Summary

  • Gorilla says a non-binding LOI covers its proposed first U.S. data-centre acquisition. The operating facility has about 6 MW of current utility capacity; the local utility has provided written support for about 30 MW more, with an indicated delivery timetable of roughly 18 months subject to infrastructure works.
  • Gorilla's target to exceed 100 MW in the United States within 12–18 months combines acquisitions, expansion and customer-led GPU deployments. It is a portfolio ambition, not a disclosure that 100 MW of owned, energized AI capacity is already secured.

Analysis

The headline number is not yet a property portfolio that can be counted site by site. In its October 9 announcement, Gorilla Technology Group Inc. says it has signed a non-binding letter of intent for an operating U.S. data centre. The facility has approximately 6 MW of current utility capacity. The company describes a route to 36 MW in total: the local utility has given written support for about 30 MW of additional power, but has indicated an approximately 18-month delivery timetable, subject to utility and site infrastructure work.

Those are three different states of evidence. The existing 6 MW is current utility capacity at a facility Gorilla has not yet agreed to buy definitively. The additional 30 MW is supported as a path, not described as delivered or energized. The 36 MW figure is the potential total after expansion. None is a published measure of usable IT load or operating GPU compute; the release does not give that conversion.

Timing matters because the company puts its broader U.S. target at more than 100 MW over the next 12–18 months. The expansion path at the first site reaches the far edge of that horizon. That does not make the wider plan inconsistent: Gorilla says the target also depends on further acquisitions, expansion at other infrastructure and customer-led AI deployments. It does mean that the first site's additional power cannot, on the disclosed timetable, explain most of the portfolio target.

The asset has a second, nearer-term proposition. Gorilla says it already has colocation and office operations and generates revenue. The company is still reviewing tenants, the rent roll, lease terms and historical performance, and says continuation of those operations depends on confirming lease and transfer arrangements. The planned purchase therefore combines current property operations with an option to build higher-density AI capacity. Neither the quality of the existing revenue nor the cost and timing of the expansion can yet be assessed from the announcement.

The company says it is evaluating electrical distribution, cooling and network architecture, direct liquid-cooling capability and the complete capital-expenditure programme. A site-power figure alone cannot establish how much computing load the building could support. Nor does an existing colocation business prove that tenants, power rights or revenue will transfer on terms that support an AI conversion.

A dated capital disclosure adds context, but not a project budget. Gorilla's SEC-filed interim statements show USD 227.97 million of group-wide commitments under property-and-equipment purchase agreements as of June 30, net of advances already paid and liabilities already recognized. That figure predates the October LOI and is not allocated to the U.S. facility. It should not be treated as the site's price, debt or funding need; it does make project-level financing and capex disclosure more useful when the company says the wider programme will use dedicated project financing.

The transaction itself remains at an early stage. Gorilla is negotiating definitive documents and targeting completion by the end of November, subject to diligence, internal approvals and closing conditions. The announcement does not identify the site or seller, state a purchase price, disclose lease-transfer terms or name customers for the planned AI expansion. Until those details and the utility milestones are clearer, the 100 MW figure describes a strategic build-out target, not an operating base.

Sources