China’s 1-gigawatt AI data center runs on domestic chips, not Nvidia
Z.AI, a Beijing-based AI company, has switched on a 1-gigawatt AI data center designed to train its GLM (General Language Model) using domestically produced semiconductors, with no Nvidia chips involved. The 1-gigawatt data center is positioned as a milestone in China’s effort to expand AI infrastructure despite US export controls that since 2025 have targeted high-end Nvidia GPUs such as the H100 and newer successors.
The article notes that Alibaba has also been ramping AI compute, launching an AI data center in April 2026 with 10,000 of its Zhenwu AI chips. Z.AI’s facility is described as larger in scale and ambition.
Policy support is central to the outlook. China’s government has proposed a five-year AI data-center plan with an estimated $295 billion budget, aiming for 80% domestic technology sourcing across interconnected facilities. New rules require domestic chips in state-funded projects, creating a demand floor for Chinese chipmakers and accelerating iteration and cost reductions.
For markets, the key “watch metric” is not only whether Chinese chips match Nvidia’s benchmarks on paper, but whether AI models trained on domestic hardware can compete with those trained on Nvidia-based stacks—an area where prior progress (e.g., DeepSeek’s efficiency gains) suggests meaningful headroom.
Neutral
This is primarily a macro/tech-industry development rather than a crypto-specific catalyst. Z.AI’s 1-gigawatt AI data center running on domestic chips highlights China’s push for AI compute self-reliance under Nvidia export controls, but it does not directly change token demand, protocol usage, or crypto market liquidity on its own.
Short term, traders may react with mild “AI theme” sentiment—similar to how markets sometimes briefly price headlines about major AI infrastructure spend or large GPU supply chain shifts. However, the linkage to crypto usually remains indirect (sentiment toward AI-related equities/ETFs analogs, and speculative flows into AI narratives) unless the news clearly points to measurable changes in on-chain activity.
Long term, the spending plan ($295B over five years) could reinforce China’s compute capacity and indirectly influence AI model competitiveness globally. That can affect broader risk appetite and technology valuations, which can spill over to crypto during market-wide rotations (bullish when overall liquidity expands). But given the lack of direct crypto linkage in the article, the expected effect on price stability is limited—hence a neutral stance.