DeepSeek and Open-Source AI

DeepSeek's R1 model, released in January 2025 under an MIT open-source license, stunned the global technology community by matching top Western AI models like OpenAI's o1 and Anthropic's Claude at a fraction of the cost — reportedly under $6 million to develop compared to the hundreds of millions spent by US competitors. The model briefly surpassed ChatGPT as the most downloaded app on Apple's iOS App Store, triggering an 18% drop in Nvidia's stock price that wiped out nearly $600 billion in market capitalization in a single day, as investors feared that efficient Chinese AI could reduce demand for expensive GPU hardware. DeepSeek, founded by quantitative hedge fund manager Liang Wenfeng, demonstrated that innovative architectural approaches like mixture-of-experts and novel training techniques could dramatically reduce computational requirements. Chinese AI models now account for approximately 15% of global market share, with other domestic players like Zhipu AI, Baidu's ERNIE, and Alibaba's Qwen also releasing competitive models. The open-source approach was strategically significant, as it allowed developers worldwide to build on Chinese AI infrastructure, expanding China's influence in the global AI ecosystem. For international observers, DeepSeek's success matters because it demolished the assumption that US export controls on advanced chips would cripple Chinese AI development, instead revealing that resource constraints can drive efficiency innovations that reshape the competitive landscape entirely.