The total covers construction of buildings, electrical hookups, other infrastructure, and equipment. Think of a crypto-mining or model-training farm: the cards account for only part of the budget, while the rest goes to walls, power, and cooling. It works the same way here. The projects are at different stages.

How Chips Still Reach China Under Restrictions?

The US began restricting shipments of advanced accelerators to China in 2022. The bans were later extended to models Nvidia had designed specifically for the Chinese market. In January 2026, Washington moved to case-by-case review of H200 export license applications from buyers that meet the established requirements. Access to American hardware has not disappeared, however. Northwestern Polytechnical University is on the US export blacklist, yet it managed to obtain two H200 accelerators through a local supplier.

Why Huang's Exit Matters?

Huang advocated easing export restrictions, while US lawmakers are pushing for tighter controls on shipments of advanced chips. His departure came amid an investigation into alleged ties between Huang and businesswoman Alice Huang, whom US authorities suspect of contacts with Chinese intelligence. An important caveat: there is no public evidence that Huang himself violated export law. The projects Reuters identified are not part of that investigation either.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Private AI Builders Community · Vibe Club

Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews

Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.

Join Vibe Club →Explore Hamidun Academy
What do you think?
Loading comments…