Gorilla Technology moves on first US data center with 6 MW live site and 36 MW path

The deal

Gorilla Technology has signed a non-binding letter of intent to acquire its first data center in the United States, according to a dev.to report on the company's October 9 announcement. The facility is an operating colocation site with roughly 6 MW of current utility capacity and what the company describes as a path to 36 MW of total site power.

The utility has provided written support for additional capacity, dev.to reports, on a delivery timetable of roughly 18 months that remains subject to infrastructure works. The acquisition has not closed: Gorilla is targeting completion by the end of November, subject to due diligence, approvals and definitive agreements. Its review covers existing tenants and future direct liquid cooling. As dev.to points out, these are company-reported plans rather than evidence of commissioned AI capacity.

An attractive starting point with strings attached

An operating colocation facility offers a ready-made entry into AI infrastructure: it already has power, tenants and revenue. It also arrives with maintenance obligations and equipment that must keep working while the site is rebuilt around it. The dev.to analysis frames the practical question for this class of deal: how a live facility moves through the transition without an expansion turning into a service incident.

Map the obligations already inside the building

The analysis argues that a brownfield AI project should begin with a tenant and dependency map, because a floor plan alone misses the constraints that determine when work can happen. That means identifying which customers share electrical distribution, cooling systems, network paths and maintenance access, and where work in an apparently unused area could affect an occupied hall. It also means recording who approves an outage, how much notice is required and which changes have to wait for a customer migration.

Commercial and engineering diligence need to meet here: a lease may promise a level of service continuity that a proposed construction sequence cannot yet deliver. Finding that conflict before committing to a deployment date gives the project team a chance to redesign the work.

Define an AI phase around a usable operating envelope

Total site power is useful for describing a development opportunity, but a deployment team needs a narrower boundary: the electrical and thermal conditions actually available to its racks, under both normal operation and agreed failure scenarios. A practical phase definition, per the analysis, names the racks being released, their supported load, the cooling configuration and any equipment still shared with existing tenants, and states the redundancy assumptions behind it. A rack that only runs while every component is healthy has a very different service envelope from one that can ride out a planned maintenance event.

For liquid cooling, procurement should follow an agreed interface specification. Ownership of the facility water connection, coolant distribution units, leak detection and emergency response needs to be assigned, along with responsibility for water quality and inspection records, all before delivery teams arrive with equipment built to different assumptions.

Isolate construction and attach evidence to milestones

A live-site upgrade needs explicit separation between construction and operations, covering access routes, dust control, temporary services and approval of changes to shared infrastructure. The useful acceptance test is broader than switching on new racks: the new phase must enter service while existing customers remain within their contracted operating conditions, and the monitoring and escalation paths should be tested alongside the physical systems. The analysis also stresses defining who can stop work, who can authorize a restart and what evidence must be captured before handover.

Commercially, it recommends splitting the transaction, utility works, site upgrades and customer acceptance into distinct milestones, each with an owner, a dependency and an observable completion condition. That gives compute customers a clearer basis for contracting capacity reservations, equipment shipments and workload migration around the phase that is actually ready, and exposes schedule conflicts early enough for the operator to adjust the plan.

Why it matters

Gorilla's proposal is a small deal that illustrates a large pattern: as AI demand presses against new-build capacity and power availability, buyers increasingly look at existing facilities. Retrofitting a live data center for AI racks, with denser power and direct liquid cooling, is as much an operational discipline as an engineering one. The distance between a stated path to 36 MW and deliverable AI capacity is filled with tenant contracts, shared infrastructure and construction sequencing, and those constraints are best mapped before deployment dates are set rather than after.