The Chinese Academy of Sciences has placed artificial intelligence inside the operating design of its next five-year development plan, not as a standalone technology theme but as a tool for changing how research is conducted and managed. The Chinese Academy of Sciences 15th Five-Year Development Plan (中国科学院十五五发展规划), covers 2026 through 2030 and sets a longer-range institutional outlook for 2035.
The plan’s AI provisions are significant because they connect scientific models, research data, national facilities, and administrative systems. Reporting by the South China Morning Post describes the document as a blueprint aligned with China’s 15th Five-Year Plan. A CAS-hosted report carrying China News coverage says the institution is actively embracing the strategic opportunity created by AI and gives AI-enabled changes in research paradigms a dedicated section.
This is a plan for proposed capabilities, not evidence that those capabilities are already operating. The supplied CAS materials do not establish a budget, staffing level, compute capacity, model release schedule, benchmark result, or commercialization timetable. They describe the infrastructure and management architecture CAS intends to build or strengthen during the plan period.
Panshi and the infrastructure for AI-driven science
At the center of the AI-for-science provisions are four named forms of infrastructure. The plan calls for building or strengthening the Panshi scientific foundation model, an intelligent scientist system, a scientific corpus, and a scientific data center. The model is referred to in Chinese as the Panshi (磐石) scientific foundation model. No verified source supplied a parameter count, training-compute figure, dataset size, version number, benchmark, public release, or open-source status for Panshi.
That boundary matters. Panshi is a plan provision, not a verified product launch. The available reporting supports describing it as part of CAS’s intended scientific-AI infrastructure, but not as a model that researchers can already access or as a system that has demonstrated a particular capability. The same distinction applies to the intelligent scientist system, corpus, and data center: the plan identifies them as building blocks for future research operations.
CAS says AI should empower work across basic science, strategic high technology, and sustainable development. Those fields give the plan a broad remit. AI is presented as a layer that can support different research domains rather than as a replacement for a particular scientific discipline or a promise to automate discovery. The sources do not say that AI will replace scientists, produce a specific discovery, or complete a defined scientific task.
The plan also calls for major national scientific infrastructure and national field observation and research-station platforms to move toward digital, networked, and intelligent operation. That proposed transition includes stronger data integration, unified standards, security protection, and data-fusion use. The wording describes an intended direction. It does not show that these upgrades have already been completed.
This data emphasis also distinguishes the CAS plan from a simple model-development agenda. A scientific corpus and data center imply attention to how research information is organized and made usable, while standards and security provisions address how that information can be integrated across facilities. The plan therefore treats AI-for-science as an institutional system involving models, data, platforms, and operating rules.
Smart CAS links research management to institutional coordination
AI also appears in the plan’s management provisions. CAS proposes a Smart Chinese Academy of Sciences special task or project, translated from 智慧中国科学院, together with an integrated intelligent scientific-research management platform. Reported functions include an institution-wide center for integrating and allocating innovation resources, a command and dispatched-operations center for scientific-innovation activities, and an information-coordination and integration center for research management.
The plan further proposes a dynamic, comprehensive capability map of research capacity. In practical terms, the stated concept would give CAS a way to organize information about institutional research capabilities and connect it with management activity. That description remains bounded by the available reports: they do not establish that the platform, centers, or capability map are operational today.
The management component is a notable extension of the AI strategy. It places intelligent coordination alongside AI-assisted scientific work, suggesting that CAS views research productivity as an issue of both technical capability and institutional organization. Resource allocation, scientific-activity command, and research-management information are all included in the proposed architecture. None of those provisions should be read as a completed administrative reform or a guarantee of specific performance.
The approach fits a broader EastFrontier pattern in which Chinese institutions are connecting AI policy to research and public systems. Earlier coverage of China’s 15th Five-Year Plan and scientific ambition examined the wider national planning context, while a separate report on AI as core national infrastructure covered the national-level framing. CAS’s document is narrower: it applies that planning logic to the country’s top academic institution and its own research-management structure.
2030 milestones and the 2035 institutional horizon
The plan sets six development goals for 2030 under the four firsts framework. According to Science and Technology Daily and ScienceNet reporting, CAS aims by that year to make major progress in seizing commanding heights in science and technology, make notable progress in becoming a source of original innovation, and achieve major breakthroughs in key core-technology efforts. The same list includes building a national highland for innovation talent, fully exercising the role of a national science-and-technology think tank, and making notable progress toward a world-class research institution.
These are stated objectives, not measured outcomes. They provide a timeline against which CAS’s AI infrastructure and management proposals may eventually be assessed, but the available reports do not provide AI-specific indicators for judging progress. It also does not assign a funding level or completion deadline to Panshi, the intelligent scientist system, or the Smart CAS platform.
The plan’s 2035 outlook is more expansive. It says CAS aims to have fully built a world-class research institution and become an original-innovation source and innovation highland with important global influence. That is a forward-looking institutional objective, not a claim that the status has already been achieved.
The planning process itself extended over two years, with strategic research reported to have begun in July 2024. ScienceNet says the document was approved after consultation and review by the CAS Academic Committee and deliberation by the CAS Party leadership group. Those process details are attributed to ScienceNet and China Science News reporting, rather than presented as independently checked against a full plan document.
CAS’s broader priority fields include integrated circuits and semiconductors, AI, quantum technology, life sciences and biotechnology, brain-computer interfaces, and controlled nuclear fusion, alongside basic and interdisciplinary research. The list shows the breadth of the institution’s 2026–2030 agenda, but it is not a list of funded projects or completed breakthroughs.
The clearest takeaway is therefore institutional. CAS has made AI a stated mechanism for transforming research and coordinating the organization that performs it. The next test will be whether the proposed models, data systems, intelligent facilities, and management platforms acquire verifiable capabilities during the plan period. For now, the document establishes goals and architecture, while leaving implementation, resources, and performance open.