CleanSpark AI Data Center Agreement: $6.6B Lease Signals a Major Miner Pivot

CleanSpark AI data center agreement news is one of the clearest signs yet that large bitcoin miners are trying to turn power access into long-term AI infrastructure revenue. CleanSpark has signed a 20-year, triple-net lease worth about 6.6 billion USD at its Sandersville, Georgia campus, with renewal options that could raise total contract value to about 11.6 billion USD over 30 years.

This is not just a large real estate transaction. It changes how investors, miners, AI infrastructure teams, and energy planners may evaluate CleanSpark. A company once viewed mainly through the bitcoin mining cycle is now positioning itself as a landlord for AI and high performance computing, or HPC.

What CleanSpark Secured in the Sandersville Deal

CleanSpark signed the lease with an unnamed, high investment-grade global technology company. The site is the company's Sandersville campus in Georgia, being developed for AI and HPC workloads.

The core figures are straightforward:

• Base term: 20 years
• Base contracted revenue: approximately 6.6 billion USD
• Renewal options: two 5-year extensions
• Potential total value: about 11.6 billion USD over 30 years
• Committed capacity: 175 megawatts of critical IT load
• Primary workload type: AI and high performance computing

Some coverage has referred to a 250 MW facility. That appears to describe the broader site scale, while the lease itself centers on 175 MW of committed critical IT load. This distinction matters. In data center planning, critical IT load refers to the power available for servers, GPUs, networking gear, and other compute equipment. It is not the same as total campus utility capacity.

That is a common place where non-technical readers get tripped up. A 250 MW utility connection does not mean 250 MW of GPU capacity. Cooling systems, power conversion, redundancy design, and building services all consume part of the available power envelope.

Why the Triple-Net Structure Matters

The CleanSpark deal is structured as a triple-net lease. Under this model, the tenant typically pays most property-level operating costs, including maintenance, insurance, and property taxes.

For CleanSpark, that structure could produce very high net operating income margins. Market commentary has cited management expectations of NOI margins close to 100 percent because many costs are passed through to the tenant.

On simple math, 6.6 billion USD over 20 years implies about 330 million USD in average annual net operating income. That is meaningful for a company whose historical performance has been tied closely to bitcoin price, network difficulty, block rewards, energy prices, and machine efficiency.

The deal also changes the quality of revenue. Bitcoin mining income can move sharply month to month. A long-term lease with an investment-grade tenant looks more like digital infrastructure income. That is why several analysts have framed CleanSpark as a potential digital landlord rather than only a miner.

Capex, Financing, and the Hard Part Nobody Should Ignore

The attractive economics come with a serious build requirement. CleanSpark's disclosures and related analysis estimate landlord project costs of about 10 million to 12 million USD per MW of critical IT load. For 175 MW, that points to a total build cost of roughly 1.75 billion to 2.10 billion USD.

That is not a small retrofit. AI campuses require dense power distribution, cooling that can handle GPU clusters, backup systems, fiber connectivity, security, and staged commissioning. In the field, the painful surprises are often mundane: transformer lead times, substation interconnection work, switchgear delays, or liquid cooling design changes after the tenant finalizes hardware density.

CleanSpark has indicated it expects to finance much of the build-out through project debt rather than heavy equity dilution. Reporting has also suggested the company may use its bitcoin treasury, reported at roughly 13,941 BTC, as collateral support.

That strategy has logic, but it is not risk-free. Bitcoin collateral can strengthen financing access when markets are liquid. It can also become a pressure point if crypto prices fall while construction costs rise. To be blunt, project debt only works well when delivery schedules and tenant rent commencement stay on track.

Timeline: When the Campus Is Expected to Come Online

CleanSpark disclosed the agreement in a Form 8-K filed on July 14, after signing the lease in early July. The delivery schedule is phased:

1. Initial data hall: expected to be ready for service in Q4 2027
2. Remaining data halls: expected to ramp through early 2028
3. Management target: completion of remaining halls by Q1 2028

Reporting has noted an important contract risk: missed construction milestones could reduce rent or allow the tenant to terminate. That clause is not unusual in large data center leases, but it puts real weight on execution. You cannot announce your way into 330 million USD of annual NOI. You have to deliver powered, cooled, commissioned space.

Texas Exclusivity Could Be the Bigger Long-Term Story

The Sandersville lease is already large. Still, the Texas component may define CleanSpark's next chapter.

The tenant has signed a letter of intent and exclusivity arrangement covering CleanSpark's entire Texas portfolio. Reports describe that portfolio as 718 acres with up to 885 MW of secured and planned power capacity, including locations such as Sealy and Brazoria.

If CleanSpark can commercialize those Texas assets under similar lease structures, some analysis suggests annual AI infrastructure revenue could exceed 1 billion USD. That is the kind of shift that would make bitcoin mining only one part of the company rather than the whole story.

Why Bitcoin Miners Are Moving Toward AI Infrastructure

This deal fits a wider pattern. Large miners already hold several assets that AI infrastructure buyers need:

• Access to large power footprints
• Experience building data center campuses quickly
• Land in energy-rich regions
• Grid interconnection knowledge
• Operational experience with high-density compute environments

CleanSpark CEO Matt Schultz has argued in a CNBC segment that miners can build AI data centers faster than traditional methods because they already control energy sites and know how to deploy compute at scale. He also highlighted that CleanSpark operates about 1.03 gigawatts of active facilities, with another 1.7 gigawatts in development, and said the company won an AI data center contract in Wyoming over Microsoft.

The logic is sound, with one caveat. Bitcoin mining facilities and AI data centers are not interchangeable. ASIC mining sites can tolerate operating patterns and redundancy levels that enterprise AI tenants may reject. AI and HPC tenants care about uptime, cooling stability, network architecture, service-level commitments, and hardware security. A miner that wants to serve AI customers must move up the infrastructure quality curve.

What the 175 MW AI/HPC Campus Could Support

The tenant has not publicly disclosed its specific workloads. Still, a 175 MW critical IT load is large enough for serious AI and HPC use cases.

Large-Scale AI Training and Inference

The site could support GPU-rich clusters for training foundation models, fine-tuning domain models, or running high-throughput inference. Modern AI deployments can require extreme rack densities, especially when using accelerators such as NVIDIA H100 or newer Blackwell-generation systems. The cooling and power design choices made before 2027 will matter a lot.

Enterprise and Scientific HPC

HPC workloads could include climate modeling, financial risk simulations, materials research, computational biology, and engineering simulations. These workloads need predictable power, low-latency networking, and disciplined cluster operations.

Cloud or Platform Capacity

Because the tenant is described as a global technology company, the capacity may feed cloud, SaaS, or platform services. That would explain the long lease term. AI capacity planning is now strategic infrastructure planning, not a short procurement cycle.

Market Reaction and Valuation Implications

Reporting indicated that CleanSpark shares rose more than 20 percent in pre-market trading after the announcement. The reaction is easy to understand. Investors saw a path from cyclical mining economics to contracted infrastructure cash flow.

There has also been some confusion in secondary reporting, with mistaken references to 66 billion USD or 116 billion USD. The more reliable figures are the ones aligned with CleanSpark's disclosures and major reporting: 6.6 billion USD base value and 11.6 billion USD including extensions.

The valuation debate is now simple: should CleanSpark trade like a crypto miner, an AI infrastructure developer, or a hybrid of both? The answer depends on execution. If Sandersville is delivered on schedule and Texas follows, the infrastructure case strengthens. If financing slips or milestones are missed, the market may reprice the risk quickly.

Risks to Watch

Professionals evaluating this agreement should track four risk areas:

• Construction risk: The project requires billions in build costs and phased delivery beginning in Q4 2027.
• Financing risk: Project debt terms remain undisclosed, and bitcoin-backed collateral adds market sensitivity.
• Tenant concentration: A single unnamed tenant anchors a large part of the AI infrastructure strategy.
• Energy and regulatory scrutiny: AI campuses face questions about grid impact, energy sourcing, water use, and local permitting.

For readers building expertise at the intersection of blockchain, AI, and infrastructure finance, this is a useful case study. Blockchain Council's Certified Blockchain Expert™, Certified Cryptocurrency Expert™, and Certified Artificial Intelligence (AI) Expert™ programs help you understand how mining economics, AI compute demand, and digital infrastructure models connect.

What This Means for the Future of Mining and AI Compute

CleanSpark's deal suggests that the next generation of large miners may not be judged only by hash rate. Power strategy, site control, project finance, tenant quality, and data center delivery skill will matter just as much.

The strongest version of this strategy is compelling: use mining-era energy infrastructure to serve AI customers under long-term contracts. The weak version is risky: assume that a mining campus can become an AI campus without major engineering, capital, and operational upgrades.

Your next step is to watch the financing terms, construction milestones, and Texas exclusivity updates. If you work in blockchain or AI, study this deal as a live example of how crypto infrastructure is being repurposed for AI compute. Then build the skill stack around it: mining economics, power markets, data center design, and AI workload requirements.