China has significantly closed the gap in AI development with the United States, now just three months behind in certain areas, as companies like DeepSeek have improved chip utilization and algorithm application, said Lee Kai-fu, CEO of the Chinese startup 01.AI. Lee highlighted that DeepSeek's AI reasoning model, developed using less advanced chips, demonstrated China's progress and undercut the belief that U.S. sanctions were hindering China's advancement in AI. Despite facing challenges due to Washington's semiconductor sanctions, Lee described how Chinese firms innovated under constraints, particularly in refining algorithms. He emphasized that DeepSeek's advancements in reinforcement learning technology showcased China's catching up and potential for innovation. Lee acknowledged the surge of Chinese tech companies, including 01.AI, ZhipuAI, and Moonshot, alongside established giants like Baidu, Alibaba, and ByteDance, in the development of AI models. Investing in proprietary models had become a bold move for AI startups in the competitive landscape dominated by well-funded tech giants and evolving open-source solutions. Lee disclosed that 01.AI will concentrate on practical AI applications, offering software solutions to enhance clients' deployment of foundational models. The company recently launched Wanzhi, a platform aiding enterprises in implementing AI technology, with projections for substantial growth.