Milestones of China in AI of 2025

The year 2025 marked a pivotal period for China's AI ecosystem, as chronicled by Tony Peng (ex-Baidu Global Head of Comms) and summarized in a global macro analysis. Despite severe restrictions on leading AI chips and a fraction of U.S. capital expenditure, China's progress was remarkable. Open-source AI models from China accounted for 17.1% of global downloads, surpassing the U.S. for the first time. Generative AI adoption exploded: the user base hit 570 million by February 2025, doubling in six months (106.6% growth) thanks to super-app integrations (Doubao, DeepSeek) and other platforms. The so-called 'Six Little Dragons of AI' matured, releasing open-weight reasoning models like Kimi K2 (Moonshot AI), GLM-4.7 (Zhipu AI now Z dot AI), and MiniMax models that closed the gap with U.S. frontiers in complex reasoning. AI agents became a major trend; Manus achieved 57.7% accuracy on high-level GAIA benchmarks, and companies like Manus and Genspark developed agentic products quickly, often targeting global audiences. On the hardware side, AI chip startups saw spectacular IPOs: Moore Threads Technology Co. jumped 425% on its first trading day in Shanghai, MetaX Integrated Circuits gained 693%, and Cambricon Technologies went public; Enflame, Biren, and Kunlunxin are expected in early 2026. Huawei's in-house Ascend 950/960 series and the return of Kirin chips (Kirin 9020) now power over 50% of domestic data centers, reducing reliance on Nvidia, with Bernstein Research predicting Ascend could reach 50% market share in 2026. Physical AI and humanoid robotics companies also emerged early on the IPO front. Meanwhile, Alibaba and ByteDance are beginning to match the capital expenditure of global AI leaders, solidifying China's role as a global undercurrent in artificial intelligence.