A company that was operating Bitcoin mining farms two years ago now controls nearly a gigawatt of contracted AI data center capacity, underpinned by nearly $27 billion in long-term revenue — and it just finished filling the last available lease on its largest campus in under a year.
Hut 8 Corp. announced July 20 that it has signed a second 15-year, $9.8 billion lease at its Beacon Point AI data center campus in Nueces County, Texas. The deal covers 352 megawatts of IT capacity and was executed with the same unnamed investment-grade tenant that signed Beacon Point's Phase 1 lease in May 2026, doubling that customer's contracted footprint at the campus to 704 MW. Combined with the Phase 1 lease, the two agreements bring Beacon Point's total base-term contract value to $19.6 billion. If the tenant exercises all three five-year renewal options under each lease, that figure could reach $50.2 billion.
Hut 8 shares jumped as much as 17% on the news, according to Stocktwits coverage of the announcement.
Beacon Point Fully Spoken For
Beacon Point spans 525 acres near Corpus Christi and is secured by a 1,000 MW utility interconnection agreement with AEP Texas. The two leases together cover 704 MW of IT capacity against that power base. Hut 8 describes the remaining capacity as structural headroom rather than open inventory. Both agreements are structured as triple-net (NNN) leases — meaning the tenant pays operating costs beyond base rent — and carry a 3% annual base rent escalator baked into the lease terms.
The second lease is projected to generate average annual net operating income of $655 million on its own, raising combined campus-level average annual NOI to $1.31 billion. Across Hut 8's broader AI data center portfolio — which also includes the River Bend campus in West Feliciana Parish, Louisiana — total contracted IT capacity has risen to 949 MW, supported by 1,330 MW of utility capacity. Aggregate base-term contract value across the portfolio now stands at $26.6 billion, with average annual NOI exceeding $1.75 billion. Phase 1 energization remains on schedule for Q1 2027, with initial Phase 2 data hall delivery expected in Q2 2028.
How NVIDIA DSX Architecture Enables 57% More Compute in the Same Footprint
The technical story behind Beacon Point's commercial success starts before either lease was signed. Hut 8 originally underwrote the site on a speed-to-power thesis to serve American Bitcoin Corp., a company affiliate, before pivoting the asset toward AI leasing. Rather than simply swapping one tenant for another, Hut 8 redesigned the first data hall to conform to NVIDIA's DSX reference architecture — formally the NVIDIA Omniverse DSX Blueprint for gigawatt-scale AI infrastructure.
DSX is a full-stack blueprint that covers how a gigawatt-scale AI factory should be physically designed, digitally validated, and operationally managed. Its two principal configurational frameworks are DSX Boost, which adjusts power management and workload distribution at the data center level to achieve either approximately 30% lower power consumption or approximately 30% greater GPU density per megawatt, and DSX Flex, which links the data center to regional power grids to balance supply and demand dynamically by tapping roughly 100 GW of under-utilized grid capacity.
Before a single steel beam goes up, the DSX platform's simulation component — DSX Sim, powered by NVIDIA Omniverse — builds a physically accurate digital twin of the entire AI factory. Designers validate cooling loops, power delivery topology, and GPU rack configurations virtually, catching engineering conflicts that would otherwise surface as costly construction changes. On the power delivery side, DSX Flex models millisecond-scale power spikes from the current NVIDIA Rubin GPU generation against the facility's power network, ensuring circuit breakers do not trip during peak training runs — a documented failure mode in conventionally designed facilities. Details on how Radiant integrates Omniverse DSX for infrastructure clients illustrate how this simulation-first approach functions in practice.
The practical result at Beacon Point: Hut 8's redesign of the Phase 1 data hall to DSX enabled 57% more IT capacity within the same land and utility footprint as the original Bitcoin-mining layout. That density gain is what makes the $9.8 billion lease economically viable at the stated power base. A data center that fits more compute into each megawatt of utility power generates proportionally higher NOI per dollar of infrastructure cost — the exact tradeoff that allows Hut 8 to offer long-term leases at rates that investment-grade tenants are willing to commit to for 15 years.
Vertiv, Hut 8's thermal and power infrastructure partner, built the campus's cooling systems to support liquid cooling at densities exceeding 140 kW per rack. That specification — roughly ten times the density of a conventional cloud rack — is required to remove heat from NVIDIA's current GPU generation in a facility this size. Construction and engineering management are handled by Jacobs, the same partner Hut 8 used at River Bend.
How the Non-Recourse Bond Model Creates a New Template for the Sector
The less-visible but structurally significant development in the Beacon Point story is how the campus is financed — and what that model implies for AI infrastructure development broadly.
In June 2026, Hut 8 closed a $4.25 billion offering of 6.129% senior secured notes due 2042, issued through its wholly-owned subsidiary Beacon Point DC LLC. Moody's assigned a Baa2 rating to the notes — one notch above the BBB− that S&P Global Ratings and Fitch Ratings assigned to Hut 8's earlier River Bend financing — and the offering was substantially oversubscribed. At T+165 basis points over Treasuries, the notes priced 20 basis points inside the River Bend spread.
Three structural features of the financing matter for understanding what Hut 8 is building:
The notes are non-recourse to Hut 8's corporate entity. Beacon Point DC LLC is a ring-fenced subsidiary; if the project were to fail, lenders' recourse is limited to the project assets and cash flows — not Hut 8's balance sheet. That means Hut 8 is constructing a $4.25 billion AI campus without adding any construction debt to its corporate financial statements.
The notes are non-dilutive. Hut 8 funded the construction of both Beacon Point phases and River Bend without issuing new equity — a constraint that forced the company to find tenants with creditworthiness strong enough to support investment-grade project financing on the merits of the lease alone.
The Beacon Point notes represent the largest, tightest-priced, highest-rated investment-grade bond in a single-sponsor data center construction financing on record at the time of issuance — a standard typically reserved for operational infrastructure assets that are already generating revenue, not projects still in the ground. Combined, the River Bend and Beacon Point financings have raised $7.5 billion in investment-grade project-level construction financing — a cumulative figure that no prior data center developer had matched at the construction stage.
For the broader AI infrastructure sector, the implication is structural: developers who can secure long-term leases with AA-or-higher-rated tenants can now access the bond markets during construction, funding hyperscale campuses without equity dilution or corporate balance sheet risk. That template, if repeatable, lowers the cost of capital for an entire category of infrastructure development.
From Crypto Mining to AI: How Power Assets Became the Pivot
Hut 8 began as a Canadian Bitcoin mining operation — for years one of North America's largest by installed capacity — before systematically repositioning its power infrastructure toward AI and high-performance computing as generative AI accelerated demand for large-scale compute. The transition reflects a structural advantage Bitcoin miners accumulated: years of negotiating high-voltage power agreements, building industrial cooling systems, and managing large power loads — the same infrastructure that AI data centers require.
The River Bend campus in West Feliciana Parish, Louisiana, was the first proof point of the pivot. In December 2025, Hut 8 signed a 15-year, $7.0 billion lease with Fluidstack for 245 MW of IT capacity at River Bend, backed by a Google financial guarantee covering all lease payments — a guarantee that was itself critical in enabling J.P. Morgan and Goldman Sachs to underwrite the project financing at favorable terms. That transaction was also part of a broader arrangement under which Hut 8 and Anthropic could develop up to 2.295 GW of AI data center infrastructure using Fluidstack-operated clusters.
The River Bend deal established Hut 8's credibility as a developer capable of attracting blue-chip counterparties. The Beacon Point Phase 1 deal — signed May 2026 — carried the same financial structure but without a Google backstop, relying instead on the tenant's own AA-or-higher credit rating to satisfy project financing requirements. The fact that the same tenant returned for Phase 2 on substantially identical terms, without negotiating a lower rate or requesting structural changes, was interpreted by Hut 8's CEO as the clearest possible signal that the product delivered in Phase 1 met the tenant's expectations.
"Our tenant at Beacon Point chose to double its footprint at the site, the strongest validation an asset can receive," Hut 8 CEO Asher Genoot said in the July 20 announcement. "We took this greenfield site from first lease to full commercialization in just months."
Are Crypto Miners Really Building the AI Infrastructure Layer?
The July 20 announcement did not arrive in isolation. On the same day, fellow former crypto miner IREN raised its year-end 2026 AI Cloud annualized revenue target from $3.7 billion to more than $4 billion, after signing $2.8 billion in new customer contracts with names including Microsoft, NVIDIA, Perplexity, and Figure AI. Both stocks climbed double digits on Monday, with IREN gaining as much as 19%.
The parallel announcements point to a structural pattern. Bitcoin mining operations accumulate large contiguous power blocks — grid-connected, cooling-equipped, often in power-advantaged geographies — at a time when the dominant AI developers cannot secure equivalent power at the pace they need it. The economics reinforce the pivot: AI data center leasing can generate between three and 25 times more revenue per megawatt than Bitcoin mining at current hashprices, with project-level EBITDA margins potentially reaching 80–90% on fully contracted, triple-net campuses.
What is not yet clear is whether the anonymous tenant — whom Hut 8 describes only as an investment-grade company rated AA- or higher — represents one of the handful of hyperscalers with the balance sheet to commit $19.6 billion to a single campus, or a major AI laboratory with a financial backstop from a hyperscale backer. At 704 MW of contracted IT capacity, the scale places the tenant among a very small group of organizations worldwide that can absorb that level of compute infrastructure.
What Comes Next for Hut 8's Pipeline
With Beacon Point fully contracted, Hut 8's development focus shifts to construction execution: delivering Phase 1 on schedule for initial energization in Q1 2027, then advancing to Q3 2027 preliminary delivery, and completing Phase 2 in Q2 2028. The company has 830 MW of capacity currently under construction across its broader portfolio and another 550 MW in active development.
Long-lead equipment for Beacon Point Phase 2 was procured before the lease was signed, and site preparation is underway. The company has also flagged a potential expansion pathway at River Bend — where Fluidstack holds a right of first offer for up to 1,000 additional MW — and an optional joint-development arrangement with Anthropic covering up to 1,050 MW of additional capacity across Hut 8's pipeline beyond River Bend.
Frequently Asked Questions
What is the NVIDIA DSX reference architecture, and why does it matter for AI data centers?
NVIDIA's Omniverse DSX Blueprint is a full-stack design standard for gigawatt-scale AI factories. It specifies how power should be managed, how cooling should be integrated, and how the complete facility should be validated as a digital twin before construction begins. Its DSX Boost framework can squeeze roughly 30% more GPU density out of a given megawatt of power or achieve 30% lower power consumption at the same compute throughput. For developers like Hut 8, adopting DSX translated directly into 57% more IT capacity within the same land and utility footprint compared to the prior Bitcoin-mining layout — a density gain that is the engineering basis for the $9.8 billion lease value.
How can Hut 8 finance a $4.25 billion data center without putting its corporate balance sheet at risk?
Hut 8 finances each campus through a separate ring-fenced subsidiary — Beacon Point DC LLC for the Texas campus, a similar entity for River Bend — that issues investment-grade senior secured notes backed solely by the project's contracted lease revenue. The notes are non-recourse to Hut 8's parent company, meaning that if the project encountered difficulties, lenders could not claim against Hut 8's corporate assets. This structure, known as project finance, is standard for toll roads and utility infrastructure; what is new here is achieving investment-grade bond ratings (Baa2 from Moody's) at the construction stage, before the campus generates a dollar of revenue. The key enabler is the investment-grade credit quality of the tenant, whose long-term lease commitment gives bond investors the same confidence they would normally require from an operating asset.
Who is the mystery tenant at Beacon Point, and why won't Hut 8 name them?
Hut 8 has not disclosed the identity of the tenant and describes it only as an "investment-grade company rated AA- or higher." Anonymity in large-scale data center leasing is standard practice. Hyperscalers and major AI laboratories typically prefer not to telegraph large infrastructure commitments before facilities come online, because doing so can affect their negotiating position with other developers and signal competitive strategy to rivals. At 704 MW of contracted IT capacity, the tenant's scale places it among a small number of organizations globally that could absorb this level of compute infrastructure — most likely one of the major hyperscalers or a frontier AI laboratory with hyperscale financial backing.
What does Hut 8's Beacon Point deal mean for other AI infrastructure developers?
The most significant precedent is not the lease itself but the financing structure. By demonstrating that a single-sponsor, construction-stage data center can access the investment-grade bond market — raising $4.25 billion at Baa2 and attracting substantially oversubscribed demand — Hut 8 has established a replicable template. Any developer that can sign long-term leases with AA-rated tenants can now, in principle, fund hyperscale campuses through non-recourse project bonds without diluting shareholders or leveraging the corporate balance sheet. That structural change could meaningfully lower the cost of capital for the next generation of AI infrastructure and accelerate the pace at which new gigawatt-scale campuses can be financed.
Related Articles