State Council's First-Ever Support for AI Procurement: A Fundamental Shift

Honestly, I was a bit surprised when I saw this news.

On April 21, the State Council issued the Opinions on Promoting the Expansion and Quality Improvement of the Service Industry. In the chapter on accelerating software and information services innovation, it clearly stated: Deeply implement the AI+ action, accelerate the development and use of intelligent programming tools, support the procurement of large models and intelligent agent services.

This is the first time the State Council has included large models and agents in the government procurement support list in an official document.

The Weight of This First-Ever

Many might think: Isn’t this just writing a few sentences? How much impact can it have?

It’s really not that simple. Previously, China’s AI industry policy was mainly focused on encouraging independent R&D and technology breakthroughs. But procurement — the government directly paying for AI services — had never been clearly stated before.

This State Council document is essentially sending a signal to governments at all levels and state-owned enterprises: you can buy these now, and priority should go to large models and agents.

The Logic Behind the Policy

Why issue this policy at this particular time?

First, after years of development, AI large model basic capabilities have become relatively mature. Continuing to simply pour money into R&D has diminishing marginal returns.

Second, upgrading the service industry is currently a key focus of the economy. Large AI models and agents are essentially providing intelligent infrastructure for the service sector. The government leading by example in procurement is setting an示范 for the market.

Third, in the context of China-US AI competition, China needs to cultivate its own AI industry ecosystem. Once the government, as the largest buyer, begins procuring domestic large models, the pull-through effect on the entire industry will be enormous.

Impact on the AI Industry

The most direct beneficiaries should be companies with the capability to do enterprise-level AI products.

For the AI programming tools track, this is also positive — the document specifically mentions accelerating the development and use of intelligent programming tools.

Of course, for AI entrepreneurs, this also means increased competitive pressure. Large factories with state backing will have more advantages in government and SOE markets.

My Take

Honestly, what strikes me most about this policy is the timing.

April 2026 is precisely when domestic large model capabilities are rapidly improving. Kimi K2.6, Zhipu GLM series, Baidu Wenxin, Tongyi Qianwen… these domestic models can already compete with GPT-4o.

Issuing the procurement support policy at this timing shows the government was waiting for the right moment — waiting until domestic model capabilities caught up before pushing procurement policies. This sense of rhythm is well-calibrated.