🇨🇳 China’s National AI Strategy

Contents

This report was prepared by GINC in mid-2025 to provide a comprehensive analysis of China’s national AI strategy, drawing on the latest policy developments, regulatory frameworks, and global positioning. The table of contents below outlines the structure of the report, organized into five thematic parts covering strategic vision, innovation systems, sectoral deployment, social impact and policy evolution over time.

Part I: National Vision and Strategic Foundations
Strategic Vision & Objectives
Governance Architecture
Policy Instruments & Incentives

Part II: Innovation System and Talent Development
R&D and Innovation Ecosystem
Talent, Education & Mobility
Data, Compute & Digital Infrastructure

Part III: Sectoral Integration and Global Influence
Industrial Deployment & Tech Diffusion
Regulatory, Ethical & Safety Frameworks
International Engagement & Standards
Defence, Security & Dual-Use Considerations

Part IV: Performance, Resilience and Social Impact
Performance Metrics & Monitoring
Strategic Foresight & Resilience
Foundational AI Capabilities
Public Trust, Inclusion & Social Equity

Part V: Evolution of China’s National AI Strategy (2015–2025)

Strategic Vision & Objectives

China’s AI strategy is ambitious, state-led, and mission-oriented, positioning AI as a transformative tool for economic, technological, and geopolitical advancement. Through top-level policy plans such as the 2017 State Council AI Plan and the 14th Five-Year Plan, the country aims to become a global AI leader by 2030. The strategy integrates innovation goals with industrial upgrading and social development priorities, framing AI as essential for both national competitiveness and social stability.

Time-Bound Leadership Targets

China’s ambition to become the global leader in artificial intelligence is clearly articulated through the State Council’s New Generation Artificial Intelligence Development Plan (2017). This document established a strategic trajectory toward achieving world-class AI capabilities by 2030, broken into three concrete phases. The first milestone, to catch up with leading countries by 2020, has largely been met through rapid industrialization, patent activity, and data infrastructure deployment. By 2025, China aims to attain global competitiveness in core technologies and build an initial framework for AI governance. The ultimate objective for 2030 is to establish China as the “primary global AI innovation center,” a goal reiterated in the 14th Five-Year Plan for Digital Economy Development and Politburo study sessions in 2023 and 2024.

Mission-Driven Innovation

The state-led nature of China’s AI strategy sets it apart from many market-driven models. AI is treated not merely as a general-purpose technology but as a tool of national strategic importance, embedded within a broader innovation framework that includes initiatives like “Made in China 2025” and “Scientific Innovation 2030”. These policies direct attention and investment toward mission-critical areas such as brain-inspired computing, intelligent robotics, and natural language understanding. The goal is not only to upgrade China’s industrial base but also to achieve technological independence in critical sectors. Government guidance funds, academic incentives, and regulatory support are coordinated to accelerate frontier innovation, allowing China to leapfrog existing paradigms and shape future global norms on its own terms.

Economic & Social Development

The economic rationale behind China’s AI push is closely tied to national goals for industrial upgrading, productivity enhancement, and digital transformation. As outlined in the State Council’s AI Plan and 14th Five-Year Plan, AI is expected to deliver efficiency gains across manufacturing, logistics, agriculture, and services—improving competitiveness while addressing challenges such as aging demographics and environmental pressures. Moreover, the strategy positions AI as a solution to societal issues, such as expanding healthcare access, improving disaster response, and delivering intelligent education in rural regions. These dual-use narratives, embedded in speeches by President Xi Jinping and policy framing from the National Development and Reform Commission (NDRC), frame AI not only as a technology of economic power but also of social stability and national rejuvenation.

Governance Architecture

China’s governance system for AI is centralized in design but decentralized in implementation. Ministries like MOST, MIIT, and CAC coordinate AI policy at the national level, while industry alliances and designated innovation platforms facilitate collaboration with tech firms. Pilot zones in cities like Beijing and Shanghai function as regulatory sandboxes, enabling regional adaptation and experimentation while feeding insights back into national policymaking.

Central Ministries and Agencies

China’s governance structure for AI is centralized yet diversified across a network of ministries, commissions, and regulatory agencies. At the highest level, strategic priorities are set by the State Council and the Central Commission for Comprehensively Deepening Reform, reflecting the importance of AI to both administrative and political planning. The Ministry of Science and Technology (MOST) oversees research policy, key technology programs, and scientific coordination. It manages national labs, funds basic research through the National Natural Science Foundation of China (NSFC), and supports innovation platforms across the country. The Ministry of Industry and Information Technology (MIIT) is tasked with industrial implementation and works on AI standardization, manufacturing integration, and corporate policy guidance.

The Cyberspace Administration of China (CAC) plays a leading role in regulating AI-related algorithms, content moderation, and personal data use. It has issued pioneering regulations on algorithm recommendation systems, deep synthesis content, and generative AI platforms. This alignment of technical regulation with ideological supervision reflects the dual function of AI as both a driver of productivity and a tool of social governance.

Public-Private Alliances

To bridge policy and practice, China has fostered several public-private consortia. Chief among them is the Artificial Intelligence Industry Alliance (AIIA), a collaborative network of over 200 enterprises, universities, and research institutes coordinated by MIIT. AIIA facilitates alignment on industrial standards, policy feedback, and the deployment of AI across key sectors. In parallel, the Open Innovation Platform Program established by MOST designates leading firms to spearhead national development in domain-specific AI subfields—such as Baidu for autonomous driving, Tencent for medical imaging, and Alibaba for smart cities.

These alliances help coordinate China’s innovation strategy at scale, encouraging knowledge-sharing, standard-setting, and public policy alignment across a fragmented and rapidly evolving ecosystem. They also serve as semi-formal governance channels, ensuring that tech giants remain aligned with state priorities while enabling government access to proprietary innovation capacities.

Local Governments and Pilot Zones

Implementation of the national AI strategy is carried out through a decentralized network of city-level and provincial actors, guided by central targets but locally customized. As of 2024, China has approved over 20 National AI Innovation Pilot Zones, including in Beijing, Shanghai, Shenzhen, Hangzhou, and Hefei. These zones are empowered to experiment with regulatory sandboxes, tax incentives, and infrastructure clustering tailored to regional strengths.

For example, Shanghai has established the Pudong AI Innovation Zone as a hub for large model training and urban AI deployment. Beijing’s Zhongguancun AI Pilot Zone emphasizes smart governance applications and supports local compute platforms. These pilot zones provide case studies and data for broader national rollout, forming a virtuous cycle of innovation and policy learning. Local governments often co-invest in computing centers and industry parks and are responsible for hosting AI demonstration projects aligned with both economic development and Party-building objectives.

Policy Instruments & Incentives

China uses a full suite of policy tools to stimulate AI development, including public investment funds, tax breaks, subsidized computing resources, and procurement mandates. National funds such as the ¥60 billion AI Industry Investment Fund and the “Big Fund” for semiconductors help mitigate R&D risk. Regulatory sandboxes empower cities to fast-track AI applications, while preferential policies target SMEs and startups to broaden participation.

Public Funding

China’s approach to AI funding is anchored in large-scale, coordinated public investment. In 2024, the government launched a new National AI Industry Investment Fund with an estimated capital base of ¥60 billion (US$8.2 billion), aimed at accelerating foundational research and strategic enterprise support. This fund complements a wider ecosystem of local and provincial guidance funds, such as those in Shanghai, Shenzhen, and Hangzhou, which co-invest with private capital in startups and frontier technologies. The National Integrated Circuit Industry Investment Fund (or “Big Fund”) provides similar support for AI-relevant chipmaking. Together, these tools help de-risk R&D, stimulate innovation, and ensure continuity amid market fluctuations.

Tax Credits and Financial Tools

To bolster investment returns, China offers generous tax policies for certified “High and New Technology Enterprises” (HNTEs), including a reduced corporate tax rate of 15% under the Corporate Income Tax Law. AI firms benefit from R&D expense deductions, tax deferrals, and direct rebates. Additionally, state-backed financial institutions such as the Bank of China have rolled out multi-year credit lines and AI-specific loan programs. In cities like Beijing and Chengdu, AI startups can access compute vouchers that subsidize GPU time on domestic clouds like Alibaba Cloud and Huawei Cloud. These instruments reduce barriers for small firms and research teams, expanding participation in AI innovation.

Government Procurement and Mandates

The Chinese state also plays a catalytic role in AI market formation through procurement. State-owned enterprises (SOEs) have been instructed by the State-owned Assets Supervision and Administration Commission (SASAC) to integrate AI into logistics, HR, and manufacturing systems. Local governments procure AI-driven urban management platforms, like Alibaba’s City Brain, and AI-based diagnostic systems approved by the National Medical Products Administration (NMPA). These contracts often include domestic sourcing clauses that ensure Chinese suppliers capture market share, promoting national champions. Public procurement thus serves not only to scale up adoption but also as a sandbox for testing and refining products in real-world conditions.

Regulatory Sandboxes and Pilot Policies

To support experimentation, regulatory sandboxes have been embedded into AI Innovation Pilot Zones across cities like Beijing, Shanghai, and Shenzhen. These zones grant temporary exemptions from national rules, allowing technologies such as autonomous vehicles or algorithmic government services to operate under controlled supervision. For example, the Beijing High-Level Autonomous Driving Demonstration Zone enables full-stack self-driving trials under policy guidance from the Ministry of Transport. These experiments inform broader policy evolution and act as testbeds for regulatory refinement, helping China maintain legal adaptability amid rapid technological change.

R&D and Innovation Ecosystem

China’s AI research landscape combines elite academic institutions, state labs, and corporate R&D centers, coordinated through national programs like the Science and Technology Innovation 2030 Agenda. Universities collaborate closely with firms like Baidu and Huawei, accelerating the translation of research into industrial products. Government-designated open innovation platforms support the standardization and dissemination of AI tools across sectors.

Research Institutions and Labs

China’s AI research landscape includes a mix of academic, state-backed, and hybrid labs. Leading institutions like Tsinghua University, Zhejiang University, and the Chinese Academy of Sciences anchor fundamental research. The Beijing Academy of Artificial Intelligence (BAAI) produces open-source foundation models (e.g. WuDao), while the Shanghai AI Lab focuses on multimodal learning and urban intelligence. In Shenzhen, Peng Cheng Lab is developing compute-intensive platforms and general AI models. These labs often operate with joint funding from central ministries and local governments, and partner with firms like Huawei, Alibaba, and Baidu.

National R&D Programs

China’s long-term S&T policy embeds AI within the Science and Technology Innovation 2030 Agenda, listing it alongside quantum and brain science as a priority megaproject. These programs are coordinated by the Ministry of Science and Technology (MOST), which allocates funding, evaluates deliverables, and organizes cross-institutional consortia. Additional support flows through the National Natural Science Foundation of China (NSFC), which funds blue-sky AI research and helps cultivate early-stage theoretical breakthroughs. AI-specific competitions and application scenarios, such as the “Intelligent Robot” and “Cognitive Computing” tracks, align academic output with national needs.

Academia-Industry Collaboration

University-industry linkages are central to China’s AI R&D model. Baidu, SenseTime, and iFlytek sponsor research chairs, fellowships, and competitions across leading universities. These firms co-author papers with academics, support open-source releases, and provide access to proprietary datasets or training compute. Through joint labs and fellowship programs—often supported by MOST—this collaboration ensures that theoretical advances are rapidly commercialized. For example, the Baidu-Tsinghua joint lab has published influential work on ERNIE models and multilingual NLP benchmarks. This symbiosis boosts academic citations while aligning research with industrial relevance.

Innovation Platforms and Open Infrastructure

China’s Open Innovation Platform initiative, launched in 2017 under MOST, delegates specific firms to lead R&D and ecosystem development in vertical domains. Baidu manages the national autonomous driving platform, Tencent handles medical imaging, Alibaba leads smart city systems, and iFlytek oversees intelligent speech. These platforms offer standardized APIs, curated datasets, and compute services to startups and research institutions. For example, OpenI (co-led by BAAI) provides shared LLM repositories, evaluation benchmarks, and collaborative development environments, accelerating ecosystem-wide experimentation.

Talent, Education & Mobility

China is building a deep AI talent pipeline through expanded university programs, K–12 STEM initiatives, adult reskilling policies, and global recruitment. Over 345 universities now offer AI degrees, while online platforms and incentive programs retrain millions in key industries. Through initiatives like the Thousand Talents Plan and AI youth competitions, China blends local capacity building with global talent circulation to drive long-term workforce growth.

Higher Education Reform and Capacity Expansion

China’s Ministry of Education launched the AI Innovation Action Plan for Colleges and Universities in 2018, which seeded AI programs at over 345 institutions. Universities like Peking University and Shanghai Jiao Tong University offer interdisciplinary degrees in AI theory, robotics, and computational neuroscience. The Ministry supports the establishment of “AI+X” dual-degree programs, combining AI with medicine, law, and agriculture. These programs are tasked with producing both academic researchers and applied technologists to meet growing national and industry demand. By 2022, China trained more AI-related graduates than any other country, according to Stanford's AI Index.

K-12 Integration and STEM Promotion

STEM instruction has been introduced earlier in China’s national curriculum through coding courses, robotics contests, and AI-themed textbooks. Local governments in Beijing, Guangdong, and Zhejiang have rolled out AI pilot curricula at the elementary and secondary levels, supported by AI-focused education companies like Squirrel AI. The Ministry of Education also supports the National AI Youth Challenge as part of science popularization, reinforcing AI as a patriotic and aspirational career pathway. These initiatives are designed to normalize AI as a literacy skill and foster interest in computational thinking from a young age.

Reskilling the Workforce

The Internet Plus Vocational Education Plan supports online and blended learning platforms such as XuetangX and iCourse. These services offer AI and data science certificates for workers in logistics, manufacturing, and public administration. The National Skilled Talent Development Outline (2021–2025) sets targets for re-skilling tens of millions of workers, with subsidies and employer incentives for AI upskilling. Local HR departments partner with major platforms and SOEs to deliver customized modules, ensuring that digital transformation does not widen social inequality.

International Talent Recruitment

Through programs like the Thousand Talents Plan and Changjiang Scholars Program, China continues to recruit foreign-trained scientists and entrepreneurs. These programs offer competitive research grants, housing allowances, and lab setup support. Cities like Shanghai have launched international innovation hubs such as the Zhangjiang AI Island to host foreign startups and joint labs. Meanwhile, Chinese firms including Huawei and ByteDance maintain R&D outposts in Toronto, Seattle, and Paris to scout and co-develop with global AI talent.

Diaspora Engagement and Brain Circulation

China encourages “reverse brain drain” by engaging the global Chinese diaspora through visiting fellowships, conference invitations, and hybrid appointments. Many of the country’s leading AI researchers are returnees from top U.S. institutions like Stanford, MIT, and Carnegie Mellon. The annual World Artificial Intelligence Conference (WAIC) in Shanghai serves as a platform to showcase their work and connect them with domestic funding. National labs and universities also offer tenure-track packages for returnees with support from central and local governments. This policy mix has turned China into a global hub for “brain circulation,” combining inbound recruitment with outbound collaboration.

Data, Compute & Digital Infrastructure

China is investing in a sovereign AI infrastructure stack, including data governance frameworks, national compute clusters, and domestic chip ecosystems. The creation of the National Data Administration and the expansion of “Eastern Data, Western Compute” are reshaping access to computing resources and public datasets. Platforms like PaddlePaddle and MindSpore support a shift away from foreign dependencies, enhancing China’s digital autonomy.

National Data Resources and Governance

China’s AI ecosystem is underpinned by vast data resources, which the government has increasingly formalized into a strategic national asset. Recognizing that data is the “new oil” for AI, the state has designated it a factor of production, on par with land and capital, in official economic planning documents. The establishment of the National Data Administration (NDA) in 2023 marked a turning point in national data governance. Its mandate includes integration of public datasets, oversight of cross-border data flows, and regulation of data transactions.

Urban data exchanges in Shenzhen and Shanghai allow enterprises to buy and sell anonymized datasets—ranging from traffic flow statistics to retail behavior—under strict compliance protocols. These exchanges operate under the framework of the Data Security Law and the Personal Information Protection Law, both of which impose stringent localization and consent requirements. While private data sharing remains cautious due to competitive concerns, government-led data pools for health, agriculture, and education provide structured access to researchers and startups.

Compute Power and Cloud Infrastructure

China has invested aggressively in compute infrastructure, viewing it as essential to both AI training and digital sovereignty. Through the “Eastern Data, Western Compute” initiative, the country is developing a network of national supercomputing hubs, which concentrate compute-intensive workloads in less densely populated western provinces where energy is cheaper and land more available.

By mid-2024, China had reached an estimated 246 EFLOPS of AI computing power, placing it second globally behind the United States. Much of this capacity resides in national AI centers, state-supported cloud platforms, and university supercomputing clusters. Chinese cloud providers such as Alibaba Cloud, Tencent Cloud, and Huawei Cloud offer scalable infrastructure for model training and deployment. These firms are increasingly using domestically developed chips—such as Huawei’s Ascend and Baidu’s Kunlun processors—in line with national targets for reduced reliance on U.S. semiconductor technology.

Sovereign Compute and Localization

As part of its digital sovereignty strategy, China is working to indigenize its AI hardware and software stack. Export controls from the U.S. on advanced GPUs and chipmaking equipment have accelerated domestic innovation. In response, firms like Cambricon, Biren Technology, and Tianshu Zhixin are developing high-performance AI accelerators, though they still lag behind NVIDIA’s most advanced offerings.

Beyond chips, sovereign infrastructure extends to the software layer. China is promoting homegrown machine learning frameworks like PaddlePaddle (Baidu) and MindSpore (Huawei) as alternatives to PyTorch and TensorFlow. These frameworks are integrated into government training programs and recommended for use in state-backed AI projects. Sovereign OS platforms (e.g., Kylin OS) and domestic code repositories (e.g., Gitee) round out a localized stack designed to ensure continuity and security in the face of geopolitical uncertainty.

Industrial Deployment & Tech Diffusion

AI adoption is a core policy priority in sectors such as manufacturing, healthcare, transportation, and agriculture. Through initiatives like Made in China 2025 and AI pilot zones, the government accelerates deployment via smart factories, autonomous vehicles, and AI-enhanced public services. Demonstration zones and SME-targeted incentives broaden access and validate the commercial and policy value of AI deployments at scale.

Sector Prioritization and National Goals

China’s AI strategy goes beyond R&D and infrastructure—it places strong emphasis on deployment across critical sectors. Strategic documents like the State Council’s 2017 AI Plan and the 14th Five-Year Plan identify manufacturing, transportation, healthcare, agriculture, finance, and energy as primary beneficiaries of AI transformation. AI is seen not only as an enabler of economic efficiency but as a means of ensuring national resilience and modernization.

In manufacturing, AI-driven automation and robotics are key components of the Made in China 2025 initiative. Sectors such as steel, electronics, and automotive now integrate vision-guided robotics, AI-based quality control, and predictive maintenance systems. In transportation, smart traffic systems and autonomous vehicle pilots are underway in over 30 cities. Healthcare has seen adoption of AI diagnostic platforms, especially in radiology and pathology, approved by the National Medical Products Administration (NMPA). These deployments are backed by policy documents that set quantitative targets—for example, the State Council calls for AI to contribute ¥1 trillion in core industry output by 2030.

Pilot Zones and Demonstration Projects

To facilitate deployment, China has launched a series of pilot zones and demonstration projects that act as live sandboxes. Local governments are empowered to test AI applications in real-world environments under flexible regulatory conditions. Hangzhou’s City Brain system, developed with Alibaba, is a flagship example: it manages traffic lights based on real-time analytics, reducing congestion by up to 15% in some districts.

In agriculture, smart farm initiatives use drone-based crop monitoring and AI-based irrigation management to optimize yields. In education, platforms like Squirrel AI offer adaptive learning tools tailored to student behavior, deployed in both urban and rural classrooms. These demonstration zones serve a dual purpose: they validate technical feasibility while also producing evidence for policy scaling.

SME Access and Ecosystem Inclusion

Recognizing that AI adoption should not be limited to tech giants, the government has introduced specific policies to bring small and medium enterprises (SMEs) into the fold. In AI pilot zones, SMEs can access subsidized computing resources, training programs, and technical consulting services. Local governments issue digital transformation vouchers that can be used to purchase AI solutions or participate in cloud-based diagnostics for production optimization.

Open-source tools developed under national projects—such as PaddlePaddle and dataset libraries maintained by the Beijing Academy of Artificial Intelligence (BAAI)—are promoted for SME usage. These efforts democratize AI access and prevent concentration of benefits among a few elite firms. Surveys indicate growing uptake of AI in logistics, retail, and precision manufacturing among mid-sized firms, a trend the government is actively monitoring and supporting through trade associations and regional innovation bureaus.

Regulatory, Ethical & Safety Frameworks

China has implemented one of the world’s most extensive AI regulatory regimes, encompassing algorithmic accountability, content synthesis, and generative AI. Key policies from CAC mandate algorithm registration, fairness, and safety testing. Ethical principles from MOST and BAAI shape value alignment across industry and academia, while emerging audit and standards frameworks enhance governance of high-risk systems.

Algorithm Regulation and Governance

China has developed one of the world’s most comprehensive regulatory regimes for AI, starting with the CAC’s algorithm regulations enacted in early 2022. These rules require major platform operators to register their recommendation systems with government authorities, submit to audits, and allow users to opt out of algorithmic personalization. They also mandate fairness, explainability, and prevention of discriminatory practices—marking an early and decisive effort to embed values into algorithmic decision-making.

Further refinements followed with the 2023 Deep Synthesis Regulation, targeting generative technologies like deepfakes. Providers must ensure AI-generated content is clearly labeled, free from misinformation or illegal material, and subject to model risk assessments. The regulation reflects Beijing’s concern about social stability and the potential misuse of AI in information warfare, fraud, or political disruption.

Generative AI Oversight

In 2023, the rapid emergence of ChatGPT-like models prompted the Cyberspace Administration of China to issue the Interim Measures for the Management of Generative AI Services. These rules require model developers to obtain a license before public release, ensure that training data is legal and representative, and maintain mechanisms for content moderation. Notably, models must align with "socialist core values," reflecting the ideological filter applied to technical development in China.

By mid-2024, over 20 large-scale language models—including those developed by Baidu, Alibaba, and Zhipu AI—had received approval for public deployment. Developers are required to watermark outputs, retain logs of generated content, and allow regulators access to training datasets. This legal architecture aims to establish accountability for AI-generated media and maintain state oversight over the information ecosystem.

Ethics and Value Alignment

Beyond compliance, China has released several high-level ethical frameworks for AI. MOST’s Ethical Norms for New Generation AI emphasize principles such as human-centeredness, privacy protection, and controllability. The Beijing AI Principles developed by BAAI and endorsed by industry and academia, add commitments to fairness, transparency, and long-term safety.

These documents guide the creation of internal AI ethics boards in tech firms and shape university ethics curricula. Government-backed standardization bodies like the China Electronics Standardization Institute (CESI) have also begun defining metrics and protocols for algorithmic audits, robustness testing, and “trustworthiness” certifications. While enforcement remains limited compared to regulatory compliance, these ethical frameworks help harmonize expectations, promote responsible innovation, and demonstrate China’s engagement in international AI safety dialogues.

International Engagement & Standards

China is shaping global AI norms through multilateral diplomacy, international standardization, and digital infrastructure exports. It promotes a sovereignist AI governance model through the Global AI Governance Initiative and plays a growing role in ISO and ITU standards bodies. Through the Digital Silk Road, China exports AI technologies and offers training to developing nations, reinforcing its influence in the Global South.

Multilateral Engagement

China has actively sought to shape the global governance of artificial intelligence, positioning itself as a key player in multilateral institutions. In October 2023, China proposed the Global AI Governance Initiative, presented by President Xi Jinping at the Belt and Road Forum. The initiative called for an open, inclusive, and fair approach to AI governance under the framework of the United Nations. It emphasized mutual respect for sovereignty, equitable access to AI technologies, and international cooperation in preventing misuse. This was followed by China’s co-sponsorship of a UN General Assembly resolution on AI in December 2023, which highlighted capacity building for developing countries and the promotion of ethical AI standards.

China uses these diplomatic avenues to counterbalance U.S.- and EU-led AI governance initiatives, arguing that no single country should dominate rule-setting for emerging technologies. It has urged the UN to play a central role in coordinating international cooperation on AI and has called for equal voice and participation by the Global South. China's multilateral strategy reflects its broader diplomatic agenda of reshaping global governance systems to reflect a multipolar world order.

Standardization Efforts

China also plays an increasingly active role in global technical standardization for AI. Through participation in bodies like the International Organization for Standardization (ISO) and the [International Electrotechnical Commission (IEC)] Joint Technical Committee 1, Subcommittee 42 (JTC 1/SC 42), Chinese experts have contributed to draft standards for facial recognition, algorithm transparency, and machine learning frameworks. In fact, Chinese proposals have often led working groups or served as reference frameworks for international standards, especially within the International Telecommunication Union (ITU), where Chinese firms like ZTE and Dahua have shaped guidelines on surveillance system interoperability and AI-powered video analytics.

Domestically, the China Electronics Standardization Institute (CESI) coordinates national efforts to align internal AI standards with global practices. It works closely with major tech firms and the Artificial Intelligence Industry Alliance (AIIA) to develop guidance on algorithmic trustworthiness, auditability, and performance evaluation. China's proactive standardization strategy not only ensures regulatory consistency at home but also gives its firms an advantage in international markets by aligning technical designs with globally accepted norms.

South-South Cooperation

In addition to participation in multilateral and standards-based frameworks, China has promoted AI development through its broader Digital Silk Road initiative. This strategy emphasizes the export of AI infrastructure, surveillance systems, and smart city technologies to developing countries in Southeast Asia, Africa, and Latin America. Companies like Huawei, Hikvision, and CloudWalk have provided facial recognition systems, data centers, and AI-enabled telecom infrastructure in countries such as Kenya, Zimbabwe, Laos, and Venezuela—often financed through state-backed loans.

China also facilitates training and capacity building through programs hosted by the Ministry of Industry and Information Technology (MIIT), which organizes seminars and certification courses for AI professionals and regulators from ASEAN, the African Union, and Belt and Road countries. These efforts position China not just as a vendor of AI technologies, but also as a long-term partner in digital transformation for emerging markets. In this way, China is leveraging AI diplomacy to enhance geopolitical influence while expanding its companies’ global reach.

Defence, Security & Dual-Use Considerations

AI is integral to China’s military modernization strategy via Military-Civil Fusion, where civilian R&D is leveraged for battlefield technologies. The PLA is advancing in swarm robotics, ISR, and cyber capabilities, while export controls regulate the outflow of sensitive AI systems. Internationally, China supports selective AI arms control but resists constraints on military AI development, preferring voluntary norms over binding treaties.

Military-Civil Fusion

AI is deeply integrated into China’s broader military modernization strategy through its doctrine of Military-Civil Fusion (MCF). This policy blurs the lines between civilian innovation and defense applications, enabling the People’s Liberation Army (PLA) to benefit directly from breakthroughs in commercial AI research. The Central Military-Civil Fusion Development Commission, chaired by President Xi Jinping, has prioritized AI-enabled command systems, predictive analytics, and battlefield robotics.

State-owned defense firms, as well as private tech giants like Huawei and SenseTime, contribute to dual-use technologies through joint labs and PLA-affiliated research programs. Universities such as the National University of Defense Technology (NUDT) conduct R&D on autonomous navigation, facial recognition, and cyber defense systems. These partnerships ensure that civilian AI advances in areas like computer vision, natural language processing, and edge computing can be quickly adapted for military purposes.

AI in Warfare

China's military thinkers increasingly describe future warfare as "intelligentized" (智能化战争), where AI will play a decisive role across command, control, communications, intelligence, surveillance, and reconnaissance (C4ISR). PLA strategy documents, including those from the Academy of Military Sciences, envision the use of AI for decision support, autonomous drone swarms, and real-time battlefield analytics.

Chinese firms have demonstrated capabilities in AI-powered loitering munitions, unmanned aerial vehicles (UAVs), and underwater autonomous systems. Trials of swarm drones capable of coordinated strikes, as seen in PLA drills and exhibitions, suggest rapid progress in distributed AI for combat applications. The PLA Strategic Support Force also employs AI for cyber and electronic warfare, leveraging pattern recognition and anomaly detection to counter foreign threats in information domains.

Export Controls

To manage the risks of dual-use AI technologies, China has expanded its Export Control Law, updated in 2020. The law includes provisions for sensitive technologies such as recommendation algorithms and facial recognition software. In 2020, the government restricted the export of TikTok’s core recommendation engine and limited the overseas sale of high-end surveillance chips and smart camera platforms.

These controls serve multiple purposes: safeguarding national security, protecting proprietary intellectual property, and leveraging technological assets in geopolitical negotiations. For example, during the attempted sale of TikTok’s U.S. operations in 2020, China’s Ministry of Commerce intervened under the Export Control Law to block the transfer of key algorithmic assets.

International Norms

China has participated in international dialogues on the ethical use of military AI, including United Nations meetings on Lethal Autonomous Weapon Systems (LAWS), where it has expressed conditional support for limiting fully autonomous weapons. China has called for “meaningful human control” in lethal systems, while opposing blanket bans on AI military development or research.

This nuanced stance allows China to continue advancing military AI under the guise of ethical restraint. China abstained from the 2023 UN General Assembly vote on initiating formal negotiations for a ban on autonomous weapons, signaling its preference for voluntary codes of conduct over legally binding treaties. Nonetheless, China’s engagement in these forums demonstrates an effort to shape global rules in a manner consistent with its strategic interests.

Performance Metrics & Monitoring

China tracks AI progress through indicators covering industry output, innovation performance, talent development, and technology adoption. MIIT, local governments, and think tanks like Tsinghua publish regular benchmarks, while metrics such as AI patent filings and sectoral GDP contribution shape policy adjustments. Adoption reporting ensures accountability, reinforcing China’s data-driven, iterative model of AI implementation.

Industry Output

China uses industry-scale metrics to track the growth of its AI ecosystem. According to the Ministry of Industry and Information Technology (MIIT), the core AI industry surpassed ¥500 billion (~$70 billion) in output by the end of 2023. This figure encompasses companies specializing in AI chips, vision systems, natural language processing, and autonomous platforms. Broader related industries—such as smart manufacturing and AI-enabled robotics—are also monitored to assess spillover effects and productivity gains.

Local governments set AI growth targets within their own jurisdictions. For example, Beijing’s AI Action Plan calls for a ¥300 billion AI industry by 2025. These economic targets are integrated into performance evaluations of municipal leaders and state-owned enterprise managers, ensuring alignment between national strategy and local execution.

Innovation Metrics

To gauge the quality and global relevance of AI research, China closely monitors publications and patents. As of 2023, Chinese institutions filed more AI-related patents than any other country, according to WIPO data. These include inventions in computer vision, edge computing, and generative models. China also ranks second globally in AI research citations, trailing only the U.S., according to the Stanford AI Index.

The China AI Development Report, published by Tsinghua University, provides annual updates on academic outputs, industrial innovation, and comparative global benchmarks. These reports are used by policy planners to assess progress and calibrate research funding.

Talent KPIs

Human capital is a core KPI in China’s AI performance system. The Ministry of Education tracks enrollment in AI-related majors, with over 345 universities offering dedicated AI programs as of 2023. The number of AI professionals trained each year exceeds 100,000, according to national figures cited in the China Artificial Intelligence Industry Development Alliance (AIIA) reports.

Returnee researchers from the U.S. and Europe are also logged as part of the “brain circulation” strategy. Talent development is measured not only in enrollment numbers but also through the number of doctoral theses defended in AI subfields, the frequency of international conference participation, and success in national AI competitions and challenges.

Adoption Tracking

Finally, China monitors the penetration of AI into public services and traditional sectors. Local governments submit implementation reports covering AI in education (e.g., adaptive learning platforms), healthcare (e.g., AI-aided diagnostics), and manufacturing (e.g., intelligent robotics). For instance, the Shanghai Municipal Government publishes detailed deployment KPIs as part of its digital economy dashboard.

The National Bureau of Statistics integrates AI adoption into sectoral productivity analyses, estimating the contribution of intelligent systems to GDP growth. Provinces like Guangdong and Zhejiang conduct AI-readiness assessments to identify digital infrastructure gaps and adjust subsidy programs accordingly. These metrics help close feedback loops between planning and deployment, reinforcing China’s iterative approach to AI policy execution.

Strategic Foresight & Resilience

China’s AI policy includes built-in foresight mechanisms to address geopolitical risks, technological bottlenecks, and emerging threats. Policies emphasize hardware self-sufficiency, algorithmic safety, and adaptability to disruptive innovations like generative AI. By embedding scenario planning and agile regulation into its approach, China seeks to future-proof its AI ecosystem against external shocks and internal instability.

Bottlenecks

Despite its rapid progress, China’s AI strategy acknowledges persistent structural vulnerabilities. Foremost among these is the country’s dependence on foreign semiconductor equipment and high-performance GPUs for training large models. The tightening of U.S. export controls, particularly those targeting advanced AI chips from NVIDIA and the tools needed for 7nm or smaller semiconductor fabrication, has exposed a critical supply chain bottleneck. The U.S. Department of Commerce’s October 2022 rules and subsequent updates in 2023 effectively curtailed China’s access to top-tier AI training hardware.

Alongside hardware, China still trails in the quality of original AI research—especially in frontier theoretical areas such as AGI safety and unsupervised learning. While China leads in the volume of AI papers, many remain undercited relative to the U.S. and Europe, a gap highlighted in the Stanford AI Index. Reliance on U.S.-led software ecosystems such as PyTorch and TensorFlow is another area of concern. These dependencies not only limit China’s self-reliance but also leave its innovation ecosystem exposed to potential geopolitical disruption or licensing restrictions.

Scenario Planning

To mitigate such vulnerabilities, Chinese policymakers increasingly embrace scenario-based strategic planning. This includes preparing for worst-case situations such as total technological decoupling from the West. The idea of “indigenous controllability” (自主可控) has become a central theme in Chinese policy discourse. It is emphasized in Politburo study sessions and embedded in strategic blueprints such as the 14th Five-Year Plan for the Digital Economy.

Chinese agencies have responded by accelerating domestic chip design (via firms like Biren and Cambricon), launching national AI model repositories, and developing cloud computing clusters that do not rely on U.S. infrastructure. AI capabilities are increasingly benchmarked not only on performance but on autonomy from foreign components. State guidance funds and local governments have been directed to prioritize investments in "stranglehold technologies" (卡脖子技术), and “Eastern Data, Western Compute” initiatives have aimed to rebalance infrastructure resilience geographically.

AI Safety

Safety and reliability have emerged as pillars of China’s AI strategy, especially in light of the rise of foundation models and generative AI. In 2023, the Interim Measures for the Management of Generative AI Services were issued to ensure that large models undergo risk assessments, are trained on lawful datasets, and include controllability mechanisms. These measures require generative AI outputs to be watermarked, a rule that came into force in 2024 and set China apart as one of the first major economies to implement enforceable provenance labeling for synthetic content.

In parallel, model developers are required to maintain logs, ensure safety testing, and adapt outputs in accordance with “core socialist values”. National standards for AI safety testing are also being drafted by CESI and AIIA. These include protocols for robustness, red teaming, bias evaluation, and error propagation. While enforcement mechanisms are still maturing, the regulatory infrastructure shows clear foresight in anticipating challenges such as disinformation, hallucinations, and algorithmic bias.

Agile Regulation

China’s AI governance is marked by its responsiveness to emerging technologies and public sentiment. Regulatory frameworks are often released in draft form for public comment, enabling feedback from companies, academics, and citizens. For example, draft rules on deep synthesis content and AI-generated media were adjusted following civil society concerns and corporate lobbying. Iterative revisions and fast rollouts show that Chinese regulators are seeking to strike a balance between control and flexibility.

This adaptive approach also extends to national policy. When generative AI technologies exploded globally in late 2022, China’s regulators acted within months to establish a licensing framework—underscoring a degree of regulatory agility uncommon among major economies. The ability to update guidelines in response to new technologies reflects a strength in China’s top-down but tech-savvy governance system, bolstered by real-time monitoring of trends on platforms like Weibo and feedback through AI industry alliances.

Foundational AI Capabilities

China is developing an independent AI stack spanning chips, compute, frameworks, and foundation models. Domestic processors like Ascend and Kunlun support national cloud platforms, while frameworks like PaddlePaddle provide a sovereign alternative to Western ML tools. China’s LLM ecosystem now includes over 20 CAC-cleared models, reflecting a strategic push for control over both infrastructure and cognitive capabilities.

Compute and Chips

At the core of China’s AI strategy lies the development of sovereign computing infrastructure. Domestic AI accelerators such as Huawei’s Ascend 910 and Baidu’s Kunlun 2 are now deployed in public and private cloud environments. These chips power national compute centers across pilot zones and are used to train large-scale language models and support edge AI applications.

Despite these gains, fabrication remains a critical weak link. China’s top chipmaker, SMIC, has made progress on 7nm and 5nm processes, but lacks access to extreme ultraviolet (EUV) lithography machines due to export bans imposed by the U.S., Netherlands, and Japan. As a result, China’s most advanced chip production lines still lag behind global leaders in performance and yield. This gap has driven Beijing to pour billions into national semiconductor funds and establish domestic alternatives to critical supply chain nodes, from materials to tooling.

Software and Platforms

On the software side, China is striving to build a homegrown machine learning stack. Baidu’s PaddlePaddle is the most prominent open-source deep learning framework in China, widely used in industrial applications, smart devices, and academic projects. Huawei’s MindSpore supports cloud-edge collaboration and AI inference across Ascend chip architectures. These frameworks are promoted through government-backed curriculum reforms, open-source challenge competitions, and preferential procurement policies.

Although TensorFlow and PyTorch remain dominant among elite researchers, Chinese frameworks are gaining traction due to regulatory favor and integration with domestic infrastructure. Moreover, state-linked platforms such as OpenI and ModelScope serve as national repositories for pretrained models, datasets, and training tools, reducing dependency on U.S. GitHub and Hugging Face platforms.

LLM and Foundation Models

Since 2021, China has ramped up its development of large-scale foundation models. The Beijing Academy of Artificial Intelligence (BAAI) released WuDao 2.0, a multimodal model with 1.75 trillion parameters, which briefly claimed to be the world’s largest at the time. Baidu’s ERNIE Bot (based on its ERNIE 4.0 model) and Alibaba’s Tongyi Qianwen are widely deployed in search, customer service, and productivity tools.

By mid-2024, over 20 Chinese-developed large language models (LLMs) had been approved for public deployment following regulatory review under the CAC’s generative AI rules. These include models from Zhipu AI, SenseTime, iFlytek, and MiniMax. Most models are trained on bilingual corpora and tuned to align with domestic norms and censorship standards. Collectively, these efforts signal China’s intent to achieve foundational independence—not just in infrastructure, but in the AI cognition layer itself.

Public Trust, Inclusion & Social Equity

Public engagement and equitable access are central to China’s AI rollout. Through media campaigns, regulatory responsiveness, and regional outreach, the government fosters trust while aligning deployment with national unity goals. Multilingual AI tools, digital inclusion programs, and draft policy consultations ensure that AI serves broad societal interests, even as public participation remains guided and state-managed.

Public Communication

China’s AI rollout is accompanied by a state-driven campaign to shape public perceptions. Government media regularly report on AI’s role in economic development, medical breakthroughs, education, and urban safety. Flagship programs like Alibaba’s City Brain and Baidu’s autonomous taxis are featured on CCTV and Xinhua as examples of AI’s societal value. This narrative fosters public trust while positioning AI as a tool of national progress.

When issues such as deepfake scams or algorithmic discrimination emerge, regulators respond swiftly. For example, following public concern over algorithmic pricing discrimination (“big data kills the loyal customer”), the CAC issued guidance requiring transparency in personalization and banning exploitative pricing algorithms. These visible interventions serve both as governance tools and reassurance signals to the public.

Multilingual Access

China’s AI strategy emphasizes cultural and linguistic inclusivity. Major voice recognition and translation tools from iFlytek and Tencent are trained on regional dialects and minority languages, including Uyghur, Tibetan, and Mongolian. These tools are used in courts, social services, and education programs to reduce administrative friction and promote equity. For example, in Xinjiang, natural language processing (NLP) is deployed for bilingual court records and voice translation in hospitals.

The push for multilingual AI also aligns with political objectives: improving public services in ethnic regions while enhancing state control through digital platforms. While some critics argue this serves surveillance ends, the government frames it as bridging digital divides and reinforcing national unity through inclusive design.

Equity in Access

To ensure that AI benefits are not limited to urban elites, Beijing promotes programs that extend technology to rural and underdeveloped regions. Initiatives like “Digital Village” use AI for smart irrigation, crop diagnostics, and rural e-commerce. In healthcare, AI-assisted remote diagnosis tools are deployed in township clinics via platforms developed by Ping An Good Doctor and Tencent Health.

The National Health Commission has supported trials where AI triage systems aid doctors in underserved areas by flagging high-risk patients based on medical imaging and patient records. In education, AI tutors like Squirrel AI are deployed to provide personalized learning in low-income schools. These projects demonstrate how AI is used to enhance service quality and bridge gaps in human resource availability.

Feedback Mechanisms

China’s AI governance is not completely unidirectional. The publication of draft regulations for public comment has become standard, as seen with the CAC’s algorithm and deep synthesis rules. Firms, universities, and even citizens use Weibo, Zhihu, and government portals to offer input—some of which results in revisions.

While civil society groups remain weak relative to other countries, industry associations like AIIA and CAAI host forums where ethical concerns are debated, and technical guidelines are developed. These feedback loops help regulators update frameworks in a timely and technically informed manner. In combination with sentiment analysis tools that scan public opinion online, they form part of an evolving model of guided participatory governance—aligned with state priorities but responsive to public needs.

15. Evolution of China’s National AI Strategy (2015–2025)

This section traces the development of China’s national AI strategy over the past decade, mapping key milestones and policy shifts year by year. From early investments in semiconductors and institutional architecture to global outreach and AI governance diplomacy, China’s AI trajectory has blended domestic industrial priorities with global ambitions.

2025: Global Outreach & Strategic Self-Sufficiency

China continues to pursue self-reliance in semiconductors amid U.S. restrictions, while promoting its AI governance model through international workshops aimed at the Global South. The rollout of homegrown large language models (LLMs) like DeepSeek, designed for cost-effectiveness and lean training, signals a maturing ecosystem that positions Chinese alternatives to Western models like ChatGPT. AI integration advances across sectors such as healthcare, surveillance, and smart cities. Ten capacity-building workshops planned by the Cyberspace Administration for developing nations demonstrate a growing push to export Chinese AI standards and norms. Meanwhile, U.S. chip curbs continue to hinder domestic fab capabilities, reinforcing China's prioritization of sovereignty in compute and software stacks.

2024: Capacity-Building & Norm-Setting

In September 2024, China unveiled the AI Capacity-Building Action Plan for Good and for All, emphasizing multilateralism under UN frameworks and AI development in the Global South. The plan outlines cooperative infrastructure, talent, safety, and data governance frameworks. Huawei’s Ascend 910C was introduced as a competitor to Nvidia’s H100, signaling China's hardware resilience. China led globally in generative AI patents, filing more than 38,000 since 2014. New AI startups and established national champions drove optimism and private sector dynamism. Strategic themes included global cooperation, digital sovereignty, and a shift toward self-sufficient chip design and cloud architectures.

2023: Generative AI Rulebook

China moved quickly to regulate foundation models with the Interim Measures for the Management of Generative AI Services, which came into force before the EU finalized its AI Act. These rules mandated training data disclosure, content watermarking, and alignment with “core socialist values.” By mid-2023, over 20 domestic models, including those from Baidu, Alibaba, and Zhipu AI, were cleared for public release under the CAC's licensing framework, establishing China as a regulatory first mover in generative AI.

2022: From Algorithms to Compute Topology

Two landmark policies defined 2022: the Deep Synthesis Provisions, the world’s first binding regulation on deepfake labeling; and the “Eastern Data, Western Compute” initiative, which built eight national compute hubs to rebalance resource distribution and power usage. This year marked the convergence of content integrity regulation and infrastructure decentralization in China’s AI roadmap.

2021: Algorithmic Governance & FYP Lock-In

The Provisions on Algorithmic Recommendation Services required online platforms to disclose algorithmic ranking logic and allow users to disable personalization. These were coupled with the 14th Five-Year Plan, which elevated AI to national basic infrastructure and embedded it in broader digital economy development priorities.

2020: Pandemic Testbed & 'New Infrastructure'

The COVID-19 crisis accelerated AI integration in public health. Computer vision was used in triage, and contact tracing relied on AI-powered QR health codes. The central government earmarked over ¥1 trillion for AI, 5G, and cloud infrastructure under the “new infrastructure” stimulus policy, setting the stage for post-pandemic digital transformation.

2019: Race of Two Giants

AI was framed as a geopolitical competition between China and the U.S. Policymakers highlighted challenges in top-tier talent, AI software, and semiconductor self-reliance. At the same time, China led globally in AI research publications and patents, and saw commercial success in computer vision (SenseTime) and consumer drones (DJI).

2018: Institutional Build-Out & Military Integration

Key developments included the Ministry of Education’s AI Innovation Action Plan for Colleges and Universities, and the launch of military AI centers under the National University of Defense Technology (NUDT). CAICT released its AI Security White Paper, warning of arms races while encouraging domestic AI framework development.

2017: New Generation AI Development Plan

The State Council issued the New Generation Artificial Intelligence Development Plan (AIDP), which became the cornerstone of China’s AI policy. It outlined three developmental phases to 2030 and explicitly tied AI to economic modernization and national security.

2015: Made in China 2025 & First Chip Fund

The Made in China 2025 initiative was launched, identifying AI and semiconductors as strategic sectors. The same year, China established its first National Integrated Circuit Industry Investment Fund to reduce foreign tech dependence and bolster domestic chip manufacturing.

Prior to 2015: Vision & Early Bets

Even before formal plans, AI was identified as a strategic priority by President Xi Jinping, who framed it as a “new engine of national rejuvenation.” Public R&D funding and early-stage patent activity in NLP, vision, and machine learning laid the groundwork for China's rapid acceleration in the following decade.