---
title: "China's \"AI+\" roadmap just set 2027 and 2030 adoption bars for smart terminals"
date: 2026-10-09
category: Policy & Governance
site: NeuroAI
canonical: https://neuroai.site/a/na-policy-ai-plus-action-opinion
language: en
---

# China's "AI+" roadmap just set 2027 and 2030 adoption bars for smart terminals

> The State Council's August 2025 Opinion on Deeply Implementing the "AI+" Action lays out staged targets to 2035 across six domains, with intelligent terminals and agents targeted at 70% by 2027 and 90% by 2030.

A factory floor, a hospital clinic and a city traffic center are about to be told the same thing: bring AI into the core of the work, not just the edges. China's cabinet turned that instruction into a national roadmap with dates attached.

## The document and who issued it

On 26 August 2025, the State Council (国务院) published the *Opinion on Deeply Implementing the "AI+" Action (关于深入实施"人工智能+"行动的意见)*, document number 国发〔2025〕11号, dated 21 August 2025. It is the highest-level framing of the "AI+" concept that first entered the government work report in 2024.

The word "action" is deliberate. The opinion is less about research leadership and more about forcing AI into the everyday machinery of the economy.

## The staged targets

The opinion sets three checkpoints:

- **By 2027** — AI is deeply integrated with six key domains; new-generation smart terminals (智能终端) and agents (智能体) reach an adoption rate above **70%**; the smart-economy core industry grows fast.

- **By 2030** — AI fully empowers high-quality development; smart terminals and agents pass **90%** adoption; the smart economy becomes a major growth pole.

- **By 2035** — the country fully enters a smart-economy and smart-society stage.

The 70%/90% figures are the headline hooks, but the harder requirement is the breadth: six domains, not one.

## The six domains

The opinion deploys six key actions:

- **"AI+" science and technology** — AI-driven research, scientific large models (科学大模型), smarter national research platforms.

- **"AI+" industry** — full-chain industrial intelligence, smart agriculture with agricultural drones and robots.

- **"AI+" consumption upgrade** — smart homes, connected vehicles, AI phones and computers, AI wearables, new forms blending with the metaverse and brain-computer interface (脑机接口).

- **"AI+" people's wellbeing** — assisted diagnosis, health management, medical insurance services.

- **"AI+" governance capability** — smarter public infrastructure and environmental governance.

- **"AI+" global cooperation** — shared AI governance frameworks.

## Eight support pillars

Behind the domains sit eight enabling capabilities: model capability, data supply innovation, intelligent-computing coordination (智能算力统筹), application environment, open-source ecosystem, talent, policy and legal safeguards, and security. The data-supply line is notable — it encourages value-based compensation and revenue sharing for data, and supports data labeling (数据标注) and data synthesis as industries.

## Why the terminal number matters

Targeting smart terminals and agents at 70% by 2027 is a demand-side policy. It tells device makers, model vendors and app developers that the state expects AI-native hardware and autonomous agents to become default, not novel. That pulls compute, models and applications into a single coordinated push rather than leaving them as separate markets.

## Honest limitations

The adoption percentages are official policy targets, not current measured penetration; independent baselines for 2025 terminal/agent usage are not uniformly published. The opinion is a directional and coordinating document — its force comes from how ministries and local governments translate it into concrete rules and procurement. This piece summarizes the official text; it is not a forecast of market size.

## What readers can do now

- **Track the six domains for procurement signals.** If you sell or build AI tools, the listed domains (industry, health, governance, consumption) are where state-backed demand will concentrate.

- **Prepare for agent-era interfaces.** The 70%/90% agent target means product teams should design for autonomous-agent interaction, not only chat windows.

- **Watch the eight pillars for openings.** Model evaluation systems, data-supply incentives and open-source support are where secondary policies and funding will appear first.

---

Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-policy-ai-plus-action-opinion
Free to quote with attribution and a link to the original.