Beijing High-End Industry Plan Unveiled, 22 Measures Across 7 Key Areas

On September 9, 2026, Beijing municipal government released the “Beijing 15th Five-Year High-End Industry Development Plan,” outlining 22 concrete policy measures to accelerate new quality productive forces. The plan covers seven key areas: commercial aerospace, full-stack chip autonomy, AI global competition, space applications, new-energy aircraft, aviation key manufacturing, and technical infrastructure development.
Key implementation facts:
- Release date: After-market close, September 9, 2026
- Implementation period: 15th Five-Year Plan (2026–2030)
- Core domains: Commercial aerospace, RISC-V chips, compute-in-memory, large language models, multimodal AI, satellite IoT, embodied intelligence, world models
- Policy tone: Strong directives including “fully push,” “accelerate layout,” and “intensify cultivation”
Commercial Aerospace: Breaking Reusable Rocket Tech, Joining National Constellation Projects
The plan commits Beijing to becoming a national hub for commercial aerospace, with priorities including:
- Breaking reusable launch vehicle and high-thrust engine technologies
- Strengthening satellite and key component batch production and supply capability
- Participating in national large-scale constellation projects
- Developing differentiated commercial constellations
- Promoting “Beidou+” integration and “+Beidou” timing applications
- Developing new-energy aircraft
- Enhancing aviation maintenance and MRO services
A subtle but critical shift emerges: the plan emphasizes both production capabilities and service ecosystem building simultaneously, contrasting with previous local plans that prioritized lab-scale technology demonstrations. This signals the transition from controlled experiments to supply chain integration.
Full-Stack Chip Autonomy: RISC-V and Compute-in-Memory as Strategic Pivots
For technical infrastructure, the plan stresses end-to-end autonomy:
- Elevate chip design to global competitive level via RISC-V, chiplet, and photonic integration platforms
- Advance compute-in-memory architectures and products
- Build a nationally leading integrated circuit manufacturing base with improved yield and service efficiency
- Develop advanced packaging, aligning design and manufacturing
- Establish domestic-leading EDA tool and PDK support capabilities
- Accelerate specialty process and metrology equipment commercialization
RISC-V is an open instruction set architecture viewed as a potential alternative to x86/ARM monopolies; compute-in-memory integrates processing and storage to reduce data movement energy—both critical for overcoming Moore’s Law limitations.
AI Global Competition: From Model Training to Agent Internet Services

The AI section is the longest, targeting leadership in global AI competition. Key initiatives include:
- Strengthening autonomous ecosystem resilience and system software stack capabilities
- Accelerating国产算力 architecture development
- Enhancing algorithm theory; advancing large language and multimodal models; building scientific intelligence and embodied intelligence heights; prioritizing world model research
- Moving intelligent agents toward autonomous execution; developing key standard protocols; constructing engineering software stacks; developing agent internet services
- Enriching model service supply; constructing token economies with production, distribution, and consumption workflows
- Implementing “AI+” initiatives via national AI application pilot bases in healthcare, manufacturing, and education
- Cultivating globally influential open source projects and developer communities
“World model” refers to next-generation AI frameworks capable of causal reasoning about physical systems; “agent internet services” denote distributed networks of cooperating AI agents via standardized protocols—a rare inclusion in sub-national policy.
Practical Implications and Actionable Guidance
For practitioners:
- Invite entrants: Semiconductor equipment/material companies (especially EDA, PDK, advanced packaging), satellite IoT application developers, AI model services and token economy builders, embodied intelligence hardware-software integrators
- Recommend wait: Pure algorithm startups (need application-specific integration), early RISC-V chip designers (require ecosystem compatibility validation), AI application developers lacking public-sector partnerships
Investors should monitor the “national AI application pilot base” construction timeline and “token factory” supply chain opportunities—both already in regulatory implementation phase.
Final Thoughts
Beijing’s plan marks a policy shift from isolated breakthroughs to ecosystem-wide coordination, positioning chips, aerospace, and AI within a unified infrastructure framework. This “full-stack autonomy plus full-domain application” approach responds to global technological competition while reflecting China’s transition from follower to rule-shaper in high-end industry.
