China’s AI Hierarchy Shifts from Four Dragons to Six Little Tigers

China’s artificial intelligence hierarchy has undergone a dramatic transformation as a new generation of generative AI startups overtakes the previous industry leaders. Known as the Six Little Tigers, this elite group has displaced the aging Four Little Dragons by pivoting from computer vision toward large language models and multimodal research. These newcomers have already secured multi-billion dollar valuations and heavy backing from tech giants like Alibaba and Tencent.

The original titans, SenseTime, Megvii, CloudWalk, and Yitu, rose to prominence through security and surveillance applications during the early vision era. Today, they face intensifying pressure from the new vanguard consisting of Zhipu AI, MiniMax, Moonshot AI, Baichuan Intelligence, 01.AI, and StepFun. This transition marks a fundamental shift from project-based revenue to scalable AI infrastructure and subscription models.

Zhipu AI has emerged as a frontrunner for government and enterprise solutions, while MiniMax and Moonshot AI focus on consumer-facing super-apps. Baichuan Intelligence, founded by former Sogou CEO Wang Xiaochuan, leverages open-source models for specialized medical and vertical sectors. Meanwhile, 01.AI, led by industry veteran Kai-Fu Lee, and the multimodal specialist StepFun, sometimes referred to as Leap Star, round out the group with significant technological breakthroughs.

This software evolution has sparked a massive windfall for domestic hardware providers according to recent industry reports. Cambricon Technologies reported a 4,230 percent revenue surge in early 2025, driven by the explosive demand for chips capable of training these massive domestic models. The company is currently raising nearly 5 billion yuan to further enhance its computing power infrastructure to meet this demand. While the Six Little Tigers currently dominate the spotlight, the landscape remains volatile and open to new challengers. Emerging players like DeepSeek and specialized multimodal firms are expected to join the unicorn ranks as the industry matures further. Analysts anticipate a period of rapid consolidation where only the most computationally efficient firms will survive the staggering costs of model development.