Key Milestones of China in AI of 2025

2025-12-27 · Source: AI Supremacy · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation, Robotics & Autonomous Systems · Depth: Intermediate, medium

Summary

China made significant strides in AI during 2025, particularly in open-source models, AI agents, and semiconductor self-sufficiency, despite limited access to leading AI chips and less capital allocation than the U.S. Chinese open-source AI models accounted for 17.1% of global downloads, surpassing the U.S. in this metric. The country's generative AI user base reached an estimated 570 million by February 2025, doubling in the first six months of the year, driven by "super-apps" and new platforms like Doubao and DeepSeek. Advanced reasoning models such as Kimi K2 and GLM-4.7 emerged, closing the gap with U.S. frontiers. China also saw the rise of AI agentic product companies like Manus, achieving 57.7% accuracy on GAIA benchmarks, and a surge in AI chip startups going public, including Moore Threads Technology Co. and MetaX Integrated Circuits Shanghai Co. Huawei's Ascend 950/960 series and Kirin 9020 chips powered over 50% of domestic data centers, reducing reliance on Nvidia, while China also progressed in EUV lithography machine development.

Key takeaway

For CTOs and AI Architects evaluating global AI infrastructure and model sourcing, China's rapid advancements in 2025 necessitate a re-evaluation of supply chain dependencies and open-source model strategies. Your teams should consider integrating Chinese open-source models, which now lead in global downloads, and assess the increasing viability of domestic Chinese AI chips and compute infrastructure to diversify your technology stack and mitigate geopolitical risks.

Key insights

China's 2025 AI progress, driven by open-source models and domestic innovation, significantly shifted the global AI landscape.

Principles

• Open-source AI fosters rapid ecosystem adoption.
• Domestic chip development reduces reliance on foreign suppliers.
• Agentic AI models enable autonomous task execution.

In practice

• Explore Chinese open-source models like Qwen for foundational AI.
• Monitor Huawei's Ascend series for domestic compute alternatives.
• Investigate AI agent platforms for autonomous task integration.

Topics

• China AI Development
• Open-Source AI
• AI Chips
• Generative AI Adoption
• AI Agents

Best for: CTO, VP of Engineering/Data, AI Architect, Director of AI/ML, Investor, Business Analyst

Related on AIssential

• Bloomberg Tech at CES — AI compute demand is rapidly accelerating across cloud, enterprise, and consumer sectors, driving innovation in chip design and autonomous systems.
• Open Source & the Bifurcated AI Frontier — Open-source AI's frontier viability is conditional on capital, governance, safety architecture, and China's strategic posture.
• Beyond the Hype: Deep Learning Advancements in 2026 — Deep learning's future lies in a connected ecosystem of specialized, multimodal, and trustworthy AI systems.
• China Is Winning the AI Race. You Barely Noticed. — China is rapidly closing the AI gap with the US, demonstrating significant advancements across multiple key sectors.
• What’s next for Chinese open-source AI — Chinese open-source AI models offer near-frontier capabilities at a fraction of the cost, rapidly reshaping global AI innovation.
• 2025 Open Models Year in Review — Open models significantly advanced in 2025, rivaling closed-source performance and fostering a more open ecosystem.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI Supremacy.