Skip to content
Tekoälli

Layer 5

Business

Who does what, who pays, and where the money finally settles.

On this page

The same euro passes through several hands. An application company is paid by its customer and pays a model provider, which pays for cloud capacity, which spends on chips, electricity and buildings. The euro does not split evenly along the way, and where it stops is not where the work was hardest.

What happens in this layer

  1. 01

    There are many ways to be paid

    Selling hardware, renting capacity, billing compute by the hour, metered access to an interface, monthly subscriptions, enterprise licences, consulting, and pricing tied to a result.

  2. 02

    Money collects wherever something is scarce

    The largest share goes to whoever holds something hard to copy: the cheapest compute, unique data, access to customers, a deep hook into how the work is done, or a regulator's approval.

  3. 03

    When a layer becomes a commodity, the margin leaves

    As soon as a layer is easy to swap out, its prices fall and the money moves along to the next bottleneck.

  4. 04

    The largest firms sit in several layers at once

    One company can own the building, design the chips, sell the compute, train the models and ship the finished applications.

A common misconception

Inference
Using a finished model. Every question and answer consumes computation, and therefore money.
MLOps
The practices around releasing, monitoring and maintaining models.
Evaluation (eval)
A systematic test of the quality of a model or of a whole application.
Open weights
The model's weights can be downloaded and run on your own hardware.
Sovereign AI
Being able to use and develop a model without someone else being able to stop you.

Articles on this topic

Is there an AI bubble

Four companies will invest 760 billion dollars this year. Meanwhile most companies get no measurable effect on their results. Both can be true.

5 min