The headline says 1 GW. The lease monetizes 704 MW of contracted IT capacity.
That difference is not a footnote. In fact, it is the first lesson infrastructure buyers should take from Hut 8’s Beacon Point development in Texas.
That difference is far from a mere footnote—rather, it is the first key lesson infrastructure buyers should take from Hut 8’s Beacon Point development in Texas.
On July 20, 2026, Hut 8 announced a second 15-year lease covering 352 MW of IT capacity at its Beacon Point campus in Nueces County. Under the terms of the deal, the agreement has a stated base-term contract value of $9.8 billion, effectively doubling the tenant’s contracted capacity at the site to 704 MW.
Together, the two leases bring the campus-level base-term contract value to $19.6 billion. Hut 8 expects the fully stabilized campus to contribute an average of approximately $1.31 billion in annual net operating income during the base term.
The campus has 1,000 MW of utility capacity secured through an interconnection agreement with AEP Texas. Crucially, however, that does not mean the site contains 1,000 MW of contracted servers. In technical terms, utility capacity, facility capacity, and IT capacity are fundamentally different metrics.
For this reason, one should never use the gap between 1,000 MW of utility capacity and 704 MW of IT capacity to calculate PUE directly. This is because interconnection capacity can include facility loads, cooling, electrical losses, redundancy, operating headroom, and future design considerations. Ultimately, a reliable PUE calculation requires actual energy measurements under defined operating conditions.
Field Note: Ask every data center developer for three separate figures: secured utility capacity, deliverable facility capacity and contracted IT capacity. If these numbers are combined into one headline, the financial model is not ready.
Power Has Become the First Product
Beacon Point was initially underwritten to serve Hut 8’s affiliated Bitcoin mining customer, American Bitcoin Corp. It is now fully contracted under two long-term AI leases with an investment-grade counterparty.
The ASIC fleet was not the most transferable asset.
The transferable assets were the power position, the land, the utility relationship, the development team and the ability to deliver energy-intensive digital infrastructure.
This is the logic behind Hut 8’s “power-first” development model. A site is secured before its final compute use is fixed. Once power and land are controlled, the developer can evaluate which workload produces the best risk-adjusted return: Bitcoin mining, AI training, cloud infrastructure, high-performance computing or another form of industrial compute.
Go power-first, but value power at the date it can actually be energized.
A utility agreement without a realistic energization schedule is not equivalent to operating capacity. Beacon Point’s announcement includes specific delivery milestones: initial energization remains scheduled for the first quarter of 2027, while the first Phase 2 data hall is expected in the second quarter of 2028.
The timeline also shows why long-lead procurement matters. Hut 8 stated that site preparation was underway and critical long-lead equipment had already been procured.
For mining infrastructure owners, this changes the definition of a valuable site. The market no longer values only low electricity prices. It increasingly values:
Firm and contractually secured power
A credible energization date
Expandable land
Transmission and substation access
Fiber availability
Permitting progress
Cooling feasibility
Proven construction and commissioning capabilities
Cheap electricity attracts miners. Deliverable megawatts attract infrastructure capital.
From Hashprice Exposure to Contracted Infrastructure Income
Fundamentally, Bitcoin mining revenue depends on Bitcoin price, network difficulty, transaction fees, machine efficiency, and hashprice. Consequently, even a well-operated mining site remains exposed to market conditions that the operator cannot control.
A long-term AI infrastructure lease changes the revenue architecture.
Hut 8’s Phase 2 agreement is structured as a 15-year triple-net lease with a 3% annual base-rent escalator. The counterparty is described as a high-investment-grade company. Three additional five-year renewal options could increase the potential contract value of the full Beacon Point campus to $50.2 billion if all options are exercised.
This does not make the project risk-free.
The risk moves.
Instead of carrying direct exposure to Bitcoin economics, the developer must manage construction costs, financing, equipment lead times, tenant specifications, delivery deadlines, power availability and long-term facility performance.
The announced contract value is not immediate cash and should not be treated as profit. It represents expected value over the lease term and includes annual rent escalation. Construction expenses, financing costs, debt service and non-reimbursable operating expenses still matter.
The commercial lesson is clear: contracted infrastructure revenue can be easier to finance than volatile self-mining revenue when the tenant, lease structure and delivery plan are bankable.
Field Note: When comparing mining income with an AI lease, do not compare gross annual revenue alone. Compare revenue visibility, counterparty credit, required CAPEX, financing cost, residual asset value and the cost of missing the delivery date.
A Bitcoin Mining Site Is Not Automatically AI-Ready
A mining site may already have industrial land, large electrical feeds, transformers, switchgear, security systems and experience operating high-density compute.
That is a strong starting point. It is not a completed AI data center.
| Infrastructure Area | Typical Bitcoin Mining Requirement | AI Data Center Requirement |
|---|---|---|
| Power | High capacity and low cost | High capacity, power quality, redundancy and controlled distribution |
| Cooling | High-volume airflow, hydro cooling or immersion cooling | Precision cooling matched to rack density and server architecture |
| Network | Sufficient mining-pool connectivity | High-bandwidth, low-latency and often redundant fiber routes |
| Uptime | Revenue optimization based on mining economics | Contractual availability and service-level obligations |
| Rack Design | ASIC-specific shelving or manifolds | GPU/server racks with service access and structured cabling |
| Controls | Miner monitoring and basic environmental control | BMS, DCIM, leak detection, alarm management and capacity tracking |
| Fire Protection | Industrial equipment protection | Data-center-specific detection, suppression and code compliance |
| Commissioning | Functional equipment checks | Integrated systems testing under defined load and failure scenarios |
The conversion question should therefore be:
“How much of the existing power, land and delivery infrastructure can be reused without limiting the AI tenant’s uptime, rack density or expansion plan?”
Some mining assets can transfer directly. Others must be redesigned.
A container built around high-volume outdoor airflow, for example, should not simply be filled with GPU servers. AI equipment may require precision temperature control, defined humidity conditions, cleaner air, tighter filtration, different rack geometry and much stronger redundancy.
The enclosure is the visible part. The operating architecture creates the value.
Cooling Determines Whether the Conversion Works
Power may win the site. Cooling decides how much IT capacity can be monetized inside it.
Hut 8 stated that redesigning the first Beacon Point data hall around NVIDIA’s DSX reference architecture enabled 57% more IT capacity within the same land and utility footprint.
That is one of the most important figures in the announcement.
The commercial objective is not merely to secure more acreage or install more utility capacity. It is to convert a higher percentage of available infrastructure into dependable, billable IT load.
Higher rack density changes the cooling system. Depending on the server architecture and heat load, the project may require:
Precision air-cooled containment
Direct-to-chip liquid cooling
Coolant distribution units
Secondary cooling loops
Dry coolers or other heat-rejection equipment
Redundant pumps and fans
Water-quality control
Leak detection
BMS and DCIM integration
Factory and site acceptance testing
A Bitcoin mining container is usually designed around a defined ASIC quantity and electrical load. An AI module must be designed around rack-level heat density, coolant temperatures, flow requirements, redundancy targets and server maintenance access.
Do not select the CDU after selecting the container.
Start with server heat load and rack density. Calculate coolant flow, supply and return temperatures, pressure requirements and redundancy. Then size the CDU, piping and heat-rejection system as one connected loop.
Field Note: The maximum usable AI capacity of a site is the lowest capacity among its power, cooling, distribution, network and commissioned rack systems. A 100 MW electrical connection does not create 100 MW of revenue if the cooling plant supports only 70 MW of IT load.
Why Modular Infrastructure Matters
Beacon Point is a gigawatt-scale campus, but the delivery lesson also applies to smaller projects.
Large AI campuses are rarely energized as one completed block. They are developed through phases, data halls and repeatable infrastructure packages. The customer wants capacity delivered on schedule, not an impressive master plan that takes too long to become operational.
Modular AI data center infrastructure can shorten part of that delivery curve by moving enclosure fabrication, electrical assembly, cooling integration, controls installation and testing into a factory environment.
A properly engineered modular AIDC solution can include:
Prefabricated container or modular building structure
AI server racks and service aisles
Busway, PDU and branch distribution
Precision air-cooling or liquid-cooling architecture
CDU and external heat-rejection equipment
Fire detection and suppression interfaces
BMS and DCIM monitoring
Factory acceptance testing before shipment
Repeatable expansion modules
Modularity does not remove the need for substations, civil work, fiber, permits or site commissioning. It makes the compute infrastructure more repeatable and allows capacity to be delivered in controlled phases.
This is where experienced mining-container manufacturers can create value. They already understand high-power enclosures, thermal integration, electrical distribution, logistics and factory testing. The opportunity is to upgrade those capabilities for AI-level uptime, monitoring, redundancy and serviceability.
A mining container is a product.
An AI-ready facility is a system of systems.
The Mining-to-AI Conversion Scorecard
Before describing a mining site as AI-ready, evaluate it across seven categories.
1. Power Control
Confirm the contracted utility capacity, voltage level, interconnection status, substation scope and realistic energization date.
“Available power nearby” is not secured power.
2. Deliverable IT Capacity
Calculate the IT load that remains after cooling, electrical losses, pumps, fans, controls, support systems and operating reserves are considered.
Keep utility megawatts and IT megawatts separate.
3. Cooling Architecture
Define rack density, design-day ambient temperature, coolant supply temperature, redundancy level and heat-rejection method.
The cooling concept must work at peak conditions, not only at the average annual temperature.
4. Network Infrastructure
Verify carrier availability, fiber route diversity, latency and expansion capacity. A remote mining site with excellent electricity pricing may still be unsuitable for latency-sensitive AI workloads.
5. Reliability and Compliance
Specify the required power redundancy, cooling redundancy, fire-protection strategy, monitoring system and applicable electrical and building codes.
6. Delivery Capability
Review equipment lead times, factory production capacity, construction partners, commissioning procedures and contractual completion milestones.
A delayed megawatt produces no lease revenue.
7. Commercial Structure
Evaluate tenant credit, lease duration, rent escalation, renewal rights, financing conditions, construction obligations and residual use of the facility.
The best conversion project is not the one with the highest theoretical IT density. It is the one that reaches commercial operation with an acceptable return and retains useful infrastructure value if tenant requirements change.
What Beacon Point Means for Infrastructure Buyers
Hut 8’s announcement does not mean every Bitcoin mining site can become a billion-dollar AI campus.
Beacon Point combines utility-scale power, an investment-grade tenant, long-term financing potential, a large development pipeline and the ability to procure and deliver complex infrastructure. Most mining operators do not have all five.
Even so, the broader signal is difficult to ignore.
Bitcoin mining companies spent years learning how to locate power, negotiate with utilities, build industrial compute sites and operate high electrical loads. AI infrastructure buyers now need many of the same capabilities. However, they also require stricter standards for reliability, networking, cooling and delivery documentation.
As a result, the most valuable mining asset may no longer be the current ASIC fleet. It may be the right to use power at a site where additional digital infrastructure can actually be delivered.
For infrastructure buyers, the procurement question is therefore changing from:
“How many machines fit in the container?”
to:
“How many commissioned IT megawatts can this supplier deliver, cool, monitor and support by the required date?”
Field Note: A serious AI infrastructure RFQ should include IT capacity, rack power density, server type, cooling method, coolant temperatures, redundancy target, ambient design conditions, utility voltage, network requirements, fire-protection standard and required commercial-operation date.
Final Verdict for 2026
The Beacon Point deal shows how Bitcoin mining infrastructure can become the first layer of a much larger digital infrastructure business.
Hut 8 secured power and land, preserved flexibility across compute markets, redesigned the facility for higher AI density and converted the site into long-term contracted infrastructure income.
In this case, the winning asset was not a generic warehouse. It was a power-secured site combined with credible engineering and delivery capability.
For mining companies, the strategic question is no longer whether AI is more fashionable than Bitcoin. Rather, the real question is whether their power position, cooling system, network access and development team can satisfy an AI tenant’s technical and commercial requirements.
For equipment manufacturers, the opportunity is equally clear. The next generation of modular infrastructure must connect the enclosure, electrical distribution, precision cooling, CDU, heat rejection, monitoring, fire protection and factory testing into one deployable system.
Power creates the option.
Then, engineering converts that option into IT capacity.
Finally, delivery turns the IT capacity into revenue.
Frequently Asked Questions
Is Beacon Point a 1GW IT data center?
No. Instead, Beacon Point has 1,000 MW of secured utility capacity. Hut 8 has announced two 352 MW leases, bringing contracted IT capacity to 704 MW.
Can an existing Bitcoin mining site be converted into an AI data center?
Possibly. However, power and land are only the starting point. The site must also satisfy requirements for fiber, cooling, electrical redundancy, rack design, fire protection, monitoring, compliance and uptime.
Why are AI leases considered more stable than Bitcoin mining revenue?
Long-term leases can provide contracted income supported by tenant credit and rent escalation. By contrast, Bitcoin mining revenue remains exposed to Bitcoin price, network difficulty, hashprice and machine efficiency. Even so, AI projects still carry construction, financing and delivery risk.
What cooling upgrades are normally required?
Depending on the server architecture, projects may require precision air cooling or direct-to-chip liquid cooling, CDUs, secondary loops, heat-rejection equipment, redundant pumps and fans, leak detection, water-quality control and BMS/DCIM integration.
Why use modular AI data center infrastructure?
Factory-prefabricated modules can support phased expansion, repeatable engineering, controlled assembly and pre-shipment testing. Although they do not replace site work, they can reduce on-site integration and delivery risk.
