In a recent essay, Dario Amodei, CEO of AI company Anthropic, has entered the debate on whether Chinese AI company DeepSeek's success implies that US export controls on AI chips are ineffective. Amodei, who has previously advocated for stronger export controls, argues that current measures are indeed slowing the progress of Chinese companies like DeepSeek.

Amodei compares the performance of DeepSeek's flagship model, DeepSeek V3, to Anthropic's Claude 3.5 Sonnet, which he claims cost a "few $10M's" to train. Despite being trained 9-12 months ago, Sonnet remains ahead of DeepSeek V3 in internal and external evaluations. Amodei notes that DeepSeek's model was trained in November or December, and attributes the gap to the ongoing cost reduction curve.

Amodei acknowledges that DeepSeek has made significant strides in cost reduction, but believes that US companies are also achieving similar efficiency innovations. He foresees a fork in the road depending on which export policies the Trump Administration adopts. If the US strengthens export rules and prevents China from obtaining millions of chips for AI development, Amodei claims that the US and its allies could establish a "commanding and long-lasting lead" in AI.

On the other hand, if the US doesn't make it more challenging for China to import AI chips, Amodei fears that China could "direct more talent, capital, and focus" to "military applications" of AI technologies. This, combined with China's large industrial base and military-strategic advantages, could help China take a commanding lead on the global stage.

Amodei's concerns are echoed by billionaire businessman Howard Lutnick, Trump's pick for commerce secretary, who accused DeepSeek of stealing American IP in a Senate hearing on Wednesday. Lutnick advocated for higher Chinese tariffs, stating that current export controls are like a "whack-a-mole model" that needs to be backed by tariffs.

OpenAI, Anthropic's chief rival, has also called on the Trump Administration to take more aggressive steps to ensure US dominance in AI. In a recently published policy document, OpenAI warned that if the US 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."

The debate highlights the complexities of balancing national security concerns with the need to promote innovation and collaboration in the AI sector. As the US and China continue to vie for dominance in AI, the implications of export controls and tariffs will be closely watched by industry stakeholders and policymakers alike.

In conclusion, Amodei's essay serves as a timely reminder of the need for nuanced and informed discussion on the intersection of AI, national security, and global competition. As the AI landscape continues to evolve, it remains to be seen how the US and China will navigate the complex web of export controls, tariffs, and innovation to achieve their respective goals.