Overview
According to CCTV.com on November 22, China’s first batch of pilot programs for trusted data spaces has reached a phased milestone. Launched in 2025, the pilots now cover 16 key industries including industrial manufacturing, transport and logistics, and healthcare, and have produced 457 high-quality datasets and 136 AI vertical models. These support high-value scenarios such as multimodal freight, mutual recognition of carbon-footprint accounting, new-material R&D and coordinated vehicle-charge-grid scheduling. The program is now deployed across 13 cities, 22 industries and 28 enterprises.
In parallel, the National Data Administration has set out a new focus for building the data-element system. Liu Liehong, the administration’s head, said at the opening of the 2026 World Computing Conference that the token links value transmission among computing power, models and data, becoming the value anchor of the digital-intelligent economy. Next steps include expanding token-service applications and improving standards for token metering, billing and quality assessment, to guide orderly development and avoid low-quality duplication. Shanghai, one of the pilot cities, has built a city-level trusted data space: at the Hongqiao transport hub, with over 1.1 million daily passengers, travelers can obtain real-time information on metro, bus routes and taxi queues via mobile positioning.
Together, the developments mark data elements moving from institutional design toward infrastructure that is measurable, deliverable and verifiable.
Background & Interpretation
1. Technical background and evolution
A trusted data space is infrastructure for trustworthy data circulation, built around making data “usable but invisible” — so providers know who uses their data, in what context and how. Data markets have long faced difficulties in confirming rights, pricing and circulation; Arrow’s information paradox makes buyers unable to value data before seeing it. Trusted data spaces provide a trust layer through technical controls that keep usage traceable. Moving from pilot launch in 2025 to results in 2026 reflects a gradual path of institutional design, pilot validation and scaled rollout.
2. Core drivers and mechanisms
Token-based metering effectively establishes a common yardstick for transmitting value among compute, models and data in the AI era. The National Data Administration notes that “token factories” elevate compute leasing, model supply and data capabilities into production services, turning high-quality datasets into measurable, deliverable intelligent services priced by call volume. The 457 datasets and 136 vertical models from the pilots are the tangible carriers of this data-to-model-to-service chain. Ministry of Industry and Information Technology data show Chinese open-source models have passed 10 billion global downloads, and Chinese models topped OpenRouter weekly call volumes for 29 consecutive weeks — making measurable, priceable, tradable attributes increasingly real.
3. Market structure and value-chain impact
The pilots pull the value chain in two directions. Roles for data providers, users and third-party services are becoming clearer, spurring demand for data governance, privacy computing and asset valuation. Meanwhile, combining vertical models with industry datasets lowers the barrier to intelligence in manufacturing, logistics and healthcare. Risks remain: duplicated trading platforms, inconsistent rights and pricing mechanisms, and the need for cross-platform metering alignment and unified quality yardsticks — leaving room for fragmented rules in the near term.
Implications & Outlook
Near term, the published results should draw more cities and industries into trusted data-space construction, with local supporting policies likely to follow. Medium-term variables include the pace of token metering and quality standards, whether metering aligns across platforms, how pilot experience scales nationally, and the reuse rate of high-quality datasets. Unified standards could accelerate data elements from policy-driven to market-driven growth.
Takeaways for Industry Participants
Enterprises should fold data governance and assetization into long-term strategy, prioritizing circulatable resources within trusted data-space frameworks. AI providers should watch token metering standards and adjust pricing and cost accounting early to avoid disputes over incompatible metrics. Regulators and platform operators should put cross-platform mutual recognition and unified quality yardsticks first, preventing duplication and fragmentation from eroding the value of the pilots.
This column compiles industry information and shares technical perspectives; it does not constitute investment advice.