Chinese AI projects worth over $3.2 billion specify restricted Nvidia chips, according to a Reuters investigation. The figure covers whole projects, not confirmed chip purchases or deliveries.
Reuters examined more than 150 procurement notices, investment plans and corporate records for Chinese AI data centres. They include a proposed $873 million computing centre in Fujian that specifies Nvidia H100 servers but is still seeking financing partners.
Restrictions on sales of advanced AI chips to China were introduced by the United States in 2022. H200 export applications moved to case-by-case review in January 2026, but sales still require licences and H100 shipments remain more tightly restricted.
For businesses planning AI computing capacity, the project pipeline does not establish how much hardware is available. The investment totals include buildings, electricity and other infrastructure, while switching from Nvidia to alternative processors can require software changes and additional engineering.
It remains unproven that developers have obtained all the processors specified, so the headline investment figure is not a measure of available AI capacity.
Builders considering alternative chips need to account for software migration and engineering work, not just hardware access.
Projects still seeking financing should not be treated as committed infrastructure when planning access to computing capacity.
Written automatically by the Made in 10 Newsroom from the source above.
Sierra has released Personal Agent Protocol (Poppy) 1.0, with Shopify enabling it for every store. The standard controls how AI agents identify themselves and get permission to act, but its specification remains a draft.
Google is adding a Google Wallet connection for every US user of the Gemini app. Access is read-only, but Google may train AI on transaction summaries if users keep Gemini activity on.
LangChain has raised a $300 million Series C funding round led by CapitalG. The funding values the AI agent software company at $3 billion.