Friday, August 28, 2026
DarkSubscribe
AI Infrastructure · News & Analysis
HomeChips & HardwareReport
Chips & Hardware · Report

China is permitting small batches of Nvidia H200 chips to enter the mainland market to help domestic AI firms maintain pace with U.S. competitors.

Controlled access to H200 accelerators extends the runway for Chinese LLM training without triggering immediate export-control escalation, keeping global demand forecasts intact.
Trade pressSlicast · August 20, 2026 · Global · Source: The Decoder
importance 85

China is permitting limited shipments of Nvidia’s H200 chips to the mainland to help domestic AI firms compete with their U.S. counterparts. According to the Financial Times, ByteDance and Tencent each received around 10,000 H200 processors, and more companies could follow soon. The H200 is at least two generations behind Nvidia’s most powerful chips, which China cannot purchase due to U.S. export controls.

The U.S. allows up to 100,000 H200 units per company, but Beijing aims to support domestic manufacturers like Huawei. Shipments to Hong Kong are also permitted, but the city lacks sufficient data centers and power infrastructure. Additionally, all purchases require sign-off from the NDRC planning agency.

Chinese laboratories are making rapid technical progress. Recent milestones include Moonshot’s K3, Alibaba’s Qwen3.8, DeepSeek’s DeepSeek V4, and most recently, Z.ai’s GLM-5.3. Despite these advancements, domestic providers still lag far behind U.S. competitors in available inference capacity. Following a surge in demand, Moonshot was forced to turn customers away. Nvidia currently maintains a stockpile of approximately half a million H200 chips.

Read the original
China is permitting small batches of Nvidia… · Slicast