The 2026 "data elements ×" contest (数据要素×大赛) held its national final and award ceremony in Qingdao on 23 October, alongside a smart-city conference. The National Data Administration published the readout.
The numbers given: the contest opened in April, drew more than 29,000 teams and more than 145,000 participants, and 330 projects reached the national final. 2026 is the closing year of the three-year action plan behind the contest.
Administration head Liu Liehong described a "531" implementation system for market-based allocation of data and said the circulation and use of data is moving from isolated points to broader coverage. He added that AI is reshaping how data is governed, widening the space in which data can circulate, and that AI plus high-quality datasets should be pushed together. The administration's research institute released an insight report on the contest, which states that a "flywheel" between AI and high-quality datasets is emerging. A housing-ministry vice minister said the urban-governance track asked projects to "deepen AI empowerment" while holding a security baseline. First-prize teams from three tracks presented their work at the ceremony.
Our takeA contest is a poor instrument for judging dataset quality, because nobody publishes the splits. The readout gives counts of teams and finalists and no baseline: no statement of what a winning dataset has to contain, how it was labelled, or how a model trained on it was tested. The "flywheel" language is directional rather than measured. If the 330 finalist projects published schemas, labelling rules and held-out sets, the programme would leave behind something reusable; a list of awards does not. That is the part to check before the next cycle starts.