A 15-Year Lease Is Only as Good as the Tenant

The investment question has moved beyond the lease announcement

A powered-shell landlord exchanges upfront capital for a stream of contractual payments. It avoids direct ownership of rapidly depreciating GPUs, but accepts construction, tenant-credit and residual-property risk. The commercial announcement establishes demand. It does not, by itself, establish the cash ultimately available to common shareholders.

Three questions therefore belong beside every headline contract value. Which entity owes the rent? What happens before and after a default? How much further capital must be supplied before the contract produces distributable cash? These questions connect the operating asset to its financing and, ultimately, to equity value.

The distinction is particularly important during a development cycle. A company can have high-quality customers and still face an equity shortfall if construction costs rise, delivery slips or borrowing proceeds are less flexible than the headline cash balance suggests. Conversely, a less highly rated tenant can be acceptable if the price, contractual protection and building’s alternative uses provide sufficient compensation.

Capacity terminology matters financially. Gross power includes facility overhead; critical IT load describes power available to computing equipment. Neither should automatically be equated with capacity delivered, accepted and paying rent. Dividing enterprise value by a mixture of these measures can create apparent discounts that disappear once the denominator is made consistent.

For example, Cipher’s Q3 filing describes Black Pearl and Barber Lake as 300 MW facilities and Stingray as 100 MW. Those site figures should not be silently relabeled critical IT load. A project comparison needs a reconciled capacity definition, ownership share and delivery date before it can support a valuation conclusion. ^[9]

A guarantee is a set of rights, not a label

TeraWulf’s November 4 filing illustrates why legal structure matters. Google’s support for the Anthropic leases at Justified becomes effective only upon commencement of the final hall under each corresponding lease. It can terminate when specified conditions are met, including sustained qualifying investment-grade ratings. Following a qualifying default, Google can assume the lease and pay rent, or in certain circumstances pay an amortizing termination fee and acquire an equity interest in the project company. ^[6]

This is meaningful credit support. It is not a promise that common shareholders retain the same asset, rental stream and economics in every outcome. The period before activation also requires separate funding and risk assessment.

Nominal support divided by undiscounted lease value is at most a description of two disclosed amounts. It is not an expected recovery rate. The supported period, rent escalation, permitted claims, amortization and beneficiary rights all affect the result. A guarantee from a major technology company on one project should not be imputed to other facilities simply because they serve the same AI ecosystem.

Financing costs reveal a difference, not its entire cause

Hut 8’s Beacon Point notes were priced at 6.129% in June, against a high-investment-grade tenant. Galaxy’s Helios financing completed in July at a 9.875% coupon. The difference is economically material, but the securities also differ in maturity, amortization, structure and issuance date. It cannot be attributed entirely to tenant credit. ^{[7], [8]}

A controlled illustration isolates the financing effect more cleanly. Assume $10m of development capital per MW, funded with $8m of debt and $2m of equity. Annual rent is $1.8m; debt amortizes evenly over 15 years. Holding everything else constant, the higher coupon leaves about $300,000 less annual cash per MW before operating costs and other charges.

These are Dilikoglu Research assumptions. The $10m capital intensity is an illustrative input, not reported expenditure or a sector midpoint. The amortization schedule is deliberately simplified and does not reproduce either financing above.

The illustration also exposes liquidity risk. At the higher coupon, six months of debt service is approximately $0.662m per MW. Such a reserve buys time; it does not finance an indefinite interruption. Its sufficiency depends on the actual debt schedule, cash sweep provisions, operating expenses and timing of replacement income.

This is why similar headline rent does not imply similar equity returns. Contract value per MW per year is itself only an average: escalation, power reimbursement, tenant improvements and service obligations can make two apparently comparable leases economically different. A landlord may rationally accept a lower quoted rent if the contract enables more resilient financing and lowers its required equity contribution.

In bankruptcy, the claim and the building are separate assets

For damages from termination of a real-property lease, Section 502(b)(6) limits the allowed claim using the greater of one year’s rent or 15% of the remaining term, capped at three years, measured from the earlier statutory petition or repossession/surrender date; unpaid rent is addressed separately. Under a simplified time-based calculation with flat rent and 12 years remaining, 15% is 1.8 years. This is an illustration, not a universal outcome for every data-center agreement. Classification, interpretation and mitigation matter. An allowed claim is not cash recovery. ^[10]

Section 365 generally requires performance of relevant post-petition lease obligations pending assumption or rejection, subject to its exceptions. For covered nonresidential real-property leases, the decision deadline is generally 120 days or earlier plan confirmation; a court may grant an initial 90-day extension for cause, with later extensions requiring the lessor’s written consent. These rules do not establish a fixed date for cash recovery or a replacement tenant. ^[11]

Separate guarantor obligations, collateral rights and project-level financing documents require their own analysis. A recovery estimate for a tenant’s unsecured bonds cannot simply be applied to a landlord’s claim, and a landlord’s recovery does not flow directly to its shareholders. This note provides an investment framework, not legal advice.

Residual value depends on how quickly rent can restart

The building remains potentially valuable when a tenant fails. Power access, location, cooling, rack density, connectivity and the cost of adapting the facility determine the pool of replacement tenants. A modern, well-located site can recover substantial earning power. That does not eliminate lost rent or the senior claims that accumulate during the interruption.

To illustrate, normalize the old annual rent to 1 and assume 12 years remain. At an 8% discount rate, uninterrupted year-end receipts have a present value of 7.536. The cases below vary vacancy, replacement rent and reletting expenditure. They do not assign a probability of tenant failure.

Even a six-month interruption with no rent reduction consumes value. With a two-year vacancy, a 30% lower rent and costs equal to 1.5 years of old rent, the retained value falls to roughly one-third. This is an asset-level result. Leveraged equity can experience a much larger percentage loss because debt and other senior obligations must still be satisfied.

Any recovery on a termination claim should be analyzed separately, including timing and mitigation. Adding a haircut-adjusted claim to replacement rent without checking the underlying damages calculation risks counting the same economic loss twice.

A large cash balance does not finish the building

Cipher reported $656.0m of cash and equivalents at September 30, alongside $2,196.6m of restricted cash. The distinction matters when assessing liquidity: restricted proceeds may finance construction or secure obligations rather than fund unrelated corporate needs. ^[9]

TeraWulf’s Abernathy transaction provides another timing example. Of the approximately $530m of consideration, $250m was received in Q3; subsequent installments were due later, subject to adjustments. Treating the entire amount as cash already received would overstate immediately available liquidity. ^[5]

A simple example shows the danger of a headline EV-per-MW comparison. Assume a completed project is worth $10bn, existing debt is $6bn, restricted construction cash is $3bn and remaining construction costs are $4bn. Ignoring timing and all other claims, residual equity is $3bn: $10bn less $4bn, plus $3bn, less $6bn. Omitting the remaining expenditure produces $7bn and overstates the residual by $4bn.

The same logic applies to contracted megawatts. Giving an unfinished development a completed-asset value while subtracting its construction cash rewards it twice. A defensible comparison uses both the remaining uses of funds and the capacity that those funds are expected to deliver. Project ownership, preferred capital and other operating businesses must also be reconciled before comparing parent-company valuations.

What would change the investment assessment?

The next evidence should come from conversion of commitments into operating cash, rather than another increase in pipeline size. Successful commissioning, tenant acceptance and recurring rent collection reduce different uncertainties. They should be tracked separately.

The strongest version of the landlord thesis combines a paying tenant, enforceable support where relevant, an adequately funded delivery plan and a facility that retains alternative uses. The weakest version relies on a long contract to stand in for all four.

For investors, the consequence is practical: evaluate rent, credit support and capital needs in one cash-flow model, then determine what remains for equity. A long lease can create durable value. Its length alone does not measure that value.

Scope and limitations

This sector note does not change a company rating, price target or published model. Reported facts and management targets are identified separately from Dilikoglu Research illustrations. The financing observations are historical issuance terms, not live yields; they do not isolate a causal tenant-credit premium. Public disclosures do not provide every project’s complete economics or all negotiated contractual provisions. Capacity definitions, accounting presentation and ownership structures differ across companies.

The illustrations exclude several costs and use simplified timing to make individual risks visible. They are not forecasts of default, recovery or shareholder returns. Legal outcomes depend on the agreement, jurisdiction and facts. Future developments may differ materially from management expectations and the scenarios presented here.