Private credit's 8.6% share holds because banks are not bidding

Nonbank lenders averaged 4.6 percent of U.S. commercial real estate loan originations from 2010 through 2020, and from 2021 through the third quarter of 2025 the average was 8.6 percent, per Invesco Real Estate's reading of Mortgage Bankers Association data — the floor under that second number is what carries the argument. A share that holds while the conditions that inflated it pass is a different object from one that tracks a dislocation, and Alex Horn, founder and managing partner at BridgeInvest, makes that case in an IREI interview, one in which the servicing line, not the origination share, decides whether the word permanent survives.

The 8.6 percent measures new originations, a flow, and says nothing about the stock of outstanding U.S. commercial real estate debt, where bank balance sheets are still the market, nor about pricing. A larger share of originations can reflect banks declining to bid as easily as private lenders winning on terms, and the data series cannot tell the two apart.

Horn's account of what changed comes in three parts, the first being the cycle itself: the shift held through a full rate cycle rather than a single dislocation, which he traces to Basel III and Dodd-Frank capital constraints that outlasted the conditions that produced them. The rest of the conversation runs to what institutional investors are asking of managers, where the firm is exercising caution, and how the lending market could look three to five years out — the last item the only honest test of the word permanent.

Banks pulled back, refinancing needs grew, and private lenders took a larger role across middle-market originations, servicing and workouts; the first two are a capital story any credit shop can ride, while the third is where a lender learns whether it built a business or caught a wave and where the pitch to allocators has moved from headline yield to capability. The allocator case rests on CRE debt's steadier 9 to 10 percent across rate environments — twenty years of Cambridge data — and it is being made against a $3 trillion maturity wall rather than in spite of it.

The middle market's appeal is arithmetic: loan sizes there are large enough to price for the risk and small enough that one lender can hold the whole position, which keeps syndication and intercreditor negotiation out of the file and leaves the originator with the relationship, the servicing, and the workout if it comes to that. Turn that over and the caution IREI's framing alludes to is not hard to infer: in a single-lender deal there is no syndicate to absorb a misjudgment.

Eight-point-six percent is roughly one CRE loan in twelve, a minority share to hang a word as heavy as infrastructure on, and the more useful question than whether the label is earned is why the share is sticky at all. The likeliest answer is that banks are not competing for it: middle-market loan sizes are too small to move a money-center balance sheet, so the segment reads as one the banks conceded rather than lost. That makes the share durable for reasons that have little to do with the sophistication of the firms holding it — and reversible on a decision made in a boardroom rather than in a debt fund.

Refinancing needs are growing, and the 2025-2026 renewal wave this publication has tracked in Canada is the same repricing window appearing in the U.S. at larger scale: performing assets that cannot be financed at the old coupon, handed to lenders who can reset basis. Where private debt has already done that, it has priced stabilized assets to current cash flow rather than to a recovery. A Newark tower taken out with $70 million of interest-only nonrecourse debt is the template, and the Miami Beach refinancing below follows it.

BridgeInvest's own book is small enough to read as a case study: 33 employees and $859 million of regulatory assets, per PRED's records, against a $612 million closing recorded in August and the $114.3 million Miami Beach refinancing in September, three weeks apart. The Miami Beach loan is the more instructive of the two: a takeout on a stabilized 83-percent-leased office asset whose next 47,000 square feet require a municipal election, so the lender is paid on today's cash flow and has left the upside with the tenants, the voters and the sponsor. That is the right structure for a shop that shares the view that office price discovery remains unfinished outside the trades that print.

The honest cost of the permanence thesis is that a middle-market lender that services and works out its own loans carries a bench that earns nothing in a good year and everything in a bad one, and the spread it charges is the wage for that bench. Origination share flatters the model, while workout performance prices it. Allocators appear to be pricing it already: the survey high in alternatives appetite in September arrived alongside a five-year peak in co-investment demand, which suggests limited partners want a seat closer to the loan than a fund commitment provides.

Three to five years out, the servicing line will say more than the share. If 8.6 percent still reads 8.6 percent in 2030, the question worth asking is how much of it belongs to firms that can service what they originate and how much still belongs to the shops that arrived for the spread, and whether a bank, once its capital constraints finally lift, decides the borrower is worth taking back. BridgeInvest's arithmetic is the smallest version of that test: a 33-employee, $859 million platform writing nine-figure closings into a market it now calls permanent. The answer arrives on files like that Miami Beach loan, on the maturity date attached to them, not in the quarterly share print.