China’s state-backed Global Times has criticized Anthropic CEO Dario Amodei’s recent safety essay, characterizing his three-step pacing plan as part of a "Cold War playbook." Beijing's commentary argued that the proposals, while framed around safety risks, are intended to curb China's artificial intelligence development through technological barriers and regulatory monopolies.
Veteran tech policy observer Xiang Ligang labeled the rhetoric hostile and groundless, while Tsinghua University’s Xiao Qian suggested the stance stems from commercial pressure driven by affordable Chinese open-source models. The White House has previously flagged alleged model distillation involving Moonshot AI. In his essay, Amodei called on Washington to crack down on distillation and tighten security surrounding model weights.
The debate has drawn varied responses across the U.S. technology sector. OpenAI CEO Sam Altman stated that his company would match Anthropic’s pledge to permit third-party evaluators inside its labs, and Elon Musk expressed support on social media. Conversely, Nvidia, Meta, and Microsoft cautioned against placing restrictions on open-weight AI. Meanwhile, speaking to reporters in Doonbeg, Ireland, Donald Trump rejected any AI slowdown, stating that whoever wins artificial intelligence wins overall.
China’s Ministry of Foreign Affairs posted on social media advocating for an open and inclusive approach to AI, urging global parties to abandon fear-mongering and confrontation ahead of potential diplomatic discussions.
The policy clash carries direct implications for financial markets and asset pricing. Anthropic is currently in the initial public offering pipeline, with pre-IPO markets and futures trading active, while valuation discussions have reached substantial figures. Tighter chip and compute policies serve as a tailwind for U.S. closed-model laboratories, and the U.S. government has cleared Anthropic’s Mythos rollout for selected organizations.
Anthropic also holds a 20-year, approximately $19 billion contracted lease through a TeraWulf data-center deal. At the same time, DeepSeek is reportedly eyeing its own IPO to compete with OpenAI and Anthropic, signaling that Beijing does not intend to slow its technological growth.
