August 11, 2026 (InvestinChina.asia) – The Shanghai Municipal Commission of Economy and Informatization has released the “15th Five-Year” Plan for Software and Information Services, setting a target of building the city’s software and IT sector into a 4-trillion-yuan (approximately US$560 billion) industrial powerhouse by 2030, with AI positioned as the overarching driver across every segment of the value chain.
The plan—the most ambitious blueprint for the sector since Shanghai elevated software and information services to its third-largest pillar industry in 2025—raises the previous 2030 revenue guidance from around 3 trillion yuan to a stretch goal of 4 trillion yuan, with industry value-added surpassing 1.1 trillion yuan. The number of companies exceeding 10 billion yuan in revenue is targeted to reach 35.
AI at the Helm: Seven Tracks, One Backbone
The plan explicitly puts artificial intelligence at the center of seven digital transformation tracks: the internet, software, digital content, cybersecurity, intelligent computing cloud, fintech, and industrial services. Beyond integration, it carves out six frontier arenas where Shanghai aims to lead:
- Agent Economy: Developing general-purpose agents and multi-agent collaboration systems, and promoting “Agent-as-a-Service” as a new commercial model, including testing, certification, marketing, managed operations and compliance consulting.
- AI Gaming: Breakthroughs in multi-agent collaborative decision-making, dynamic environment generation, real-time emotional interaction, long-horizon memory reasoning and accelerated physics simulation.
- Next-Generation Device Software: Purpose-built operating systems, control software and high-end embedded software for embodied intelligence, low-altitude equipment, smart wearables and brain-computer interfaces.
- Virtual Reality: Lightweight terminals, real-time spatial rendering, and blended virtual-physical interaction, deeply coupled with large models and domain knowledge bases.
- Native Security: A “model-defined security” architecture, with breakthroughs in bottom-layer secure inference, full-chain attack attribution and adversarial defense.
- Physics Simulation: “World models” capable of physical cognition, real-time simulation and autonomous iteration, deployed in embodied intelligence, aerospace and biomedicine.
Hard Tech: Non-Transformer Architectures and Compute Hardware
In a notable departure from mainstream AI development, the plan calls for exploring non-Transformer architectures—including state space models, recurrent neural network variants and liquid neural networks—alongside frontier foundations such as physical intelligence, world models, quantum intelligence and brain-inspired intelligence. This signals Shanghai’s ambition to diversify beyond the dominant Transformer paradigm and reduce dependence on foreign-originated architectural IP.
On the hardware side, the plan outlines a sweeping “hundred-thousand-ten thousand” intelligent computing cluster project:
- 100,000-card scale: Ultra-large intelligent computing clusters in Songjiang, Lin-gang and Qingpu.
- 1,000-card scale: Clusters in Baoshan, Pudong and Jiading.
- 100-card edge: Traditional data centers and telecom rooms will transition to edge intelligent computing centers.
The plan further targets breakthroughs in ultra-large-scale intelligent computing cluster networking, and explicitly names the core hardware supply chain: GPU/NPU high-performance computing chips, QPU (quantum) chips, CPO (co-packaged optics) high-speed interconnect, HBM (high-bandwidth memory), and heterogeneous servers. It commits to deepening the integration of domestic chips with mainstream large models.
Basic and Industrial Software: Closing the Gap
The plan emphasizes a full-stack assault on foundational software: OS kernels, graphics engines, SQL engines and high-concurrency processing, plus performance upgrades for operating systems, databases, middleware and tooling software, backed by a hardware-software adaptation and verification system.
For industrial software—long a chokepoint for Chinese manufacturing—the plan calls for breakthroughs in geometric modeling, constraint solving, mesh generation and physics simulation, and accelerated R&D in CAD, CAE and EDA, supported by an intermediate-test verification system for rapid product iteration.
In a forward-looking move, the plan also encourages software enterprises to adopt coding platforms, build intelligent R&D production lines across the full workflow, and explore counting tokens, data governance, and large-model deployment and retraining as eligible inputs for technical transformation subsidies.
Spatial Layout and Ecosystem
To support the expansion, Shanghai will construct a “one ring, two poles, three spokes” (一环两极三射) industrial spatial layout, optimizing enterprise cultivation, open-source ecosystems and industrial base systems. The plan also innovates on compute as a factor of production—advancing market-oriented reform of compute resources and exploring models such as using compute power as equity investment.
Why It Matters for the AI Industrial Chain
The plan’s significance extends well beyond Shanghai’s municipal boundaries. By naming specific hardware components—GPU/NPU, QPU, CPO, HBM—and committing to non-Transformer architectural exploration, the document reads as a de facto industrial chain roadmap that benefits domestic suppliers across chips, optical interconnect, memory, EDA, industrial software, embodied intelligence and agent platforms.
The raised 4-trillion-yuan target, combined with explicit support for compute-as-a-factor-of-production and token-based R&D subsidies, provides rare policy visibility into 2030 for investors across the AI value chain. For context, the sector’s ascent to Shanghai’s third-largest pillar industry in 2025—surpassing both real estate in 2024 and wholesale/retail in 2025—underscores the strategic weight now attached to software and information services in China’s commercial capital.