KEEL Moses Lake Site

Moses Lake is Keel's 18 MW gross development in Washington, with a 2027 service target and a modular infrastructure programme involving Vertiv. Its current status differs materially from the old article's description of an AI retrofit inside an existing mining building: Keel reports that the former building has been demolished.

That changes how the physical project should be understood. The existing electrical position and location can remain valuable, but the demolished structure cannot be credited with supporting the new racks or reducing the cost of a building conversion.

Open the Keel Data Center Tracker to compare the location with the rest of the portfolio. The new Keel Stock Forecast 2030: The Price of Patience separately examines the financial assumptions behind the development story.

The project has moved beyond mining

Keel's Q2 presentation describes mining as fully decommissioned, the previous building as demolished and development with Vertiv as in progress. It identifies zoning approval, with land-development and environmental permits still progressing. The earliest ready-for-service year remains 2027. Keel Q2 development presentation.

The August business update also reports receipt of the first Vertiv modules. Delivered equipment is an achieved procurement milestone. It is not the same as installed, commissioned equipment or a completed tenant handover. Keel Q2 operating update.

What the 18 MW measures

The presentation labels the site's capacity as gross power. Customer IT load will depend on the delivered design, cooling overhead and contractual configuration. Applying a rental rate quoted per kilowatt of IT capacity to the full gross figure would skip that conversion.

The site's modest scale can make it an important execution test without making it representative of every future campus. A completed and leased first project would establish a record of delivery. It would not prove that a much larger site has the same cost, schedule or tenant demand.

Infrastructure procurement is not a GPU purchase

Keel associates Moses Lake with a $128 million Vertiv agreement for critical infrastructure. The older study used that commitment in broader owned-compute comparisons. The agreement should not be treated as evidence that a complete GPU fleet is included or that every future project will follow the same ownership model. Keel's project and financial materials.

The historical calculations below remain unchanged, but their labels make this limitation explicit. A project procurement commitment is a specific input. Scaling it across an entire portfolio requires additional assumptions about buildings, power systems, tenant fit-out and equipment ownership.

September backup-power agreement

On September 9, PowerSecure announced an agreement to design and deliver a modular standby-power solution for Moses Lake. The planned system includes backup generation, controls and electrical infrastructure, with peak-shaving capability. The supplier announcement does not disclose completed commissioning, a new tenant, a rated MW addition or a delivery date. It advances the procurement record without establishing that the campus is operating.

What determines the next stage

Moses Lake needs the remaining approvals, completed installation and a commercial agreement aligned with the delivered facility. Keel's August update described active negotiations across its priority sites. Negotiations should remain distinct from signed leases and rent commencement.

The important next disclosures are therefore a lease, notice to proceed where applicable, completed infrastructure and tenant acceptance. Those would progressively narrow the uncertainty around the 2027 service target.

The Keel company page provides the wider research. This factual study was reviewed through September 17, 2026.

Historical calculations retained from the original study

The following limited examples are preserved from the original publication. Their assumptions and outputs have not been revised. They are not current operating measurements, a new Northwise forecast or a statement of current contracted prices. The factual discussion above uses the latest reviewed evidence.

The original examples extrapolate project commitments and hypothetical per-MW costs. The Vertiv critical-infrastructure agreement does not establish inclusion of a complete owned GPU fleet. These examples retain the original assumptions rather than validating them as current project economics.

Shell-style development, where KEEL prepares land, power, utilities, cooling, and site infrastructure and then leases to a tenant, runs roughly $3M to $7M per MW in illustrative terms. At that range, developing the 430 MW secured power base alone implies $1.3B to $3.0B in total capital need. The full 2.2 GW pipeline sits in an entirely different category, somewhere between $6.6B and $15.4B depending on development approach and cost per MW.

Owned-compute development, where KEEL also owns the AI hardware placed inside its sites, is heavier still. The Moses Lake project itself carries a disclosed $128M funding package for 18 MW, or roughly $7.1M per MW. That figure already signals that Moses Lake is not being treated as a minimal shell build. At a broader illustrative range of $15M to $25M per MW for owned-compute strategies, developing even the secured power base implies $6.5B to $10.8B. The full pipeline at those figures runs to $33B to $55B.

The $128M Vertiv commitment represents a substantial portion of current liquidity allocated to a single 18 MW site. At roughly $7.1M per MW, Moses Lake is being developed at a cost intensity that reflects genuine infrastructure depth rather than a minimal shell preparation. That level of per-MW investment supports the colocation and powered-shell leasing pitch but also concentrates near-term capital risk on a single site’s execution and leasing outcome.