Editor's Note — Today is being called "AI Sovereignty Day" in Chinese tech circles. Alibaba's company-wide Claude ban takes effect at midnight, displacing 120,000 developers onto domestic coding tools. Ant Group drops a world-model bombshell. MiniMax is reportedly building a 2.7 trillion-parameter monster. And in the robot world, AgiBot's 15,000th humanoid rolls off the line. The throughline: China's AI ecosystem is accelerating on every axis — sovereign tools, sovereign models, sovereign hardware, sovereign infrastructure — simultaneously.
The shift from "catching up" to "sovereign capability" marks a structural turning point. Chinese enterprises are no longer just consuming Western AI — they're building a parallel stack that competes on cost, scale, and specialization. For anyone tracking the global AI race, this is the story to watch.
Track 1: Foundation Models
Chinese AI startup MiniMax is developing an LLM with 2.7 trillion parameters, according to Reuters. If released, it would be the largest open-weight AI model from a Chinese firm and possibly the largest in the world. The push toward trillion-parameter models is driven by demand for autonomous systems capable of complex, multi-step reasoning without human intervention.
DeepSeek V4 launches with a 1M token context window (up from 128K) and 40-60% throughput improvement. Most notably, it introduces China's first time-of-use pricing for LLM APIs — off-peak tokens at just 0.5 RMB/million. The Pro-tier API has been permanently cut to one-quarter of its original price.
CNBC reports that Chinese AI models are seeing rapidly expanding adoption among US enterprises, driven by 60-90% price advantages below comparable American frontier models. Data from OpenRouter shows Chinese model token share peaking at 46% among US enterprise users. US startup Lindy estimates savings of millions of dollars after switching to DeepSeek.
Track 2: Enterprise & Infrastructure
Tencent's new Hunyuan Hy3 features a 295B total parameter MoE architecture with 21B active parameters and a 256K context window. Free agent functionality is rolling out to Tencent's Yuanbao app and WeChat Work, targeting industrial automation and office productivity.
The company-wide ban on Anthropic's Claude takes effect today, pushing 120,000 developers onto domestic alternatives including Tongyi Lingma. This is the largest single forced migration in enterprise AI history and a landmark moment for China's AI sovereignty push.
⏱️ China AI in 5 Minutes, in your inbox
We read China's AI news so you don't have to. One five-minute briefing, every weekday morning — free.
No spam. Unsubscribe anytime.