US’ Bessent warns ‘nothing would matter’ if China wins the AI race

US Treasury Secretary Scott Bessent warned on Tuesday (Sep 8) that 'nothing else would matter' if China wins the artificial intelligence race, saying 'there is no day after tomorrow if China wins at this' and that even the US defence budget would fail to protect the nation. His comments come as US agencies said Alibaba, DeepSeek, Moonshot AI and other China-based AI companies extracted billions of tokens from US frontier AI models, including Anthropic's Claude, Alphabet's Gemini and SpaceXAI's Grok, since at least late 2024, potentially with Chinese government awareness.

US’ Bessent warns ‘nothing would matter’ if China wins the AI race Comments appear aimed at US voices upset over data centre projects getting built in their backyards WASHINGTON US Treasury Secretary Scott Bessent warned that the US faces dire consequences if it loses out in the artificial intelligence race with China, highlighting the angst in Washington about its rival’s tech advances https://www.businesstimes.com.sg/companies-markets/china-narrows-ai-gap-us-open-source-shift-could-hit-valuations-george-yeo . “There is no day after tomorrow if China wins at this,” Bessent said at a Breitbart News event in Washington on Tuesday Sep 8 , indicating also that the US’ large defence budget would fail to protect the nation. “If they were to pull away from us on AI, then nothing else would matter.” The comments appear aimed at some voices in the US upset that large data centre projects are getting built in their backyards https://www.businesstimes.com.sg/opinion-features/what-americas-fight-over-ai-infrastructure-really-about . Late in August, US President Donald Trump warned that a growing public backlash to the projects in the US poses an economic threat and risks giving China the upper hand in AI. In China, many new projects are going up far from big cities in less populated regions in the north and northwest. Bessent’s remarks also highlight worries in the US about open-weight Chinese models https://www.straitstimes.com/business/companies-markets/chinas-open-weight-ai-models-to-be-spared-us-tests-sources-say that are nearly as powerful as American versions at a fraction of the cost for the user. That is happening even though the US has tried to deny China access to its most advanced tech https://www.businesstimes.com.sg/international/us-takes-step-halt-nvidia-ai-chip-shipments-chinese-firms-outside-china . Also on Tuesday, US government agencies said Alibaba, DeepSeek, Moonshot AI and other China-based AI companies “extracted billions of tokens across millions of exchanges/ requests from US frontier AI models”. These include variants of Anthropic’s Claude, Alphabet’s Gemini and SpaceXAI’s Grok since at least late 2024. Tokens are units of data processed by models during training and inference. They enable prediction, generation and reasoning. The actions were potentially taken “with Chinese government awareness”, the Federal Bureau of Investigation, National Security Agency and Cybersecurity and Infrastructure Security Agency said in a statement. Reuters has reported that the US and China were preparing to discuss AI safety risks in mid-September, talks that if they happened would come just before Chinese leader Xi Jinping meets Trump in Washington on Sep 24. BLOOMBERG Decoding Asia newsletter: your guide to navigating Asia in a new global order. Sign up here to get Decoding Asia newsletter. Delivered to your inbox. Free. Share with us your feedback on BT's products and services