Hut 8 signs a 15-year, $9.8bn lease for the first phase of its Texas AI data centre

The Beacon Point lease, anchored by an unnamed investment-grade tenant, lifts Hut 8’s contracted AI capacity to 597 MW and $16.8bn in base-term value. The Bitcoin-miner-to-AI-landlord pivot is now substantially complete.

There is a particular kind of corporate transformation that is supposed to take years, and Hut 8 has done it in roughly 18 months. On Tuesday, the company announced that it has commercialised the first phase of its Beacon Point campus in Nueces County, Texas, through a 15-year lease covering 352 megawatts of IT capacity at a base-term contract value of $9.8bn. With three five-year renewal options included in the contract, the cumulative value could reach approximately $25.1bn if exercised.

The tenant is described as a confidential high-investment-grade company. The lease structure is triple-net. The workloads are AI training and inference. Initial data-hall delivery is expected in the third quarter of 2027.

On those facts alone, this is one of the largest individual data-centre leases announced this year. The deeper structural story is what Hut 8 has, in the past two quarters, become.

From Bitcoin miner to AI landlord

Hut 8’s pivot has been visible for a while. In December 2025, the company signed a 15-year, 245 MW lease at its River Bend campus worth $7bn over the base term. The Beacon Point announcement adds 352 MW more, bringing total contracted AI data-centre capacity to 597 MW and aggregate base-term value to approximately $16.8bn. The company’s compounded contracted revenue line, before any renewals are exercised, now exceeds the entire combined market capitalisation of most of its Bitcoin-mining peers.

What makes the trajectory notable is the speed at which Bitcoin-mining infrastructure has been re-purposed for AI training. What differs is the customer base. Bitcoin pays in volatile cryptocurrency. Hyperscalers pay in 15-year triple-net leases against investment-grade balance sheets.

How the deal is structured

The Beacon Point lease covers the first phase of a planned 1 GW campus, with the remaining ~650 MW available for subsequent commercialisation. The 15-year base term, triple-net structure, and confidential investment-grade tenant designation are the standard features of large hyperscaler-anchored data-centre transactions. Triple-net means the tenant pays operating expenses, taxes, and insurance in addition to base rent, so Hut 8 captures the cash flow without absorbing the operating cost variability.

The renewal-option mechanics are what produce the $25.1bn ceiling figure. Three successive five-year extensions, each at pre-agreed terms, would extend the lease to 30 years and lift cumulative contract value to roughly $25.1bn. Whether those options are exercised is a 2042-and-beyond question. What matters now is that the renewal pricing is locked at deal structures Hut 8 already considers attractive.

The Beacon Point announcement landed alongside Hut 8’s Q1 2026 results. The earnings release confirmed that the company’s transformation is now visible at the financial-statement level: AI data-centre lease revenue is becoming the dominant line, with Bitcoin mining contributing a steadily smaller share of forward earnings.

The geographic choice reflects a wider trend. Texas has emerged as one of the most accommodating US jurisdictions for AI infrastructure, with grid interconnection processes faster than in most coastal markets and zoning regimes more permissive than in Northern Virginia. The same pattern is visible elsewhere on the demand side. Meta’s $13bn El Paso data-centre financing announced last week concentrates capacity in the same state, and Blackstone’s BXDC data-centre REIT IPO last week explicitly identifies Texas as a top-tier market alongside Northern Virginia, Ohio, Maryland, Phoenix, and Austin.

The wider AI-infrastructure context tightens the logic. Thailand’s $29bn data-centre approval cycle led by TikTok this week describes Southeast Asian capacity scaling at hyperscaler-equivalent commitments. Samsung Electronics crossing $1tn on AI memory demand on the same Wednesday Hut 8 announced is the supply-side signal that the AI compute build-out is still tightening rather than slowing. Hut 8’s Beacon Point lease is the operator-side outcome of the same dynamic. Tenants want capacity now, are willing to commit for 15 years, and will pay investment-grade premiums to anyone who can deliver megawatts on the timeline they need.

What are the risks behind the deal? Q3 2027 initial data-hall delivery is an aggressive timeline given the construction, permitting, and grid-interconnection cycles that typically govern projects of this scale, and slippage is a real risk.

The second is tenant concentration. A confidential investment-grade tenant locked into a 15-year lease is a structural strength only as long as the tenant remains creditworthy and committed; the same lease becomes a liability if the underlying workload economics change.

The third is the broader AI-capex durability question. If hyperscaler capex moderates from current $725bn-plus annual run-rates, lease pricing on subsequent Beacon Point phases compresses with it.

None of which is fatal. The 15-year base term insulates Hut 8 from short-cycle volatility, and the triple-net structure pushes operating-cost risk onto the tenant. What changes for Hut 8 over the next 18 months is whether the remaining ~650 MW of Beacon Point capacity attracts comparable terms, and whether the company can maintain the construction velocity its 2027 delivery commitment requires. On Tuesday’s evidence, the pivot from Bitcoin miner to AI landlord is no longer a strategic intent. It is a contracted, nine-figure-revenue reality.