China Accelerates National Strategy to Integrate Data Assets with AI Development

I tuned into the livestream from across the Pacific and watched a stage crowded with 372 companies and more than 16,000 registered guests. Xinhua reported that China's data industry hit 6.78 trillion yuan, roughly a trillion U.S. dollars, last year. Beijing, the dispatch makes clear, is now racing to coordinate that growth with artificial intelligence — and the room in Guiyang felt like a coordinated push in real time.

The Coordinated Push

The expo's theme — "Token: A New Path to Value of Data Elements" — signals exactly where policymakers want to steer the conversation: toward treating data as a tradable, verifiable asset class rather than a by-product. Liu Liehong, head of China's National Data Administration, framed the moment at the opening ceremony as a mandate to build high-quality datasets wherever AI advances. Yang Jie, secretary-general of the World Data Organization, pushed the same thread harder, arguing that the richness, purity, authenticity and professionalism of data directly determine the cognitive depth, knowledge accuracy and capability ceilings of large models.

That language matters more than it sounds. It is a state-backed bet that AI performance will be bottlenecked less by chips and more by the cleanliness of the data feeding the models. International delegations from Singapore, Morocco and Canada reportedly turned up hunting exactly that kind of partnership, drawn by the promise of a coordinated national market rather than a patchwork of provincial pilots.

A Grid That Learns to Listen

The most vivid proof point came from Guizhou Wujiang Hydropower Development Co., where Fan Sheng, the deputy general manager, walked reporters through a unified dispatch platform that now bundles hydropower, wind and solar into a single digital brain. Years of local meteorological and hydrological data were cleaned, integrated and standardized, then fed into models trained to predict conditions up to 15 days ahead. Company test data shared at the expo pegs wind forecast accuracy at 91.95 percent, solar at 94.54 percent, and cascade hydropower inflow at 94.1 percent — roughly four percentage points better than the year before.

For an industry long haunted by the volatility of renewable generation, that is not a vanity metric. It is the difference between curtailing clean power and actually putting it on the grid. And it is the kind of unglamorous data plumbing that decides whether the AI era hums or hemorrhages value.

What to Watch

The signal for founders and operators watching from the Valley is straightforward: the data infrastructure layer — the cleaning, the standardization, the tokenization of who owns what — is now where the policy oxygen is flowing. If the question of how to verify the integrity of traded data feels familiar, the same trust architecture is being hashed out in adjacent markets. Independent verification frameworks are moving quickly from a crypto curiosity to a baseline expectation in any industry that trades in data.

Watch whether the expo's "token economy" theme translates into binding rules by year-end, and whether other provinces replicate Guizhou's unified dispatch model. The winners in this coordinated push will not be the ones with the most GPUs, but the ones whose datasets survive audit.