How Hut 8 Fully Leased Its 1GW AI Campus Before Launch

Could a data centre be fully leased before it is even fully built?
Hut 8 has shown that, with the right mix of power availability and long-term partnerships, it can.
The company has now fully commercialised its 1GW Beacon Point campus in Texas after securing a second 15-year lease with the same investment-grade customer that signed up for the site's first phase, cementing the campus as one of the largest contracted AI infrastructure developments in the US.
The latest agreement covers 352MW of IT capacity and is worth US$9.8bn over its initial 15-year term.
Combined with the Phase 1 deal, the tenant has now committed to 704MW at Beacon Point, leaving the campus fully contracted against its 1GW utility capacity.
A fully leased AI campus
Beacon Point, located in Nueces County, Texas, has become Hut 8's first fully commercialised AI data centre campus.
Instead of searching for customers after construction, the company has secured long-term agreements before completing development, creating predictable revenue backed by an investment-grade customer.
The second phase will see Hut 8 deliver another 352MW AI factory designed to NVIDIA's DSX reference architecture, supported by 500MW of utility capacity.
Initial delivery of the first Phase 2 data hall is expected in the second quarter of 2028.
- Across both leases, the campus now carries a contracted value of US$19.6bn over the base lease terms.
- If all available renewal options are exercised, the total value could rise to US$50.2bn.
Asher Genoot, CEO of Hut 8, says: "The real test of our power-first approach is what our partners are willing to commit against it.
"Our tenant at Beacon Point chose to double its footprint at the site, the strongest validation an asset can receive.
"We took this greenfield site from first lease to full commercialisation in just months. That speaks to the quality of the sites we originate, the credibility of our delivery, and the long-term orientation of our partnerships.
"The opportunity ahead of us is to apply the same model across our development pipeline."
Why power comes first
The project reflects Hut 8's strategy of securing power infrastructure before locking in compute demand.
Beacon Point's entire 1GW capacity is already covered through an interconnection agreement with AEP Texas, meaning no additional utility capacity is required for the second phase.
The site was originally underwritten to support its affiliated customer American Bitcoin Corp., before being fully leased to an investment-grade AI customer under two separate 15-year agreements.
The company also adapted the campus design during development. The first data hall was redesigned around NVIDIA's DSX reference architecture, increasing IT capacity by 57% without requiring additional land or utility infrastructure.
That redesign helped support the customer's decision to expand its footprint on substantially the same commercial terms as the original lease.
Construction preparations are already underway, with site works in progress and long-lead equipment secured.
Initial energisation remains scheduled for the first quarter of 2027.
Building a larger AI portfolio
The Beacon Point announcement also expands Hut 8's wider AI infrastructure portfolio. The company now has 949MW of contracted AI data centre capacity, including 704MW at Beacon Point and 245MW at its River Bend campus.
Together, those projects represent a cumulative base-term contract value of US$26.6bn and are expected to generate more than US$1.75bn in average annual net operating income once stabilised.
Just as significantly, every AI data centre in Hut 8's contracted portfolio is leased to, or financially backed by, investment-grade counterparties.
That combination of long-term customer commitments and secured utility capacity has become a key part of the company's development strategy.
Beacon Point demonstrates how securing energy infrastructure early can shorten the path from greenfield site to fully contracted AI campus, even before the final data halls come online.


