In January 2025, a Chinese AI company called DeepSeek released its R1 model — and the AI world stopped. The model performed at a level comparable to the best American systems from OpenAI and Anthropic, but had reportedly been trained at a dramatically lower cost, using fewer advanced chips.
The implications were staggering. If true, DeepSeek's achievement suggested that the massive capital investments being made by American AI companies weren't the only path to frontier AI. Stock markets reacted immediately — Nvidia's stock dropped nearly 17% in a single day, the largest single-day market cap loss in U.S. history.
The Export Control Question
DeepSeek's achievement raised uncomfortable questions about the effectiveness of U.S. export controls on advanced semiconductors. If Chinese companies could develop competitive AI with restricted chips, were the controls achieving their goals?
The Broader Race
The AI race is genuinely global. DeepSeek was a reminder that American technological leadership can't be taken for granted — and that the next chapter of this story hasn't been written yet.