China's 'AI Plus' Initiative, Translated for Smaller Businesses

On August 26, 2025, Xinhua, China's state news agency, released the full text of the State Council's Opinions on Deepening the Implementation of the "AI Plus" Initiative (State Council Document No. 11 of 2025). The policy had been adopted at a State Council executive meeting on July 31 and formally issued on August 21. It is a high-level document, long, and written in the language of industry and governance. Three weeks on, interpretations abound; this piece answers just one question — what should a company of fifty or a few hundred people take from it?

The skeleton first: three dates, six directions

The document is essentially a set of targets plus a set of actions. The targets come in three stages: by 2027, adoption of new-generation smart terminals, AI agents and similar applications is to exceed 70%; by 2030, to exceed 90%, with the intelligent economy becoming a major growth engine; by 2035, China is to fully enter a new stage of intelligent economy and society. The actions run in six directions — "AI Plus" science and technology, industrial development, consumption upgrading, public wellbeing, governance capacity and global cooperation — resting on eight supports, from model capabilities, data supply and computing power to the application environment, the open-source ecosystem, talent, regulation and safety.

Most clauses address industrial ministries, local governments and large institutions. Read through the filter of a smaller company's concerns, though, and three places touch you directly.

Touchpoint one: industrial development does not size-gate intelligence

The industrial development action encourages companies — wherever conditions allow — to fold AI into strategic planning, organisational structure and business processes, and explicitly pushes services industries toward intelligence-driven service models, promoting smart terminals and AI agents across software, information services, finance, commerce, law, transport, logistics and trade. Look at that list again: it is precisely where small and medium businesses live. The policy does not frame intelligent transformation as a big-company privilege — "fold AI into business processes" reads the same for a 50-person trading firm as for a conglomerate.

Touchpoint two: consumption upgrading is a demand-side signal

The consumption action calls for new intelligent service scenarios, naming everyday sectors — entertainment, e-commerce, housekeeping, property services, mobility, elder care, childcare — and for new service entry points such as AI assistants. For consumer-facing businesses this is a signal about demand: the way customers reach you and deal with you is being nudged toward intelligent interfaces. Today's booking call or group-buy page may increasingly become one conversation with an assistant — and whether your services can be understood and surfaced by those new entry points will become a practical question.

Touchpoint three: the wellbeing chapter is about how your team works

Inside the public wellbeing action sits a line that is easy to skim past: creating more intelligent ways of working — exploring human-machine collaborative organisational structures and management models, and strongly supporting AI skills training. This clause is not addressed to an industry; it is addressed to every organisation. It effectively promotes "employees who can work with AI" from an internal efficiency topic to a national capability-building direction. Smaller companies, which rarely have a dedicated digital team, lean on ready-made tools and external training — and that supply is now likely to grow.

Why the three-stage targets deserve to be taken seriously

Policy targets often get read as slogans. This one merits a second look: the adoption rate of AI agents and smart terminals has been written down as a quantified national target — over 70% by 2027, over 90% by 2030. Setting the metric on the application side, not the technology side, means the state wants intelligent applications running through daily business the way electricity does — not a handful of star models. In other words, agents and intelligent applications are a decade-scale direction, not a passing wave. No company needs to sprint, but planning with this as a long-term background variable is the rational move.

Getting practical: three things to do now

First, watch local implementation rather than rereading the central text. The State Council document sets direction; the money usually arrives through local programmes. Shanghai, for example, committed in its March 2025 service-sector measures to issue computing-power, model and corpus vouchers each year, letting companies use intelligent computing and corpus resources at a discount. Whether your own city runs similar schemes, and whether your industry authority keeps a pilot list, deserves more attention than the original prose.

Second, translate "intelligent transformation" into your own language. In a smaller company it starts far more plainly: which step carries the most repetitive labour? Where do errors cluster? At which stage do customers wait longest? Answer those and the entry scenario picks itself — the method laid out in which scenario your first AI workflow should target.

Third, do the digital homework first. "AI Plus" presupposes something to add AI to: a business whose processes live on paper and in chat groups, and whose data sits scattered across personal laptops, gives AI nothing to grip. Rather than fret about missing the policy window, clear the common pitfalls of small-business digitalisation and level the foundation. Unglamorous, but it decides how well everything after it lands.

A note of restraint

Over-reading is a risk too. The Opinions are a directional document: they do not mean subsidies land tomorrow, and they are no reason to launch an AI project you cannot justify; the adoption targets are national-scale, not a KPI for any single firm. From an implementation viewpoint, the judgment we care about never changes: in your specific processes, does this technology pay for itself? Policy warms the water — whether and when to swim still depends on your own strength.