Will FDE become a form of high-end outsourcing?

If an FDE writes custom code for customers every day, how is that different from premium outsourcing?
This may be the biggest trap of the FDE model.

Previously, we have been saying: FDEs understand AI, understand business, and understand the field.
It sounds extremely scarce.
But here comes the problem:one customer, one requirement, one factory, one solution, one project developed from scratch each time.
The more customers, the more FDEs must be hired.
What happens in the end?

More and more people, yet profits increasingly resemble software outsourcing.

So a truly excellent FDE cannot merely "solve the customer's problem."
They must also ask:

Can this problem be solved a second time?

For example, a factory needs equipment failure prediction.
The FDE might spend three months the first time:
connecting equipment data, organising maintenance records, training models, building a prediction system.
The project is complete.
If, at the second factory, they start all over again, that is merely project delivery.
But if the FDE abstracts everything accumulated from the first project: data models, fault knowledge, Agent capabilities, interface standards, and abstracts them all out.
The second factory only needs configuration, not redevelopment.
At this point, the FDE truly begins to create value.

So the true endpoint for an FDE is not "finishing the project," but "turning the project into a product."

This is also the biggest difference between FDEs and traditional outsourcing.

Outsourcing solves: "What does the customer need this time?" FDEs should truly solve: "Is the problem behind this customer actually an industry problem?"

01An Interesting Flywheel

This creates a very interesting flywheel:
The first customer, solves one problem.
The second customer, reuses part of the capability.
By ten customers, industry know-how begins to form.
By one hundred customers, a true industry product may ultimately emerge.

So the strongest FDE team in the future
may not be the team with the most engineers.
But rather:

every customer they serve makes the next customer easier.

This is the true scaling path for FDEs.

02Especially Important for Industrial Software Companies

And this is especially important for industrial software companies.
Because manufacturing appears to be "one factory, one way,"
but in reality:
equipment management, quality inspection, production scheduling, predictive maintenance – behind them lie a large number of common problems.
Whoever can extract these common problems from projects
will have the opportunity to turn: project capability into product capability.

So FDE is not an endpoint.
It is more like a road.
One end connects to the customer site, the other connects to product R&D.
And truly strong companies
will accumulate continuously along this road, ultimately forming their own industry barrier.

03The Only Criterion for Judging Whether an FDE Company Has a Future

So to judge whether an FDE company has a future,
only one question needs to be asked:

With every project it does, is it adding people, or adding assets?

If it is only adding people, it ultimately remains outsourcing.
If every project
adds products, data, knowledge, and industry experience, then it truly has a future.

In the next issue, we discuss the final and most important question of this series:
In the next five years, will FDE become a mainstream profession in the AI era?