When the Machine Maker Becomes the Bank: Vendor Financing in the AI Power Cycle
Written by Bruno P. Allaire
The OEMs are starting to finance their own turbine and engine sales. It makes deals easier, and that is exactly why it deserves a cold look. The manufacturer financing the buyer is one of the oldest late-cycle signals in heavy industry, and it has ended badly twice before.
Share this perspective
The Pattern Arrives
The first time a sponsor mentions it, it sounds like good news. The OEM will extend the payment terms. The OEM will lease the unit instead of selling it. The OEM, through a captive finance arm or a partner lender, will wrap financing around the whole package: equipment, installation, maybe even the maintenance contract. After two years of turbines rationed by waiting list, the factory has started offering to help you pay for one. In a market defined by scarcity, with average lead times for a new combined-cycle plant stretched from roughly three and a half years in 2023 to about five today, and longer still for the heaviest frames 1, the seller suddenly has an interest in making the purchase easy.
It should be said plainly that some of this is benign. Captive finance arms are old, legitimate institutions, and a manufacturer helping a good credit buy its product is ordinary commerce. The demand behind all of it is real enough: Goldman Sachs Research projects that global power demand from data centers will grow by as much as 165 percent by the end of the decade compared with 20232. The interesting question is not whether vendor financing exists, it is why it is expanding now, into which buyer population, and on what terms. And here the answer gets uncomfortable. The buyers being offered the most aggressive structures are the ones the traditional lenders already passed on, the mid-market sponsors whose projects are real enough to want the equipment but whose balance sheets are too thin to clear a bank credit committee.
I flagged this development in my funding piece as deserving its own treatment, and this is that treatment. The short version of the argument: vendor financing is a financing innovation in the same sense that a fever is a temperature change. It tells you something real about the state of the patient. The rest of this article is about what it tells you, and what a disciplined sponsor should do differently because of it.
The Two Historical Echoes
American heavy industry has run this experiment before, twice, and both runs are worth knowing in detail because the endings are so similar. In the competitive mania of the mid-century, the locomotive builders found themselves with more production capacity than the railroads had cash to absorb. So the builders financed the purchases themselves, and then, when the railroads' own revenues disappointed, they kept financing, until the receivables of an entire industry were sitting on the balance sheets of its suppliers. When the correction came, the manufacturers fell with their customers, and several never recovered their independence.
The second echo is closer in time and closer in character. Through the late 1990s, the telecom equipment vendors financed the carriers, and the numbers eventually became extraordinary: at the peak, the largest vendors held billions in customer financing commitments, much of it to buyers whose business plans depended on the same demand curve the equipment was supposed to serve. McKinsey later put the combined vendor-financing exposure of nine equipment suppliers at roughly $25.6 billion by the end of 2000, with Lucent Technologies alone carrying about $8.1 billion in commitments, at the time nearly a quarter of its annual revenue3. The equipment was real, the networks got built, and when the demand curve bent, the vendors wrote off receivables by the billions, Lucent alone reserving $2.2 billion for bad debts in fiscal 2001 and another $1.3 billion the year after, and the weakest customers simply vanished. The lesson was not that fiber optics were a fraud. It was that financing migrates to wherever the diligence is weakest, and in a mania, the equipment maker's credit department is the weakest diligence in the room.
Both episodes share a structure, and the structure is the point. The seller of a scarce, essential good starts advancing the capital to buy it. Demand looks infinite, so the credit risk looks abstract. The receivable grows faster than anyone admits, because the same optimism that grows the order book grows the credit book. And when demand finally pauses, the party holding the risk is the one with the least information about the buyer, because it was the only lender in the market that never needed to underwrite the buyer's whole business to make the sale.
What the Signal Actually Means
Read as information, vendor financing tells you two things at once, and both are about the seller, not the buyer. First, the OEM is protecting its order book, and the order books are enormous: GE Vernova alone closed its most recent quarter with 116 gigawatts of gas power equipment backlog and paid slot reservation agreements, and it is taking reservations for delivery as far out as 20314. A financing offer that keeps a marginal buyer committed is a way of keeping announced backlog intact through the soft part of the cycle, and backlog is what the OEM's own equity story is built on. Second, and more telling, the OEM has concluded that the constraint on its sales is no longer the buyer's desire for the machine. It is the buyer's ability to pay for the machine.
That second conclusion is the signal. Demand for power equipment in this buildout is, at the level of intent, effectively unlimited; every utility, hyperscaler, and sponsor on the continent says it wants turbines. When a seller in that kind of market starts lending, it is telling you that intent has stopped converting into orders through the ordinary channel of bankable credit. The buyer's balance sheet is now the binding constraint, and the market is quietly repricing around that fact. Vendor financing is the sound of a shortage turning into a credit cycle.
The reason this deserves a cold look rather than celebration is that the structure makes easy deals easier at exactly the moment when making deals easier is the danger. Every late cycle produces some instrument whose function is to reach the buyer the disciplined lenders passed on: the second-lien loan in 2007, the vendor lease in the telecom bubble, the sale-leaseback in the shale boom. The instrument always arrives with a good reason attached. The good reason is never the problem. The problem is what the instrument reaches.
The Credit Consequences
Now the sponsor's side of the table. What does accepting vendor financing actually cost, in a market where the equipment is both scarce and delivery-constrained? The costs concentrate in one place: the same party now holds your delivery date and your mortgage. A lender ordinarily wants you to succeed; a lender who is also your sole equipment supplier holds leverage that no ordinary lender has. If the project stumbles, the OEM does not have to negotiate the way a bank does. It can slow-walk your spare parts, reprice your maintenance, adjust your delivery queue position, all within terms that looked standard at signing, and each adjustment is a negotiation you conduct with a counterparty you owe money to and cannot replace.
The structural problem shows up first at any soft patch. The projects that took vendor financing to stretch a thin capital stack are the ones carrying a mortgage on the single asset most essential to their revenue, held by a creditor whose own fortunes depend on the same demand curve the project depends on. If the buildout pauses, the sponsor's offtake softens and the OEM's order book softens at the same time, for the same reason. The recourse analysis a sponsor should run is short: in a default, who owns the equipment, who owns the slot, and what happens to the project if the OEM itself has a credit event and the captive finance arm's paper ends up somewhere unexpected.
There is a subtler cost, too, and I think it matters most for exactly the mid-market sponsors I write for and advise. Financing that comes bundled with the machine is financing you stopped shopping for. The market rate for your credit got set inside a sales conversation, by a party with an overwhelming interest in the answer being yes, and the rate you didn't pay almost certainly got folded into the equipment price where no lender can ever compete it down. A sponsor who takes the bundle never learns what its project actually costs to finance, and in my experience the discipline of knowing that number is worth more than the convenience of not asking.
How a Sponsor Should Underwrite It
None of this means refusing vendor financing categorically. It means underwriting it as credit, with the same skepticism you would apply to any other loan, and there are situations where it is genuinely the right instrument. A short-tenor structure that bridges delivery to permanent project financing, non-recourse to the sponsor beyond the asset, priced at a rate that survives comparison to third-party debt, is a legitimate tool. The test is not who the lender is. The tests are whether the structure can be repaid by the asset's own contracted cash flows, whether the terms can be understood by someone who did not write them, and whether the sponsor could refinance the paper tomorrow at a knowable price.
The diligence list follows from the failure modes. Who actually holds the paper at the bottom of a default, and is it the OEM or an entity the OEM sold it to? What does the financing agreement say about delivery delays, given that the lender controls the delivery calendar? Is the headline rate real, or has the interest been capitalized into the equipment price, and would the OEM sell the same unit for cash meaningfully cheaper? Does the paper accelerate on the OEM's own credit events, or only the sponsor's? And the question that answers all the others: would a traditional lender, seeing this project cold, make this loan? If the honest answer is no, the vendor financing is not a financing solution. It is a transfer of a declined loan from a bank that would have said no to a seller that could not afford to.
I will admit the reflex is a personal one, learned in the worst possible school. In the trough years, running a small-cap oil and gas company, I sat across from equipment vendors with empty order books and full warehouses, and they offered us financing that no bank would have considered, because they needed revenue more than they needed clean credit. I read those term sheets line by line, and what I found was that the generous headline hid everything that mattered: cross-collateralization against service contracts, rate resets tied to conditions the vendor itself controlled, cure provisions that were really options to repossess. We took some of the deals anyway, and we earned the right to regret two of them. The market has now inverted completely, the vendor holds every card, and the terms on offer are dressed far better than the ones I read. The skepticism transfers without alteration.
The Design Alternative
There is a way out of the vendor-financing trap that does not depend on out-negotiating an OEM, and it is the design decision we made before this market ever reached its current state. It is why we formed Defiant Power Solutions around standardized, modular reciprocating units rather than large single frames. The modular architecture removes the two sources of leverage that make vendor financing dangerous at exactly the wrong scale. The units are built in volume and staged in inventory, so there is no scarce delivery slot to mortgage the project for. And the fleet is distributed, so the financing conversation is about a populated class of assets with known operating histories rather than a single bespoke machine whose terms the manufacturer can set unilaterally.
The economics read as insurance, the same way they did in my turbine piece. A sponsor that buys in increments never faces the all-or-nothing capital commitment that makes a vendor-financed frame feel necessary. A project whose architecture survives one delayed delivery has no reason to hand a lender control of its critical path. Modularity does not just shorten the lead-time exposure; it removes the desperation that the financing structures are designed to exploit. The sponsor who is not cornered does not need the corner's terms.
I want to be fair to the other side of this choice. The hyperscale sponsors with firm fuel and investment-grade balance sheets will buy their frames outright on ordinary corporate credit, and for them vendor financing is a rounding error inside a balance sheet that can absorb it. I am writing this section for the mid-market, for the sponsor whose project dies if the iron is late or the loan is called, because that sponsor is precisely the buyer the OEM financing divisions are now courting with the most aggressive paper. For that sponsor, the architecture decision and the financing decision are the same decision.
The Correction Will Sort Them
How does an outside observer use any of this? The same way I have read this market from the beginning: by the paperwork, not the press release. A sponsor whose equipment financing sits at market rates, with third-party lenders and recourse terms that can be read, is a different credit than a sponsor whose capital stack quietly ends at an OEM's captive finance arm, whatever the two projects' announcements have in common. The interconnection filings and the fuel contracts tell you whether a project can be built. The equipment paper tells you who actually believes in it enough to lend against it at a price that is real.
The window has the same shape my funding piece described. Vendor financing will keep expanding through the frothy part of this cycle, because that is what vendor financing does; the OEMs will report record backlogs, the credit books will grow faster than anyone discusses, and a meaningful share of the announced buildout will be financed by the same companies supplying it. When the correction arrives, and every demand curve this steep has bent eventually, the receivables will reprice quickly and the slot assignments will move to the sponsors who financed with paper that was shopped. The shortage punished the unprepared first. The turn, as my turbine piece argued, hands the prepared their best entry prices, and the prepared will be the sponsors who can pay for the entry without their equipment lender's consent.
The machine maker becoming the bank is an old story, and it has never once been a story about the machines. It is a story about where the weakest diligence was when capital went looking for a home. The sponsors who remember the locomotives and the telecom switches will treat the OEM's financing offer as what it is, a piece of market information about who cannot get a real loan, and they will keep their delivery dates and their mortgages with different counterparties. That separation is the cheapest risk management this market offers.
Notes & Sources
- 1.Siemens Energy reported gas-turbine lead times of three years or more alongside a 69 GW gas services backlog in its fiscal third-quarter 2026 results, and industry reporting puts average lead times for a new combined-cycle plant at roughly five years, up from about three and a half in 2023, with some heavy-duty frames at seven or more. Utility Dive, "Siemens Energy's gas turbine backlog nears 70 GW as company expands manufacturing," August 2026; OilPrice, "US Power Boom Triggers Global Gas Turbine Shortage," 2026
- 2.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
- 3.McKinsey estimated the combined vendor-financing exposure of nine telecom equipment suppliers at approximately $25.6 billion by the end of 2000. Lucent Technologies carried roughly $8.1 billion in customer-financing commitments, about a quarter of its annual revenue at the time, and booked provisions for bad debts of $2.2 billion in fiscal 2001 and $1.3 billion in fiscal 2002. CFO Magazine, "What Goes Around," March 2003; TheStreet, "Cisco, Lucent and Nortel: Prime Lenders for the Network Buildout," 2001; American Affairs, "Who Lost Lucent? The Decline of America's Telecom Equipment Industry," August 2020
- 4.GE Vernova closed its second quarter of 2026 with 116 GW of gas power equipment backlog and slot reservation agreements, up from 83 GW at the end of 2025, and is taking reservations for delivery as far out as 2031. Utility Dive, "GE Vernova gas turbine backlog climbs to 116 GW," July 2026