Data & AI as Commons
Comprehensive state data policy and data exchanges; the Personal Information Protection Law (PIPL) has been in force since 2021, with the balance between state and individual interests still unresolved.
Where does China stand, and where might it break?
China treats data differently from the West — not primarily as a privacy right, but as a factor of production and a national asset. With the Personal Information Protection Law (PIPL, 2021) and the Data Security Law (DSL, 2021), China established a comprehensive legal framework. In 2022, the leadership classified data as the fifth factor of production alongside labour, capital, land and technology under the Data 20 Measures, and established state-run data exchanges in Shanghai, Shenzhen and Beijing through which data is traded, priced and potentially taxed. In 2023, China established a National Data Administration to coordinate the data economy. At the same time, the Cyberspace Administration of China (CAC) tightly regulates generative AI. The Interim Measures for the Administration of Generative AI Services entered into force in August 2023, requiring registration, security assessments, and content controls. AI adoption is nonetheless leading globally. According to CNNIC, the number of generative AI users reached approximately 602 million by end-2025. China accounts for around 70% of global AI patents and over 23% of publications, according to the Stanford AI Index 2026. Data is therefore simultaneously tradeable, state-administered, and comprehensively accessible — a commons under state direction.
If data is the central raw material of AI value creation, the rules governing its ownership and access will determine who shares in the proceeds. The fault line: China organises data as a state-administered factor of production — this permits trading, pricing and potential taxation at scale, but raises the question of whether individuals and data-supplying companies share in the value created, or whether value accrues predominantly to the state and a handful of platforms. Unlike in private-sector data markets, sovereignty here lies with the state, which simultaneously acts as regulator, exchange operator and beneficiary. This creates scale and controllability, but also a tension between commercial exploitation, the protection of the individual under PIPL and comprehensive state access. The central question through to 2040 is whether China turns data into a broadly shared commons whose proceeds are widely distributed, or whether the data rent concentrates in the hands of the state and a few corporations — and how much individual control over one's own data remains.
Paths for China
China deepens the state-run data exchanges so that data-supplying companies, and potentially individuals, share in the value created — through remuneration, data trustees and regulated revenue sharing. This is central because data is the fifth factor of production and the exchanges in Shanghai, Shenzhen and Beijing are already in place. This path turns data into a shared commons. The cost: data valuation is methodologically unproven, and as long as the state is simultaneously regulator, operator and beneficiary, equitable revenue sharing will be difficult to enforce.
Trade-off: Broad participation in the data rent versus the state's dual role
Feasibility: Fair only if the state's operator and beneficiary roles are separated
China is strengthening individual protection and control rights under PIPL so that individuals remain in control of their data and share in its use. This goes to the heart of the matter, because in the AI economy personal data constitutes the raw material. PIPL (2021) already provides a comprehensive framework. The price: individual data rights are in tension with pervasive state access and the use of data as a factor of production — genuine empowerment would limit state visibility, which runs counter to the logic of central governance.
Trade-off: Individual data sovereignty versus state governance
Feasibility: Limits arise from the dual nature of data as a state resource
China is expanding its data centres, chips, and energy infrastructure to secure the domestic computing base of the data economy. This matters because data generates value only when combined with computing power — and China remains import-dependent for advanced semiconductors. The country's lead in building out data centres and renewables is telling. The cost: sovereignty over advanced chips remains the most vulnerable point, and the energy appetite of data centres strains climate targets for as long as coal serves as a bridging fuel.
Trade-off: Data sovereignty versus hardware and climate constraints
Feasibility: Contingent on the success of semiconductor self-sufficiency
China is continuing its early, strict regulation of generative AI through the Cyberspace Administration of China (CAC) — encompassing registration, security assessments, and content controls. This is a logical path, as the Administrative Measures for Generative AI (August 2023) already deliver legal certainty within a framework of limited openness. The approach gives the state control over models and content. The cost: stringent governance may constrain innovation diversity and openness, creating tension with the ambition to lead globally in AI — control and the pace of innovation must be kept in balance.
Trade-off: Control and security versus innovation dynamics
Feasibility: Balancing oversight and openness remains a delicate act
Trigger-based phases
Trigger: Data exchanges, the National Data Administration, and the CAC's AI rules establish data as a tradeable, state-administered factor.
China is building the infrastructure to trade, price, and regulate data. AI adoption is leading. Yet who participates in the data rent remains an open question.
Trigger: Data exchanges, remuneration models, and data rights will determine whether value is broadly shared or concentrated.
China is deciding whether data-contributing companies and individuals participate in value creation from their data, or whether the rent accrues to the state and a small number of platforms. The tension between monetisation, privacy, and state access is coming into plain view.
Trigger: China has embedded data either as a broadly shared commons or as a state- and platform-concentrated rent.
The ownership, access, and remuneration framework for data is in place. Whether China makes data a broadly shared commons whilst preserving individual control will determine participation and trust through to 2040.
Real-world pilot programmes
Status: Active – state-operated exchanges for trading data products
The state-run data exchanges trade data products and pilot their pricing. They are the test case for whether data can be organised as a tradable factor of production and whether the returns from it can be shared broadly.
Status: Framework since December 2022, administration since 2023 — data as the fifth factor of production
The classification of data as a fifth factor of production and the establishment of the national data authority provide the institutional foundation of the data economy. They reflect China's distinctive approach to administering and exploiting data under state direction.
Status: In force since August 2023 — registration, security assessment, content controls
The CAC's early AI regulations provide legal certainty while maintaining centralised control over models and content. They are the test case for how China balances the pace of innovation with state oversight of AI.
Status: In force since November 2021 — comprehensive protection of personal data
PIPL provides a comprehensive framework for the protection of personal data, yet stands in tension with state access. It is the test case for how much individual data sovereignty remains within a state-administered data space.
What we are watching
What is measured is who participates in the returns from data. The threshold matters because data are the raw material of the AI economy. If the rent concentrates, the data commons remains a state- and platform-held asset. Broader participation would signal that data are becoming a shared resource.
What is measured is how much control individuals retain over their own data. The threshold matters because personal data form the raw material of AI. If control remains weak, power shifts entirely to the data administrator. Strengthened control would signal a balance between commercial exploitation and personal data protection.
This indicator measures control over the computational foundation of the data economy. The threshold matters because data creates no value without computing power, and export controls expose this flank directly. Where dependency persists, data sovereignty remains incomplete. A reduction would demonstrate success in achieving self-sufficiency.