The factory of the future will no longer need to "buy" a piece of industrial software.

Factories of the future may no longer need to "buy" a set of industrial software.

Don't rush to disagree.
This is not to say that MES, ERP, and PLM will disappear immediately.
What is really being said is:

The way industrial software is produced may be about to change.

How did we build industrial software in the past?
A software company would spend years developing an MES.
Then sell it to a factory.
The factory would then spend months, or even a year or two, on implementation, configuration, and secondary development.
Why was it so heavy?
Because the most expensive part of software in the past was development and customisation cost.
When a factory raised a special requirement, the software company had to change the code.
After the changes, there was still testing, deployment, and maintenance.
So industrial software naturally tended toward being a "big platform."
Because only by pre-building the common needs of many enterprises into the software could the software company amortise development costs.

But AI programming is changing something very important:

The marginal cost of software development is declining.

Functions that a programmer might have taken days to complete in the past can now be produced by AI in hours.
The special business logic of an enterprise that previously required dedicated development by a software company
may in the future become:

"Tell AI what I want, and it generates it directly for me."

01If Requirements Differ, Why Force-Buy a "Standard MES"?

At this point, the question arises.
If one factory's requirements differ from the next,
why force-buy a "standard MES"?

Suppose Factory A says:
"Once my orders come in, I want automatic scheduling based on equipment status, delivery dates, and material inventory."
Factory B says:
"My scheduling rules are different – I also need to consider mould life."
Factory C even says:
"We do small-batch customisation, and there are no fixed scheduling rules at all."

What happened in the past?
Custom development.
So software became more and more complex, and implementation more and more expensive.
But what about the future?
AI may directly reorganise capabilities such as orders, equipment, inventory, and processes according to each factory's own rules,
generating a set of software that belongs only to that factory.

02The Logic of Industrial Software Has Changed

At this point, the logic of industrial software changes.
In the past:
Software companies defined products, and factories adapted to the software.
In the future, it may become:
Factories define requirements, and AI generates the software.

Note!
What is really changing here is not how much code AI writes for programmers.
Rather, it is:

The "software product" itself may transform from a pre-manufactured commodity into something that can be generated at any time.

This is like buying furniture in the past.
You could only go to a furniture store and pick an off-the-shelf table.
Later, custom furniture appeared.
You tell the manufacturer the dimensions and requirements, and they make it for you.
And the AI era may go one step further:

You describe the requirements, and the software grows itself.

03A Very Big Prerequisite

Of course, there is a very big prerequisite here.
AI cannot understand a factory out of thin air.
It needs to know:
what your equipment is;
what your processes are;
how your orders flow;
what your quality rules are;
which things can be done automatically;
which things must be confirmed by people.

In other words,
what will truly matter in the future may not be "code," but depositing the factory's own capabilities.

Turning experience into knowledge,turning processes into rules,turning systems into service interfaces,turning equipment into callable capabilities.Then AI rearranges all of these.

So the biggest change in industrial software in the future is not whether domestic software can replace foreign software.
Rather, we may redefine:
what industrial software is.
It may no longer be an MES, an ERP, or a fixed platform.
But rather:

a set of "its own software" generated by AI from a factory's knowledge, data, capabilities, and services.

So at that point,
will MES still exist?
It certainly may.
But the way it exists may be completely different.

In the next episode, we discuss a more interesting question:
If every factory can ultimately have its own industrial software, will MES really disappear?