What makes a data-center site financeable: five gates

A site can be genuinely excellent — well located, real power, motivated seller — and still be unfinanceable. The two words measure different things.

"Good" is a judgement about the asset. "Financeable" is a judgement about whether a third party with its own investors and its own credit committee can put money behind it on terms it can defend. Those come apart constantly, and the gap is where most transactions actually die.

The gates below are independent and conjunctive. Clearing four of five is not eighty per cent of the way there; it is a no. This is the single most common misunderstanding among sellers, who tend to over-invest in the one gate they find most interesting and treat the rest as paperwork.

A documented power position attached to land the seller does not firmly control is two problems presented as one asset.

What capital looks for is duration and certainty that outlast the development timeline: ownership, or an option or lease long enough to carry the project through energization with margin. A short option over a site whose power will not arrive for three years is a countdown, not control.

The common failure is subtler than "no control." It is control that expires before the power lands, control over a parcel that differs from the one the interconnection was studied against, or control contingent on a rezoning that has not happened. Each of those reads as control right up until diligence.

Match the two on a page: what is controlled, on what terms, until when — against what the power position requires and when it lands.

This is the one sellers most often have not worked on at all.

Lenders and infrastructure capital look through to project cash flows. A site with power and no contracted demand produces no cash flow, so there is nothing to underwrite. Power without offtake is an input, not a project — and an input is worth substantially less than a project.

The distinction that matters is between *binding and contracted* on one side, and *indicated, forecast, or in discussion* on the other. A letter of interest from a credible hyperscaler feels like enormous progress. It is not offtake, and no amount of counterparty prestige converts it into offtake.

Where a site genuinely cannot have offtake yet — early-stage, pre-permit — that is not disqualifying. It does change what kind of capital is available and on what terms, and it should be stated rather than papered over.

Power, control and offtake are usually worked on because they are visible. These are the ones that surface in diligence, when the cost of discovery is highest:

• Environmental position. Contamination history, remediation obligations, and who carries the liability. On former industrial land this is the dominant risk, not a footnote.
• Water. Cooling demand is a real constraint and increasingly a permitting and political one. A site with power and no water answer is not solved.
• Zoning and permitting. Whether the use is permitted as of right, or requires a discretionary approval that gives someone the ability to say no.
• Community and political position. Increasingly decisive. Opposition organised around noise, water, or electricity prices can stall a permitted project indefinitely, and it is not a risk that shows up in any document.
• Fiber. Rarely fatal, occasionally is. Proximity to long-haul routes and genuine carrier diversity, not a single provider's map.

None of these makes a site financeable on its own. Each can independently make it unfinanceable.

The phrase is used to mean anything from "we have a survey" to "we could break ground next month," which is why sophisticated buyers discount it to zero on sight.

If you intend to use it, it should mean: site control that outlasts the build, a documented power position with a dated energization path, permits either in hand or permitted as of right, an environmental position that has been investigated rather than assumed, and a demand path that is contractual or honestly labelled as not yet.

The practical test is unglamorous. Assemble those five as documents in one place, in advance, before anyone asks. A seller who can produce that package on request is in a different category from one who assembles it reactively over six weeks — not because the underlying asset differs, but because the first has demonstrably done the work and the second is asking capital to fund the discovery.