In an essay on Wednesday, Dario Amodei, the CEO of Anthropic, weighed in on the debate over whether Chinese AI company DeepSeek’s success implies that U.S. export controls on AI chips aren’t working.
Amodei, who recently made the case for stronger export controls in an op-ed co-written with former U.S. deputy national security adviser Matt Pottinger, says in the essay he believes current export controls are slowing the progress of Chinese companies like DeepSeek. Compared to the performance of the strongest U.S.-produced AI models, Amodei says, DeepSeek’s fall short when factoring in the release time frame.
“DeepSeek produced a model close to the performance of U.S. models 7-10 months older, for a good deal less cost (but not anywhere near the ratios people have suggested),” Amodei said. “[This is] an expected point on an ongoing cost reduction curve. What’s different this time is that the company that was first to demonstrate the expected cost reductions was Chinese.”
Amodei compares one of DeepSeek’s flagship models, DeepSeek V3, to Anthropic’s Claude 3.5 Sonnet, which he says cost a “few $10M’s” to train. Sonnet’s training finished 9 to 12 months ago, while DeepSeek’s model was trained in November or December — yet Sonnet remains ahead in a number of “internal and external evals,” Amodei notes.
“U.S. companies [are also] achieving the usual trend in cost reduction,” Amodei added. “The efficiency innovations DeepSeek developed will soon be applied by both U.S. and Chinese labs to train multi-billion dollar models.”
Amodei, who in the essay calls DeepSeek “very talented engineers” that “show why China is a serious competitor to the U.S.,” foresees a fork in the road depending on which export policies the Trump administration embraces. Before Trump took office, the outgoing Biden administration imposed new restrictions on hardware exports that are scheduled to take effect in the coming months, but that could be curtailed should Trump wish to do so.
If Trump strengthens export rules and prevents China from obtaining what Amodei describes as “millions of chips” for AI development, the U.S. and its allies could potentially establish a “commanding and long-lasting lead,” Amodei claims. If, on the other hand, the U.S. doesn’t make it more challenging for China to import AI chips, the country could “direct more talent, capital, and focus” to “military applications” of AI technologies, Amodei fears.
“Combined with its large industrial base and military-strategic advantages, this could help China take a commanding lead on the global stage,” Amodei said. “To be clear, the goal here is not to deny China or any other authoritarian country the immense benefits in science, medicine, quality of life, and so on that come from very powerful AI systems. Everyone should be able to benefit from AI. The goal is to prevent them from gaining military dominance.”
It seems likely that Amodei will get his preferred outcome. In a Senate hearing on Wednesday, billionaire businessman Howard Lutnick, Trump’s pick for commerce secretary, accused DeepSeek of stealing American IP.
“What this showed is that our export controls, not backed by tariffs, are like a whack-a-mole model,” Lutnick said. “Chinese tariffs should be the highest.”
As commerce secretary, Lutnick would have a key role in carrying out Trump’s plans to raise and enforce tariffs.
OpenAI, Anthropic’s chief rival, has also called on the Trump administration to take more aggressive steps to ensure U.S. dominance in AI. In a recently published policy doc, OpenAI warned that if the U.S. doesn’t attract the necessary global funds for AI projects, they’ll “flow to China-backed projects” and “[strengthen] the Chinese Communist Party’s global influence.”
