China launches 3-year plan to back 10,000 new AI SMEs

Beijing’s new AI SME plan is industrial policy at scale. Rather than focusing solely on a handful of national champions, China is explicitly trying to seed thousands of smaller AI companies across application domains, data services and compute infrastructure. That broad base matters because AGI‑relevant innovation is increasingly coming from tooling, infrastructure and niche applications, not just from training ever‑larger frontier models.

Strategically, this creates a wide pipeline of firms that can adopt, fine‑tune and stress‑test models like GPT‑6‑class systems in real sectors from manufacturing to healthcare. It also underwrites domestic alternatives to Western software and cloud providers, which is critical given export controls and geopolitics. The inclusion of national incubators, service platforms and support for the AtomGit open‑source AI community suggests China wants not only commercial wins but also influence over the global open tooling stack.

For the AGI race, the implication is that China is doubling down on ecosystem depth. Even if Chinese labs lag the absolute frontier by a model generation, thousands of well‑supported SMEs can push usage, data collection and hardware deployment in ways that narrow the gap. That makes it more likely that any meaningful AGI‑adjacent breakthroughs diffuse quickly into Chinese industry.