China's tech giants move AI model training overseas to access Nvidia chips, FT reports

Nov 27 (Reuters) - Top Chinese firms are training their artificial intelligence models abroad to access Nvidia's NVDA.O chips and avoid U.S. measures aimed at curbing their progress in advanced technology, Financial Times reported on Thursday.

Alibaba 9988.HK and ByteDance are among the tech firms training their newest large language models in Southeast Asian data centres, the report said, citing two people with direct knowledge of the matter.

Reuters could not immediately verify the report. Nvidia declined to comment on the report.

According to the report, there has been a steady increase in training in offshore locations after U.S. moved to restrict sales of the H20 chip in April.

Chinese companies rely on lease agreements for overseas data centres owned and operated by non-Chinese entities, the newspaper said, noting that DeepSeek, which gathered a large stock of Nvidia chips before the US export bans, was an exception with its model being trained domestically.

Deepseek is also collaborating with domestic chip manufacturers led by Huawei, to optimise and develop the next generation of Chinese AI chips, FT added.

Alibaba, ByteDance, DeepSeek and Huawei did not respond to requests for comment.