The Ministry of Industry and Information Technology (MIIT) and the National Data Administration (NDA) have jointly issued a notice to implement the 2026 "Model-Data Resonance" action plan. This initiative aims to deepen the integration of artificial intelligence (AI) with the real economy, in line with the State Council's guidelines on promoting the "AI Plus" campaign and the specific实施意见 for "AI Plus Manufacturing."
The action plan targets key industrial sectors including steel, petrochemicals, non-ferrous metals, building materials, industrial machinery, automotive, medical equipment, power equipment, shipbuilding, aerospace, home furnishings, pharmaceuticals, biomanufacturing, historical classics, electronic components, consumer electronics, new displays, software, information and communications, and cybersecurity. By 2026, the program seeks to achieve several core objectives: develop high-value, technically feasible AI application scenarios; create specialized industry models and intelligent agents that embody industrial know-how; build high-quality general and specialized industry datasets; and foster collaborative innovation consortia. The ultimate goal is to establish a virtuous cycle where data, models, and practical applications mutually reinforce each other, thereby driving high-level AI empowerment of new industrialization.
Key tasks outlined in the notice include: 1. Developing general industry datasets and foundational industry models. Provincial regions and central state-owned enterprises are to select key sectors and create high-quality, generalized datasets for each. Based on these datasets, industry-specific models addressing common application challenges will be developed and deployed. 2. Identifying high-value application scenarios and constructing specialized datasets and intelligent agents. For selected high-potential scenarios, specialized datasets containing unique scenario knowledge will be built. Subsequently, dedicated models or specialized intelligent agents (for tasks requiring autonomous planning and execution) will be created and implemented. 3. Establishing evaluation datasets and improving model assessment mechanisms. Professional institutions will develop customized benchmark datasets to evaluate the capabilities of industry models and intelligent agents, creating a feedback loop for continuous improvement of both datasets and models. 4. Creating "Model-Data Resonance" spaces. These collaborative platforms, to be established by third-party neutral institutions, leading companies, or central enterprises, will feature integrated hardware/software infrastructure for cross-entity data aggregation and model training, along with management mechanisms for data collaboration, model co-development, and security. 5. Forming "Model-Data Resonance" innovation consortia. These groups, comprising compute providers, model developers, data firms, and application developers, will collaborate with the resonance spaces to create comprehensive, industry-level AI solutions. 6. Enhancing supporting ecosystems. This involves organizing promotional events, building training bases, cultivating multi-skilled talent, fostering AI and data technology innovation, and strengthening standardization efforts. 7. Designating "Key Cities" as pilot zones. Cities with strong AI industrial foundations and abundant data resources will be selected as focal points for implementing the action plan, serving as benchmarks for others.
Implementation requires provincial industrial authorities and central enterprises to submit detailed action plans by May 30, 2026. Progress reports are due by August 30, 2026, followed by a final summary and evaluation by November 30, 2026. MIIT and the NDA will provide policy support, establish a unified platform to showcase results, and promote successful experiences. Regions and enterprises demonstrating effective implementation will receive preferential support in relevant policies and projects.
Comments