MIIT unveils 24 innovation tasks — machines are starting to understand industry

On July 29, seven ministries including MIIT launched the 2026 Innovation Task Challenge for industry and information technology.
On the surface, it covers 24 technical areas.
Including robotics, embodied intelligence, brain-inspired intelligence, and on-device world models.
But if you only see these keywords, you might miss the point.
The real signal being released is:

Future industrial competition is no longer about building stronger machines, but about giving machines stronger cognitive capabilities.

Over the past decades, industrial development has essentially been solving one problem:
how to make machines work faster.
So we developed automation, robotics, digital systems – PLCs controlling actions, MES managing production, ERP managing resources.
These systems solved: how machines execute, how processes are optimised, how production is organised.
But one question has remained unanswered:
Does the machine itself know why?

Take a machining centre, for example.
It has been running for 5 years.
Processing thousands of parts every day.
It generates: vibration data, temperature data, current data, processing parameters, alarm logs, maintenance records.
These data contain a vast amount of experience.
But in the past, where did this experience reside?
In the minds of master technicians, in maintenance engineers' notebooks, in equipment vendors' decades of after-sales experience.
The equipment itself did not know.

Now AI is changing a fundamental logic:
In the past: humans possessed knowledge, machines executed commands.
In the future: machines will possess knowledge, machines will assist decision-making.
This is why one particularly critical direction in this task list is: on-device world models.
Many people associate world models with robots.
But what truly matters in the industrial field is: enabling equipment to begin understanding its own state.

A smart piece of equipment in the future may not wait for a failure to occur before alarming.
Instead, it will tell you before failure happens:
"Based on the past 10 years of operational data, this bearing has now deviated from normal status."
"If the current operating conditions continue, an anomaly may occur in 15 days."
"Adjusting this parameter can reduce risk."
At this point, the equipment is no longer just a production tool.
It begins to become: an industrial partner with experience.

01Industrial Competition Is Shifting

So why is there such a concentrated push in these directions in 2026?
Because industrial competition is undergoing a shift.
In the past, Chinese manufacturing competed on: equipment quantity, production scale, and manufacturing cost.
But in the future, Chinese manufacturing will compete on: industrial intelligence density.
What does this mean?
Within a factory, how many pieces of equipment can perceive? How many can understand? How many can optimise autonomously?

This is also why industrial software will become increasingly important.
Because the biggest challenge for AI entering industry has never been algorithms.
Rather, it is: where is the industrial data? Where is the industrial knowledge? Where are the industrial scenarios?

Without equipment data, large models are just chat tools.Without industrial knowledge, AI cannot enter the production site.

02A New Watershed in Manufacturing

In the coming years, manufacturing may see a new watershed:
Some enterprises own a large amount of equipment.
But the equipment is just assets.
Other enterprises also own a large amount of equipment.
But every piece of equipment has its own data, knowledge, and intelligence.
The competitiveness of these two types of enterprises will be completely different.

Therefore, what truly deserves attention in MIIT's 24 innovation tasks is not just robotics.
Rather, it is a deeper shift:
Industry is moving from the automation era to the cognitive era.
The most advanced factory of the future may not be the one with the most machines.
Rather, it will be the one where:
every machine understands itself more and more.

Because the next round of manufacturing competition is not, in essence, human versus machine.It is: who can give machines knowledge.