Firm Fuel or No Deal: The Siting Question Nobody Asks Out Loud
Written by Bruno P. Allaire
Boards have learned to ask when the power arrives. The sharper question is where the gas comes from, and whether the supply is firm. A co-located campus without contracted transport is a stranded investment waiting for a cold week.
Share this perspective
The Question Behind the Question
Two years of education have done their work. A board reviewing a new campus no longer opens with fiber, water, and incentives. It opens with power, and usually with a specific date attached. That is progress, and it is also where the diligence stops, which is the problem.
When can we get power is a schedule question. Where does the gas come from, and is it firm, is a solvency question. A campus can clear its interconnection study, place its equipment order, sign a lease with a hyperscaler tenant, and still fail, because the fuel behind the fence was never contracted at a price or a volume that survives a cold February. The failure does not look like a failure in the announcement. It looks like an escalation clause, a curtailment notice, or a phone call from a lender who has just read the supply agreement for the first time.
I have asked this question on the upstream and midstream side of the business for most of my career, and I ask it now in every siting conversation I sit in on. It is a shorter conversation than the interconnection discussion and a much more decisive one. Gas is not scarce in the United States. Firm gas, delivered to an isolated load at a known price for fifteen years, is scarce, and it is contracted years before anyone needs it.
The demand side of this is not in doubt. Goldman Sachs Research projects data center power demand growing by as much as 165 percent by the end of the decade against 20231. The supply side is where projects are decided, and the supply side does not scale on a model release cycle.
What Firm Actually Means
The word firm gets used loosely in siting decks, so it is worth being precise. Firm supply means gas you are entitled to receive, in the volumes you need, on the days the system is most constrained, at a price the contract defines rather than the spot market. Everything short of that is an intention with a delivery mechanism attached.
There are three pieces, and all three have to exist. The commodity, which is the molecules themselves and is generally the easiest to obtain. The transport, which is the capacity to move those molecules to your meter on the day you need them, and which is where most projects actually fail. And the deliverability at the point of use, which means a lateral, a meter station, and a pressure profile that matches your equipment. A sponsor who has a supply agreement and no transport capacity owns a document, not a fuel position.
Interruptible transport is the trap that catches the unprepared. It is cheap because it is interruptible, and it is interruptible on precisely the days your campus is worth the most to its tenant. The pipeline does not care that your load is a data center with a 99.99 percent availability commitment. It cares about the firm shippers who contracted ahead of you, and in a constrained basin there are more of them every quarter.
Transport Is the Contract That Matters
Here is the structural feature of this market that siting teams consistently underweight. Pipeline capacity in the regions with the best data center load growth is largely spoken for, and the incremental capacity requires a federal permitting cycle that runs on its own clock. New laterals and expansions are being proposed. They are not being energized quickly, and they are not being built for a single campus unless that campus is large enough to anchor the whole project.
That creates a negotiation with an unusual shape. The sponsor needs the pipeline more than the pipeline needs the sponsor, unless the sponsor is willing to sign a long-term firm transportation agreement at a volume that justifies the build. Those agreements are typically fifteen to twenty years and they are not free, and they are the single largest pre-development commitment a co-located project makes. Sponsors who resist that commitment are, in effect, asking the pipeline to underwrite their project.
The consequence for the deal is straightforward. Firm transport converts an interruptible fuel arrangement into a contractable one, and contractable fuel is what makes the generation financeable. A lender will lend against a project with fifteen years of firm transport and a matching offtake. It will not lend against the same project with a promise and a basis hedge. The fuel contract is not a procurement detail on the way to the financing. It is the financing.
The Cold Week Test
Every energy operator I respect runs some version of one scenario before committing capital: the coldest week of the last twenty years, replayed on the project. In that week, spot gas at the affected hub does not rise by ten percent. It rises by multiples, because residential and industrial demand both spike at once and the system rations through price. If your campus is buying at that hub on that day, you are not operating a data center. You are operating a bid.
The cold week test also exposes the basis problem, which is the quieter half of the risk. A campus can be sitting on top of the cheapest gas basin in North America and still pay a premium, because the local delivery point is constrained even when the basin is not. Basis blowouts are regional, they are seasonal, and they are exactly the moments when a sponsor with a merchant fuel arrangement discovers that its economics were underwritten on an average that never appears in the month it matters.
A project passes the test in one of two ways. Either the fuel is contracted firm at a defined price with transport behind it, or the offtake contract explicitly passes fuel cost through to the tenant with a cap the sponsor can survive. The second is legitimate and common, and it moves the risk to the party best able to manage it, which is the tenant with the balance sheet. What is not legitimate is a sponsor carrying the fuel risk implicitly, in a model, with a basis assumption nobody tested.
How to Diligence Fuel Before You Site
The diligence sequence is short enough to run in a week and it should run before the option payment, not after. Start with the pipeline: which system serves the site, what firm capacity exists on it today, who holds that capacity, and what does the expansion queue look like for the next three years. The answers are public enough to be found and specific enough to change a siting decision.
Then the contract terms. Firm or interruptible, the volume profile, the term, the primary delivery point, the credit support the pipeline requires, and the conditions under which the shipper can be curtailed. Read the curtailment provisions the way a lender will, because a lender will. Then the equipment side: does the generation you have selected run on the pressure and quality the system delivers, or does it need compression and treatment that nobody has budgeted.
Finally, the credit question that ties it together. A fifteen-year firm transport agreement is a balance sheet commitment. A sponsor who signs it and then fails to finance the project has bought a long-dated obligation and no asset. That is why, in our own development work at Defiant Power Solutions, we treat fuel, equipment, and permits as one underwriting problem rather than three procurement workstreams. A site with excellent power economics and no firm fuel is not a project. It is a land position.
What the Answer Tells You
For anyone reading this market from the outside, whether as an investor, a tenant, or a counterparty, the fuel question is the fastest way to separate the announced from the real. Ask a sponsor where the gas comes from. If the answer is a basin, a hub, and a hedging strategy, the project is earlier than its press release suggests. If the answer is a named pipeline, a firm transport agreement, a delivery point, and a term that matches the offtake, the project has done the work that most of its competitors have not.
That distinction will matter more as the cycle matures, because the constraint is migrating. The interconnection queue is slow and visible, and it has been discussed for two years2. The fuel behind it is neither. A campus that solved power on paper, at a price that assumed abundant gas delivered whenever needed, is the most likely candidate for the stranded asset category of this buildout, and it will fail quietly, in an escalation clause, long after the announcement.
The boards that learn to ask the second question are the ones that will still be operating these campuses when the first question stops being interesting. Firm fuel or no deal is not a slogan. It is the difference between owning generation and owning an obligation to buy gas at whatever the worst week of the decade costs.
Notes & Sources
- 1.Goldman Sachs Research projects that global power demand from data centers will increase by as much as 165 percent by the end of the decade compared with 2023. Goldman Sachs Research, "AI to drive 165% increase in data center power demand by 2030," February 2025
- 2.The US interconnection queue peaked at nearly 2,600 GW of active generation and storage requests at the end of 2023, and the median duration from request to commercial operation exceeded five years for projects built in 2025. Lawrence Berkeley National Laboratory, "Queued Up" (2025 and 2026 editions)