Banks left construction lending. Private credit sent the invoice.

Between October 2023 and October 2024, commercial real estate loan volume originated by alternative lenders rose 34 percent while bank-originated volume fell 24 percent. Same asset class, same borrowers in many cases, opposite lines on the chart. That is not a market finding its own level. That is one lender group leaving a room and another one walking in to take the seats.

Nobody at the door was surprised. The regulators told the banks to go.

The bank retreat from construction lending is a policy choice, not a cycle

Federal regulators have flagged banks with heavy commercial real estate concentrations relative to capital for years, and the Basel III endgame framework, layered on top of FDIC guidance on CRE concentration risk, gives that scrutiny teeth: hold more capital against the same construction loan than before, and a construction loan on a bank balance sheet gets structurally less attractive relative to almost anything else the bank could do with that capital. Regional and mid-size banks, historically the workhorse lenders for construction and land development, are the group with the least room to absorb that. The retreat reads as structural rather than cyclical for the simple reason that it survives improving credit conditions. The January 2026 Senior Loan Officer Opinion Survey found banks easing CRE underwriting standards for the first time since the 2022 tightening cycle began. Conditions improved. The retreat did not reverse with them.

By April 2026, the SLOOS showed why. Standards on construction and land-development, nonfarm nonresidential, and multifamily loans were basically unchanged in aggregate, but that aggregate hid a split by bank size: large banks eased across all three categories, while smaller banks moderately tightened construction and land-development standards and modestly tightened multifamily. The most frequently cited reason banks gave for easing CRE policy over the prior year was blunt: competition from other banks or nonbank lenders. Banks are not uniformly leaving. The ones with the least regulatory room are leaving, and even the ones staying are being priced by the ones that left.

Private credit did not fill the gap out of charity

It filled it at a price, and the price is the point of this piece. Construction and bridge loans price as a spread over SOFR, which sits at roughly 3.66 percent today. A strong sponsor on a straightforward deal can still get bank-quality pricing, something like SOFR plus 250 to 450 basis points. A mid-market developer, or any project a bank has decided sits outside its shrinking risk appetite, is more likely to see SOFR plus 400 to 600, and market participants describe spreads as having widened a further 50 to 100 basis points since late 2025 even as private lenders describe their own terms as improving, meaning better covenants and less payment-in-kind risk for the lender, which is the same thing as worse terms for the borrower.

Proceeds moved with price. What used to be a 75 to 80 percent loan-to-cost construction loan before 2022 is now commonly quoted around 60 to 70 percent, with some banks capping straightforward deals near 65 percent loan-to-cost. Run that through a $100 million project. At 75 percent loan-to-cost the sponsor brings $25 million of equity. At 65 percent, the same project needs $35 million. Ten million dollars of additional equity, for the identical building, before a single basis point of interest-rate change gets counted. Add the spread widening on top of the smaller loan and the total cost of capital on the deal has moved on two axes at once, not one.

Stack the two effects on the same $100 million project and the gap gets wider, not additive. The $65 million loan that replaced the $75 million loan is also more likely to be priced at the mid-market spread than the bank spread, since a project a bank has pushed toward 65 percent proceeds is frequently one it already viewed as marginal. At SOFR plus 500 rather than SOFR plus 300, that $65 million loan costs roughly $1.3 million a year more in interest than it would have at bank pricing, on top of carrying $10 million less proceeds in the first place. A sponsor comparing today's term sheet against a 2021 memory is not looking at one number that moved. They are looking at two numbers that both moved against them at the same time, on the same deal.

There is a real, large-scale illustration of the alternative-capital side of this sitting in the public record already. Crusoe, Blue Owl Capital, and Primary Digital Infrastructure closed an $11.6 billion financing package in May 2025 to fund six data-center buildings outside Abilene, Texas, inside a campus that runs to a $15 billion joint venture at full build. Blue Owl is a direct lender, not a bank. That is not a story about a niche corner of the market anymore. It is the largest, most closely watched infrastructure financing of the cycle, and it was priced, structured, and closed by exactly the kind of capital this piece is describing, at hyperscale, because a bank syndicate was never going to carry that concentration alone even before Basel gave it a reason to say no.

The binding constraint moved, and most sponsors have not noticed

Loan sizing used to run on the lower of loan-to-cost and loan-to-value. Increasingly it runs on debt yield instead, stabilized net operating income divided by the loan amount, because rates have moved further and faster than cap rates have. Most banks want 8 to 10 percent debt yield before they will size proceeds at all, and in a market where rates rose faster than net operating income, debt yield becomes the tightest of the three tests almost by construction. A sponsor who still negotiates a term sheet by talking about loan-to-cost first is negotiating the constraint that used to bind, not the one that binds now.

That matters because it changes what "shopping the deal" actually accomplishes. Two years ago, a sponsor unhappy with a bank's loan-to-cost quote could reasonably expect a competing bank to offer a better one. Today, if debt yield is the binding constraint and every lender is underwriting to roughly the same 8 to 10 percent floor against the same projected net operating income, shopping the deal moves the spread a little and moves proceeds barely at all. The lever that actually moves proceeds is the net operating income assumption itself, which means the pre-leasing and rent-comp work a development team does before it ever calls a lender now carries more weight on total proceeds than the lender relationship does.

What this stacks against

None of this is happening against a quiet backdrop. The Mortgage Bankers Association counts $875 billion of commercial and multifamily mortgage balances maturing in 2026, 17 percent of the $5.0 trillion outstanding, with hotel loans at 30 percent maturing, industrial at 23 percent, and office at 17 percent. Every one of those loans refinances into the environment this piece just described: fewer bank dollars chasing construction and transitional risk specifically, wider spreads on the dollars that remain, and a debt-yield test that does not care how generous the original loan-to-cost was in 2019 or 2021. A sponsor refinancing a loan originated at 78 percent loan-to-cost under 2021 underwriting into a 65 percent world is not looking at a renewal. They are looking at a capital call, on a schedule set by a maturity date, not by market convenience.

I think the split keeps moving in the same direction through the rest of this cycle. My bet: private credit's share of new construction originations passes bank share for the first time outside an outright recession before 2028, not because banks want out permanently but because Basel's capital treatment makes construction lending structurally worse capital-efficiency for a depository than it is for an unregulated fund, and that treatment does not sunset on its own. If regional banks meaningfully re-enter construction lending at pre-2022 proceeds levels by 2028, I was wrong about how permanent this was, and the industry should stop describing Basel III as a settled fact.

Whatever the split ends up being, price to the lender you can actually close with today, at the spread and the debt yield they are actually quoting, not the one your relationship banker quoted before the rules changed underneath both of you.

This analysis is a source-cited research summary drawn from public records, not legal advice. It can contain errors and should be verified independently before any investment decision.