Shanghai unveils new smart manufacturing policies — the future of industrial software is about to change.
Shanghai's plan targets 8,000 humanoid robots, 500 advanced smart plants, 300+ agents and 2,000+ industrial AI scenarios by 2028. The real signal: redefining industrial software—from tools that manage production to agents that participate in it, turning process mechanisms and veteran experience into knowledge bases, upgrading control systems. Future vendors sell not software but continuously learn

What Truly Deserves Attention in This Shanghai Document Is Not 8,000 Humanoid Robots
On October 28, Shanghai released a document very much worth the industrial software industry's attention: the Shanghai Action Plan for Accelerating the Development of a New Generation of Intelligent Manufacturing.
Several numbers stand out: by 2028, promote 8,000 humanoid robots into factories, cultivate 500 advanced-level smart factories, launch more than 300 high-level agents, and exceed 2,000 industrial AI application scenarios.
Many people, upon seeing this, may first think that the humanoid robot industry is about to explode.
But what truly deserves attention in this document is not the robots, but rather –
the future of industrial software may need to be redefined.
01.The First Change: Traditional Industrial Software Evolving Toward Software Agents
First, look at a very important detail.
The document explicitly proposes promoting the evolution of traditional industrial software toward software agents, enabling agents to penetrate the core links of manufacturing.
Note that this is different from our past understanding of industrial software.
In the past, MES handled production management, ERP handled business management, and equipment management systems handled equipment maintenance.
These software systems were essentially recording data, managing processes, and providing decision support.
But in the future, software may not only tell you that production progress is lagging, but also analyse the cause, coordinate equipment, materials, and personnel, propose adjustment plans, and even execute directly within the authorised scope.
Industrial software is transforming from a tool for managing production into an agent that participates in production.
02.The Second Change: Industrial AI Begins to Truly Value Process Knowledge
The document specifically proposes building high-value industrial datasets, transforming process mechanisms and expert experience into industrial knowledge bases and knowledge graphs, and also accumulating 10 categories of "master technician" experience datasets.
This direction is very important.
Because the most valuable knowledge in a factory is often not in the data already entered into systems.
For example, with the same machine and the same set of process parameters, why is the product qualified today but a quality problem appears tomorrow?
Why can a master technician judge in advance what is wrong with the equipment based on sound, temperature, and vibration?
This knowledge has largely existed in people's experience.
What needs to be done now is to transform this experience into industrial knowledge that machines can understand, invoke, and execute.
This is the data foundation that industrial AI truly needs.
03.The Third Change: Industrial Control Systems Are Also About to Be Upgraded
The document explicitly proposes developing a new generation of industrial intelligent control products such as PLCs, DCS, SCADA, and SIS, and promoting the integrated evolution of control systems, industrial networks, and computing infrastructure.
This means that industrial AI in the future cannot remain only on computers in offices.
It must gradually enter the equipment site, understand real-time status, and participate in control and execution under safe and reliable conditions.
And behind this, industrial software, control systems, edge computing, and AI models need to work in coordination.
04.The Final Signal: Smart Manufacturing Service Providers Are Divided into Three Levels
Shanghai divides the new generation of smart manufacturing service providers into three levels: single-point implementation, system integration, and intelligence-leading.
It particularly emphasises that future smart manufacturing service providers must understand processes, equipment, software, and AI.
This is actually redefining the competitiveness of industrial software enterprises.
In the past, a software company might survive in a niche market simply by doing MES well and delivering projects well.
But in the future, merely being able to develop software may be far from enough.
You also need to understand the factory's processes, understand the equipment's operating logic, understand the production process, and only then can you truly integrate AI into production.
Of course, Shanghai has not neglected SMEs. The document specifically proposes promoting specialised "small, fast, light, precise" solutions to lower the threshold for SMEs' intelligent transformation.
05.The Truly Profound Industrial Change
So what truly deserves attention in this document is not the figure of 8,000 humanoid robots.
But rather a more profound industrial change:
The factories of the future will compete not just on the degree of equipment automation, but on the degree of intelligence of the entire production system.The industrial software companies of the future should not just deliver a system to the factory, but help the factory build a production capability that can continuously learn, continuously optimise, and continuously evolve.
Here comes the question:
If AI can already understand processes, invoke software, and participate in production decisions, then are future industrial software companies selling software, or selling a new kind of production capability?
What do you think?