FMV vs dollar-out purchase options

The option is a single clause, and it determines the economics of everything before it. Read the two side by side and the rest of the lease becomes predictable.

The mechanism is simple and it explains every row. A lessor prices a lease to recover the asset's cost less the value it expects to have left at the end, plus a return. The larger the residual it is willing to assume, the less it needs to recover during the term, and the lower the payments.

A fair-market-value option means the lessor genuinely expects to have something worth selling. It has taken a view, priced it into the payments, and kept both sides of it — if values hold, that is its gain; if a new generation lands and values step down, that is its loss. The user's payments are lower because it sold that exposure.

A nominal option means the lessor expects nothing. It has recovered the full cost through the schedule, so the transfer at the end is an administrative act. Nothing about the residual has been transferred, because the user is buying the asset either way — it is simply paying for it over time.

Which is why the fifth row matters more than any other. Under FMV the user has a decision to make at the end and the decision has content: return, extend, or buy at the price the market sets. Under dollar-out there is no decision. An asset available for a token amount will always be taken. A user believing it retains flexibility because the document contains the word "option" has misread what it signed.

The fixed-price option sits between the two and needs the closest reading of the three. A price set years in advance is a view on where values land, and on hardware whose value moves in steps driven by other people's roadmaps the range of outcomes is wide. Whether it is favourable is unknowable at signing; what is knowable is that someone is taking a position, and that it is worth establishing which party.

The structures themselves are set out on GPU lease structures; the behaviour of the values underneath them is on GPU residual value and depreciation.

An FMV option wins wherever the user's need for the equipment might genuinely end, and wherever the user has no way to value what would be left.

The compute contract has a defined end and nothing committed after it. The clean case. The user pays for the use it needs, hands the equipment back, and carries nothing forward. Matching the option to a matched term is what makes the exposure end when the revenue does.

Obsolescence is the risk the user least wants. An FMV structure is the most complete transfer of technology risk available in lease form. For a sponsor whose competence is operating a facility rather than trading hardware, that transfer is the point.

The compute layer must stay separable from the rest of the stack. FMV is the structure that most cleanly keeps compute off a long-dated infrastructure balance sheet, which is why it recurs wherever the separation argument is being made.

Workload requirements are genuinely uncertain. The ability to return a configuration and re-specify at term end has real value where the mix of work is moving. That optionality is what the lower payments and the retained lessor upside are buying.

FMV is the wrong answer where the user is confident it will buy. Paying a lessor to hold a residual, then buying the asset at market value anyway, means the user paid for an option it always meant to exercise and then paid again to exercise it.

The two things to negotiate rather than accept: how fair market value is determined, and what the return actually requires. A valuation mechanic that leaves the number to the lessor's determination is not an FMV option in any useful sense — an independent appraisal standard, a defined process and a timetable are what make the clause real. And return conditions are where an attractive payment is recovered: de-installation, freight, configuration standards, excess wear, and a return window workable for equipment that is still in production use until the last day.

A dollar-out option wins where the user was always going to keep the equipment and would rather be paid for admitting it than pay for pretending otherwise.

Utilisation is durable past the term. An operator with a redeployment path, a pipeline and a realistic view of the product cycle is describing an asset with life left in it. Selling the residual to a lessor in that case is selling something the user is better placed to realise.

The user wants the upside. If values hold better than the market expects, the dollar-out user keeps the difference. That is a deliberate long position on the residual, and a user taking it knowingly is on solid ground.

No residual bid exists. Frequently the honest case. Where lessors will only write full-payout terms on a given configuration, the choice is not really being offered — and a nominal option at least makes the substance explicit rather than dressing financed ownership in operating-lease clothes.

Certainty is worth more than flexibility. A known end state with no valuation negotiation and no return obligation has administrative value for a user with no intention of handing anything back.

It is the wrong answer wherever the demand behind the equipment is shorter than the schedule. The obligation continues when the customer leaves, and the user is then holding both a payment stream and an asset it has no use for — the same exposure that makes term matching the first test in lease vs own compute. It is also the wrong answer where the sponsor chose a lease to keep the asset off its balance sheet: a nominal option is one of the clearest indicators pointing the other way, and the classification consequences are set out in operating vs finance lease.

One further case belongs here. A sponsor that has already taken a dollar-out structure and later wants the capital back is not choosing between these two options any more; it is choosing between the routes on sale-leaseback vs refinancing.

DigitalOcean's September 2026 equipment facility provides a useful public example of the dollar-out end of the spectrum. Advances can fund up to 90% of eligible equipment cost, the lessee supplies the remaining amount as prepaid rent, each advance amortises fully by the facility maturity, and title transfers for nominal consideration after payment in full. (DigitalOcean Form 8-K, as of September 20, 2026)

Every element points in the same direction. The user contributes equity-like cash at the start, repays the financed amount rather than only the equipment's use during part of its life, and receives the asset at the end for a token payment. The lessor holds title and security during the term but does not rely on selling the equipment at a material residual to earn its return. Technology upside and downside stay with the user.

That does not make the structure inferior to FMV. It makes it suitable for a different case. An operator expecting durable utilisation, able to redeploy equipment and willing to carry obsolescence may prefer full payout because it retains every unit of value after the financing is repaid. It also avoids an appraisal and return process at maturity. The economic question is whether those benefits exceed the value of a genuine walk-away right.

The disclosed mechanics also show how to compare proposals that use different labels. Put initial cash, all scheduled payments, fees, end-of-term amount, title outcome and remaining asset value on one timeline. Under a nominal-option structure, the user's downside case includes payments remaining after a customer leaves and the cost of owning obsolete or underutilised equipment. Under FMV, the downside case includes return cost, condition claims, holdover exposure and the possibility that the user wants to keep the equipment but must buy it at a stronger-than-expected market value.

There is no universally cheaper column. Dollar-out usually shows more payment because it buys the whole asset. FMV may show less payment because the lessor is investing in the unamortised residual. The correct comparison gives each party credit for the asset value it keeps and charges it for the risks required to realise that value.

An FMV option is only as real as the procedure for discovering fair market value. A clause that says the lessor will quote a price at expiry gives the user a future negotiation, not a reliable option.

The standard must specify what is being valued. Value in continued use at the installed site can include power, interconnect and configuration the lessor does not own. Value in exchange assumes a sale between market participants. Orderly liquidation value and forced-sale value answer still different questions. The lease should not price the purchase option on installed earning capacity while pricing a return default on removed equipment.

It must also specify condition and location. Is the equipment valued as maintained and operating, as-is at the site, or after compliant return to a designated destination? Who pays removal and freight? Are missing components, modifications or excess use reflected in the appraised price or pursued as separate return claims? Double deductions can arise when a lower condition value and a separate damage payment address the same defect.

The timetable needs room for a decision. The user should receive an indicative value early enough to compare purchase, extension and return, with a defined process if the parties disagree. Independent appraisers, appointment rules, access to data and a method for resolving divergent values should be settled before either party knows which answer favours it. A process that ends after the return deadline makes purchase the only operational choice.

Finally, the option should address market discontinuity. A thin secondary market may produce few comparable sales, particularly for integrated rack-scale systems. The appraisal method should permit evidence from broker quotes, complete-system transactions, component values and income from a realistic secondary workload without treating an aspirational listing price as a completed sale. It should also define whether taxes, duties and transaction costs are included.

These details are not legal ornament. Under FMV, the appraised number decides whether the user keeps the equipment and how much residual the lessor realises. A fair process is part of the financing economics in the same way the interest calculation is part of a loan.

The best time to discover that an option does not fit the operating plan is before the lease starts. The second-best time is early enough before expiry to change the plan without paying holdover rent.

At signing, model four end states: return after the original customer contract, extend for a defined secondary use, purchase and continue operating in place, and purchase for resale or relocation. Assign each a realistic downtime, site cost, return or purchase payment, refresh requirement and expected revenue. The decision should still work if residual value is materially above or below the central case. If only one narrow value makes the preferred option economical, the structure is a leveraged residual forecast.

During the term, update that analysis when a new product generation ships, the customer contract changes, power pricing moves or a credible secondary-market transaction appears. The purpose is not to trade the option continuously. It is to preserve enough lead time to contract for replacement equipment, market returned capacity, arrange a buyout or reserve cash for the return.

Six to twelve months before expiry, the process becomes operational. Confirm notice deadlines, appraisal dates, inspection access, data-sanitisation standards, de-installation contractors, transport capacity, replacement-system delivery and whether the hosting agreement permits equipment to remain during the transition. Obtain extension and early-purchase quotes even if they are not the preferred route; they are the alternatives against which return should be measured.

The user should also test whether buying the equipment creates value it can actually realise. A positive difference between appraised value and purchase price is not useful if the user cannot sell because of transfer restrictions, cannot remove because another creditor controls the site, or needs the equipment to keep serving a contract. Likewise, returning equipment is not genuine flexibility if no replacement can arrive before the customer needs capacity.

Option value comes from having executable alternatives at the decision date. The contract creates the rights; advance planning keeps them from collapsing into the one path time still permits.

The end-of-term clause is the most informative sentence in an equipment financing, and it can be read before any of the numbers are agreed.

What a lessor offers tells you what it expects. An FMV option on a given configuration means a party whose business is remarketing that hardware is prepared to take a position on its future value. A refusal to write anything but full-payout terms on the same hardware is a view on values, expressed in the only way a lessor expresses one. Soliciting the option structure across several configurations is a cheap way of reading the market's expectations without asking anyone to forecast.

What a sponsor asks for tells the same story in reverse. A sponsor pushing hard for a nominal option is saying it expects to keep the equipment and believes it is worth more than the lessor does. That may well be right — an operator with a redeployment path often does know better. It should be a stated position rather than an incidental consequence of preferring a lower headline structure.

Beware the option chosen for the wrong reason. The two common errors are symmetrical. A sponsor takes FMV for lower payments while privately intending to buy, and pays twice. Or a sponsor takes dollar-out because the total looked lower over the term, without noticing it has bought a residual position it cannot value.

Test the option against the compute contract, not against the equipment. The right question is never "will this hardware still be good" but "will there be demand for it that this sponsor can serve." Those come apart constantly, and the second is the one the obligation is measured against — which is why the quality of the contract behind it, tested on compute offtake as credit, governs the option choice as much as the hardware does.

The discipline is to write down the assumed end-of-term value before reading the term sheet. FMV is a decision not to take a position on that number. Dollar-out is a decision to take one. Both are defensible; taking one without noticing is not.