The first build is the slow one, and that is deliberate

Most of the time in a data centre programme is not spent building. It is spent deciding: which pad, which switchgear, which coolant loop, whose drawings, which inspector, which sequence. A second site that re-opens those decisions is not a second site. It is a first site again.

So the first Xalant build is planned as the expensive one on purpose. It is where the specification gets settled, and settling it is most of the work.

What the first build is for

South Oregon is the proof-of-concept market, with build-out targeted for Q1 2027. That build has three jobs, and only one of them is visible from outside.

The visible job is the campus itself: the pad, the hall, the generation beside it and the link out.

The second job is the drawing set. Every decision that gets made once in Oregon is written down in a form the next market can order from, which means a pad layout, a power skid, a cooling loop and a commissioning sequence rather than a folder of photographs.

The third job is the list of things that went wrong. A first build produces that list whether or not anyone writes it down; the difference is whether the second build inherits it.

Why repeatability is a power decision first

A campus that waits on a utility interconnection is not repeatable, because the queue length is somebody else's variable. The first thing a repeat site has to be independent of is the thing nobody can schedule.

That is the reason generation sits on the pad rather than arriving at it. Bridge generation is designed to come up with the hall, and the ground underneath the site — geothermal where the resource is there, a reactor class later where it is not — is designed to replace it without the hall ever coming down. The sequencing is described on the Power pages, and it is a plan rather than a plant.

What we are not claiming

We have not built this. There is no Xalant site anywhere, Xalant is still being formed, and everything above describes a design and a schedule rather than a record.

What we will say is that the design is finished enough to cost, and that the second market is chosen. When ground is actually broken, it will be an announcement on this page with a date on it.

What this means if you are the tenant

You get a boring answer, eventually. The first conversation about a repeat site is about the pad and the load rather than about what is possible, because the specification has already been argued out somewhere else.

You bring a use case. We bring the site, the power, the backhaul and the compute as one contract. Nothing on the pad depends on a public utility.