In August, China’s State Council released its opinions on implementing the national “AI Plus” initiative, laying out a development blueprint through 2035.
From policy positioning and development goals to priority tasks, the document signals a shift in China’s artificial intelligence strategy. AI is no longer framed solely as a tool for industrial upgrades but as critical infrastructure and a core driver of productivity in the country’s modernization efforts.
The plan outlines a three-step roadmap:
- By 2027, AI will be deeply integrated into six priority areas, with adoption rates for intelligent terminals and agents surpassing 70%.
- By 2030, adoption will exceed 90%, and the “intelligent economy” is expected to emerge as a major growth engine, extending benefits broadly across society.
- By 2035, AI is envisioned as the backbone of intelligent economic and social development, supporting China’s goal of socialist modernization.
The document suggests that within a decade, AI could become as ubiquitous as electricity or the internet, serving as baseline infrastructure.
2027: AI in six key areas
China’s near-term focus is on shifting from technical breakthroughs to practical applications across six domains: science and technology, industry, consumer markets, public services, governance, and global cooperation. Intelligent terminals and agents are identified as the main carriers of adoption, with a targeted penetration rate above 70%.
Guo Tao, an angel investor and AI expert, told Yicai that these areas share clear data entry points, defined business loops, and strong technology diffusion effects, which support a rollout strategy that begins with targeted applications and then expands outward to wider adoption.
In science and technology, AI is expected to evolve into a new paradigm for research, including in the humanities and social sciences. The document calls for building scientific foundation models to accelerate nascent breakthroughs and shorten the gap between lab discovery and market application.
In industry, the plan highlights both upgrading existing sectors (manufacturing, agriculture, and services) and nurturing intelligence-native industries. These would mirror how internet-native companies such as Meituan and Didi emerged in the mobile internet era. AI-powered enterprise models, from agentic customer service platforms to automated design firms, are expected to receive policy support.
The service sector is positioned as the most fertile ground for AI, with intelligent agents and terminals such as AI-enabled smartphones, PCs, and digital assistants potentially becoming new service gateways.
In consumer markets, AI will influence both production and household services. Automotives, smartphones, computers, robots, home appliances, and wearables will increasingly integrate AI as their operating “brains.” On the service side, offerings are projected to move from digital convenience to more personalized, even emotional, engagement, with AI providing companionship or creative input.
In public services, AI will integrate into education, healthcare, cultural production, and social services, promoting personalized learning and workplace efficiency. The policy also encourages embedding “Chinese elements” in AI-generated content to bolster cultural confidence.
In governance, the plan includes advancing human-machine collaboration, building intelligent security systems, and supporting ecological transition initiatives.
In global cooperation, China presents its approach as inclusive and benefit-sharing, positioning it as a counterpoint to restrictive models elsewhere.
2030: AI drives the economy
By 2030, AI Plus is expected to expand across the entire economy, with adoption of intelligent agents and terminals surpassing 90%. AI would move from specific applications to a central role in driving economic growth.
The document cites inclusivity and broad outcome-sharing as key principles, signaling concern for equitable distribution of benefits.
Outside estimates reinforce this trajectory. Nature has reported on findings from METR, a nonprofit research institute, which proposed a “Moore’s Law of intelligent agents” whereby capabilities double every seven months. At such a pace, AI could autonomously complete tasks that now take humans weeks as soon as 2028. If penetration reaches 70% by 2027, 90% by 2030 could be a plausible outcome of exponential progress.
By 2035, AI is projected to become not just an economic driver but a cornerstone of intelligent society, influencing public services, urban governance, and personal lifestyles. The policy envisions a more efficient, accessible, and sustainable ecosystem that supports national modernization.
Four pillars: Models, data, computing, and open source
Chinese AI firms have accelerated open-source releases. All of the top 15 models on the Design Arena leaderboard now originate in China. In July, Hugging Face highlighted 33 open-source models from developers including Alibaba, Z.ai (formerly Zhipu AI), Kunlun Tech, Moonshot AI, Tencent, and StepFun.
Interconnects, a research platform, identified 19 leading domestic open-source labs, from DeepSeek to smaller academic teams producing specialized models.
The State Council’s opinions call for strengthening open-source ecosystems. Universities are urged to recognize open-source contributions in credit systems, while companies and institutions are encouraged to experiment with collaborative adoption models.
Globally, models, data, and computing are viewed as the three pillars of AI competition. By aligning open-source efforts with national policy, China aims to accelerate research, industry, and applications at scale.
The plan highlights various measures:
- For models, it calls for strengthening AI theory and architecture, improving efficiency, and developing evaluation systems for reliable iteration.
- For data, it proposes high-quality dataset construction, stronger property rights, expanded access to public content where lawful, and compensation frameworks for contributors.
- For computing, it promotes breakthroughs in AI chips and supercomputing clusters, while also supporting China’s national program to shift data demand from the more developed eastern regions to data centers in the west to balance capacity.
Stronger rules for safe AI use
The State Council’s document repeatedly stresses AI security, mentioning risks such as algorithmic opacity, hallucinations, and bias. The Mandarin word for “security” appears 12 times, underscoring its importance.
This emphasis aligns with international debates on explainable AI and responsible AI.
Several new rules were also introduced this year:
- On March 14, the Cyberspace Administration of China (CAC), the Ministry of Industry and Information Technology, the Ministry of Public Security, and the National Radio and Television Administration issued measures to regulate AI-generated content, effective September 1. All AI-generated text, images, audio, video, and virtual scenes must carry visible and invisible digital watermarks. Platforms must flag such content, apps require compliance reviews, and users must declare AI-generated posts. Altering or removing marks is prohibited.
- On March 21, CAC and the Ministry of Public Security released new rules for facial recognition technology, effective June 1. These limit use to necessary, consent-based contexts, with special protections for minors. Data must be stored locally and retained only for set periods. Facial recognition cannot serve as the sole verification method, and alternatives must be available. Collection in private spaces is prohibited, and encryption standards are mandated.
China’s first national AI strategy was released in 2017. Compared with that document, the latest State Council opinions are more targeted and systematic. They emphasize integration across six domains and address practical challenges such as fragmented applications and low open-source participation.
By tailoring strategies for each sector, the policy sets out a roadmap for embedding AI into the fabric of China’s modernization.
This article was adapted based on a feature originally written by Synced (jiqizhixin.com). KrASIA is authorized to translate, adapt, and publish its contents.