AI Data Center Leases Stretch to 20 Years as Power Becomes the Asset

Host Digital outlined a 15-year take-or-pay lease for its Oklahoma Site I, representing $1.25 billion of contracted base-term rent, approximately $67 million of first-year rent, and 55 MW of gross capacity targeted for delivery in Q1 2027. The company has invested about $40 million in the project and raised an additional $17.5 million, while pursuing project-level financing and a potential second 20 MW Oklahoma site with roughly $391 million of base-term rent. The article highlights a broader AI-data-center trend of long-dated, investment-grade-backed leases and debt financing against power-connected sites, but Host's plans remain subject to construction execution, financing, related-party acquisition negotiations, and potential dilution.

Analysis

The investable signal is not aggregate AI-capex forecasts; it is the widening valuation gap between operators with contracted, financeable power capacity and miners whose capacity remains exposed to Bitcoin economics. Long-duration leases can convert volatile power assets into infrastructure-like cash flows, but only after completion, utility milestones, and tenant credit support are independently verified. The key rerating catalyst over the next 1-3 months is disclosure of construction capex per MW, debt-service coverage, tenant identity/guarantee structure, and the true economics of pass-through power pricing—not another headline contract value.

HUT appears best positioned on cost of capital if its revolver translates into faster site-control and collateral deployment; lower-cost liquidity creates meaningful option value when grid-connected sites are scarce. CLSK's secured financing at a materially higher fixed coupon signals a less forgiving execution path: construction delays or cost overruns would erode equity value disproportionately because debt service begins ahead of stabilized rent. CORZ and WULF have the most direct power-portfolio optionality, but their equity outcomes remain highly sensitive to whether incremental capacity earns infrastructure multiples or continues to be valued as crypto-mining capacity.

Second-order beneficiaries are electrical equipment and grid-build vendors—VRT, ETN, GEV and PWR—whose revenue is recognized during construction rather than contingent on eventual tenant utilization. The contrarian risk is that investors are treating take-or-pay language as equivalent to investment-grade project finance; a lease backed by a private AI customer, a limited guarantee, or uncapped power costs can be materially weaker than a conventional utility-style contract. The promoted HOST disclosure is not tradeable evidence until SEC filings reconcile lease counterparties, project-level capital requirements, related-party terms, and dilution; treat it as a liquidity/event risk rather than a data-center REIT analogue.

Key Decisions for Investors

• Prefer a 6-12 month long CORZ / short CLSK pair, sized beta-neutral: CORZ offers scalable power optionality and potential restricted-cash release, while CLSK carries higher fixed financing burden. Target 15-25% relative return; exit if CORZ's next capacity delivery slips, or if CLSK demonstrates contracted rent commencement ahead of schedule and delevers faster than expected.
• Accumulate HUT only on verified progress toward funded development milestones and disclosed tenant-credit terms; use a 12-month horizon. The favorable liquidity profile is valuable, but do not underwrite announced contract value until construction capex and collateral needs are quantified. Falsifier: borrowing costs widen materially from the initial spread or incremental equity issuance becomes necessary.
• Use VRT and ETN as lower-volatility expressions of the buildout over 6-18 months, preferably on broad AI-infrastructure pullbacks. They capture power-distribution and cooling spend before data-center operators prove stabilized cash flow; key risk is hyperscaler capex guidance cuts or lead-time normalization compressing equipment multiples.
• Avoid HOST pending independent filing review. Create an alert for definitive project financing, tenant identity/credit support, all-in capex per critical MW, and post-financing share count; absent those data, the asymmetric risk is dilution and delivery slippage rather than a valuation rerating.
• Monitor high-yield/project-finance spreads and utility interconnection approvals over the next quarter. A 100-150bp widening in financing costs or delayed utility approvals would pressure highly levered conversion names first—particularly CLSK and WULF—and is the appropriate trigger to reduce sector gross exposure.