Most national AI strategies are about building models. China's current one is about installing them.
On 30 July 2026, the Political Bureau of the CPC Central Committee, setting economic priorities for the second half of the year, called for deep implementation of the “AI+” (人工智能+) action, the development of a new form of "intelligent economy", and improvement of the AI governance system.
The phrasing matters. "AI+" is not a funding call for laboratories. It is a directive to embed AI into existing industries — manufacturing, agriculture, healthcare, transport, governance — with measurable adoption targets.
Key takeaways
- Top-level directive: the July 2026 Politburo meeting explicitly ordered deep implementation of the "AI+" action, alongside building an "intelligent economy" and improving AI governance.
- 2027 target: the action sets penetration targets for AI adoption across key economic sectors by 2027, with 70% figures cited for priority areas in policy reporting.
- Industrial base: as of June 2026, China had more than 6,600 AI companies, about 15% of the global total, spanning base infrastructure, model frameworks and industry applications.
- Factory programme: the Ministry of Industry and Information Technology reports 15 "navigator-level" smart factories, 500+ "excellent-level", and 8,200+ "advanced-level" smart factories.
- Planning frame: the 2026 government work report proposed "building a new form of intelligent economy" for the first time; the 15th Five-Year Plan outline calls for cultivating "intelligence-native" business models.
Why "AI+" is structurally different
There is a familiar pattern in technology policy: fund research, publish papers, hope for spillover. China's approach inverts it. The state identifies high-value scenarios, requires large state enterprises to open them, and then measures deployment.
The clearest example is the state-owned enterprise programme. At WAIC 2026, the State-owned Assets Supervision and Administration Commission released a second batch of 60 high-value AI scenarios and 70 industry datasets from central state enterprises. These are not speculative:
- China Baowu: an "AI furnace operator" for steel converter smelting, and an "AI chief operator" for continuous annealing lines.
- State Grid: intelligent inspection and control of distributed solar, and virtual power plant aggregation.
- China Southern Power Grid: a power meteorology large model.
- Sinopec and CNPC: AI-assisted lubricant R&D and intelligent injection-production optimisation for offshore fields.
- CRRC and China Railway Construction: rail vehicle health management and intelligent rail route selection.
Alongside this, the commission launched an open-source AI community (version 2.0) and a joint "intelligent software factory" engineering programme. The mechanism is deliberate: open the scenario, supply the dataset, let vendors compete inside it.
The honest problem it is solving
Chinese policy researchers describe the gap with unusual bluntness. Alongside the achievements, official analysis lists unresolved constraints: original and disruptive innovation capacity remains insufficient; high-end chips, core algorithms and basic software remain chokepoints; mismatched supply of high-end compute; poor circulation of data as a production factor; and a shortage of cross-disciplinary talent.
"AI+" is the response to a specific diagnosis — that China has strong deployment capability and weaker original innovation, so policy should lean into the former while the latter is addressed.
That is a rational strategy, and it has a visible cost: adoption metrics can outrun genuine productivity gains. A factory that installs an AI quality-inspection system and keeps the old manual check alongside it has "adopted AI" without transforming anything.
What to watch
The measurable question is not how many models China produces. It is whether the 8,200 advanced-level smart factories translate into measurable productivity growth, and whether the 130 state-enterprise scenarios produce reusable products or one-off pilots.
The early signs in reported cases are genuinely strong — production changeover cut from hours to minutes, clinical R&D cycles compressed by 40%, inspection labour down by thousands of hours a year. Whether that scales beyond showcase plants is the story of the next three years.
Sources: Xinhua and CAC reporting on the July 2026 Politburo meeting; SASAC releases at WAIC 2026; MIIT smart factory disclosures.