Financing a mining-site conversion

The most common structural error in these transactions is treating the conversion as an acquisition with some capital expenditure attached. It is two requirements with different risk profiles, different tenors and, usually, different providers.

The acquisition is a real-asset purchase. What is being bought is an electrical position — an interconnection, a queue standing, a utility arrangement, a substation and medium-voltage distribution, and the land under them — with a facility on top that will largely be removed. It is secured on things that exist, it can be diligenced in the conventional way, and it is the half of the transaction that most closely resembles ordinary practice.

The retrofit is a construction requirement. The existing building comes down or is stripped, a new shell goes up, and the electrical and mechanical systems are rebuilt to a standard the original site was never designed to. It is drawn over time against progress, it carries completion risk, and it is underwritten on a plan rather than on an asset.

Between them sits a third thing that is easy to miss and expensive to discover: the site stops earning before the new use starts. The current operation winds down, the connection still has to be held, obligations still have to be met, and the estate produces nothing while consuming capital. That is a short-dated requirement with its own logic, and it is the same shape as a [bridge to energization](bridge-to-energization) even though the power in this case already exists.

Why the separation matters commercially: the acquisition frequently closes on the strength of the position, and the retrofit is then approached as a subsequent problem. A conversion financed that way has bought an asset it may not be able to complete — which is the least attractive position in the sequence, because the position's value is highest when it is intact and unfinished sites do not stay intact.

This is the constraint particular to this fact pattern, and it is a credit observation rather than a view on anything else.

An operator that built and ran a mining estate generally has a balance sheet whose value and whose cash generation move together with a single volatile input. Revenue is denominated in it; the hardware on the balance sheet is worth what it is worth largely because of it; and the equity value of the enterprise moves with it too. That is correlation, not weakness — but it produces a specific and awkward dynamic.

The conditions that make conversion attractive are frequently the conditions in which the operator's own resources are thinnest. When the economics of the existing use compress, the case for redeploying the electrical position to a different load strengthens, and the operator's capacity to fund that redeployment weakens at the same time and for the same reason. The trigger for the transaction is also the trigger for the credit deterioration. Very little else in infrastructure finance behaves that way; the usual case is a sponsor whose credit is uncorrelated with the reason it is transacting.

Three consequences follow for anyone structuring around such a sponsor:

• Recourse to the sponsor is worth less than it appears. A guarantee, a completion undertaking or a cost-overrun obligation from an entity whose resources move with an external price is a contingent claim that is weakest in the scenario where it is called. It is not worthless; it is correlated, and it should be sized as correlated.
• Existing obligations sit across the estate. Equipment finance, facility-level debt and convertible instruments are common in this segment, and they frequently attach to assets a conversion needs unencumbered. What has to be released, by whom, and on what terms is a gating question rather than a closing detail.
• Timing is set by pressure rather than by readiness. A seller moving on its own balance-sheet timetable is a seller whose transaction may need to close before the diligence in [converting a mining site](/sites/mining-site-conversion) has finished. The structural answer is escrow, holdback and staged consideration rather than reliance on undertakings.

None of this makes the transaction unfinanceable. It relocates where the credit has to come from — which is the subject of the next section.

Where the sponsor's covenant is correlated and the asset is mid-transformation, the contract becomes the most bankable object in the transaction. That is not unusual in project finance generally; what is unusual here is how much weight it has to carry.

A firm contract with a creditworthy counterparty — a hosting agreement, a capacity commitment, or a compute contract behind one — does several things at once for a structure of this kind:

• It supplies a payment obligation that is uncorrelated with the sponsor. The cash flow servicing the structure comes from a party whose credit has nothing to do with the input the sponsor's balance sheet moves with. That single fact does more for the structure than any amount of sponsor support.
• It converts the retrofit from speculative to contracted. A building being built to a specification a named counterparty has agreed to pay for is a fundamentally different underwriting from a building being built in the hope of leasing it.
• It sets the specification. The counterparty's availability requirement determines the tier, and the tier determines the sellable capacity and therefore the whole revenue case. A structure sized before that requirement is known is sized against a guess.
• It can survive the sponsor. Where the contract is with the project entity and is assignable to a financing party on enforcement, the structure has an answer to sponsor failure that does not depend on the sponsor. Where it is not, the contract is an asset that leaves at exactly the wrong moment.

The mechanics of what makes such a contract carry weight — term, take-or-pay construction, termination rights, assignability, and what the counterparty is actually obliged to pay for — are worked through in [the hosting agreement as a financeable contract](hosting-agreement-as-a-financeable-contract) and, on the layer above, in [what lenders underwrite on a GPU cluster](/compute/what-lenders-underwrite-on-a-gpu-cluster). The point specific to a conversion is one of relative weight: on a conventional development with a diversified sponsor, the contract is one input among several. Here it is frequently the input, and a conversion presented without one is asking capital to take sponsor risk, completion risk and demand risk simultaneously, from a sponsor whose credit is correlated to the reason the site is for sale.

A converting mining site is an unusually heterogeneous collateral pool, and each component behaves differently. The security package has to be assembled component by component, and the ordinary assumption that a real-asset mortgage picks up most of the value does not hold — because the most valuable single item is generally the one that cannot be mortgaged at all.

The practical answer for the interconnection is indirect and is the same one used everywhere in this stack: security over the entity that holds the position rather than over the position itself, tested against whatever the documents say about change of control. That is worked through in [who holds the interconnection position](who-holds-the-interconnection-position), and on a conversion it acquires an extra dimension, because the load-change question runs in parallel — a position that requires the utility's agreement to serve a firm load is a position whose value is conditional on a consent nobody has yet given.

The gap between those two is where these transactions succeed or fail, and it can be set out in a sentence each way.

What is being sold is an operating facility at a stated nameplate capacity, with an energized connection, on land, with equipment in it — presented as a going concern that needs redirecting.

What is being underwritten is narrower and different: the megawatts of critical load the site can deliver at the availability standard a named counterparty will actually contract to; a completion path for the work required to get there, with the cost and the schedule allocated to parties obliged to perform; a payment obligation from a counterparty whose credit is uncorrelated with the seller's; and an enforceable position over the entity holding the electrical position, tested against the consents that position requires.

Everything in the first sentence that does not appear in the second is a cost line rather than an asset. That is not a criticism of the seller's presentation — an operating facility genuinely is what they hold. It is a statement about what capital can actually attach to, which is the only thing that determines whether a structure exists.

The sequencing that follows from it is unglamorous and consistent. Establish the utility's position on the change of load before the price is agreed, because the whole asset is conditional on it. Establish the sellable capacity at a stated tier, because the revenue case is built on that number and not the nameplate. Establish the offtake, because it is the credit the structure will rest on. Establish what is already encumbered and what has to be released, because those consents are slow. Then size the acquisition, the retrofit and the gap between the two uses as three requirements rather than one — and do it in that order, because a conversion that closes its acquisition before it knows the answers to the first four has bought a position it may not be able to finish.

Continuum advises on how conversions of this kind are structured and arranges the capital that finances them. It is not a utility, not a bank, not a broker-dealer and not a direct lender; it does not hold client funds.