Rockefeller did not win the oil age by drilling the most wells.

He won by controlling the pipelines, the refineries, and the distribution.

The wells were interchangeable. The rails were not.

I thought about that this week when Ornn, a startup founded by two MIT grads, raised $33 million led by Andreessen Horowitz to build a commodities market for AI compute.

Not to buy GPUs. To build the layer that prices them, trades them, and hedges them.

Their index is already on the Bloomberg Terminal, and Intercontinental Commodity Exchange has announced compute futures tied to it.

The smartest money in tech is betting that the durable value in AI sits in the connective layer, not the raw capacity.

Capacity commoditizes. Rails compound.

Now look at how many law firms are approaching AI.

A lot of energy is going into picking wells. Which model, which point solution, which vendor's demo looked best last quarter.

Those tools matter, but they should be the most interchangeable part of the stack.

Two years from now, the underlying models and platforms will have leapfrogged each other a dozen times.

What will not be interchangeable is the connective layer. How AI reaches your firm's knowledge/data.

How it moves through your workflows. How it shows up for your clients.

The firms building that layer are accumulating an advantage that survives every model/platform release.

Firms buying point solutions think they are renting wells. They are actually renting pipes.

Every prompt library, workflow, and work product that lives inside a vendor's closed platform is plumbing the firm does not own.

The model underneath will be swapped out dozens of times. The plumbing is what you cannot leave.

So here is the question to ask about every AI dollar in your 2027 budget: are we buying a well, or are we building pipes?