China's industry ministry lays out its AI application playbook for the next five years
Beijing is moving from AI slogans to a work plan: the industry ministry used its first "15th Five-Year" briefing to name exactly which AI scenarios get state backing, while one of China's best-known small-model startups told staff that the next battle isn't in the cloud at all.
China will spend the 2026–2030 plan period cultivating "AI+", "robotics+" and other flagship application scenarios as a matter of industrial policy. At a State Council Information Office briefing on August 26, officials from the Ministry of Industry and Information Technology said the country will run large-scale application demonstrations for new manufacturing technologies, products and scenarios, backed by mechanisms to select flagship projects, publish lists of new technologies and products ready for deployment, and broker supply-demand matching so innovations move through real-world testing into volume use. The same ministry disclosed it is organizing AI ethics review and service pilots across multiple cities under an April rulebook issued with nine other agencies — pairing the build-out push with governance scaffolding from day one. Read together, this is Beijing signaling that diffusion, not just frontier capability, is now the explicit goal of national AI policy.
ModelBest's co-founders marked their company's fourth birthday by telling staff that on-device intelligence will be the main battleground of the general-intelligence era. In an all-hands letter signed by CEO Li Dahai, Tsinghua professor Liu Zhiyuan and other co-founders, the MiniCPM maker framed its next phase as three battles: keep pushing efficiency gains along its self-described "densing law," scale proven vertical deployments in consumer electronics, industry and law, and rebuild the company's own workflows around AI. The numbers behind the pitch: the open-weight MiniCPM family has passed 50 million cumulative downloads, and its models already run inside car cockpits, flagship phones and even satellites. The letter explicitly rejects the ambition to become "China's OpenAI or Anthropic," positioning the startup instead as infrastructure for putting intelligence into every kind of device — a bet that mirrors, from the small-model side, what Apple's local-AI hardware play signals from the consumer end.
What to watch: whether MIIT's scenario lists turn into named projects and funding lines in the coming months — and whether ModelBest's IPO-era discipline survives the pivot from proving the model to shipping it everywhere.
Is industrial-policy muscle the fastest way to spread useful AI, or a recipe for subsidized demos that never scale? Tell us in the comments.
