Most of the unitranche premium is buying leverage, not convenience.

What does each structure cost right now?

At a 1-month Term SOFR of 3.65% (CME, 4 August 2026), a lower-middle-market borrower faces a ladder rather than a choice between two prices. Commercial bank senior cash flow debt runs S+350 to 425 below $10m of EBITDA, or 7.15% to 7.90% all-in. Non-bank senior and unitranche runs S+550 to 750 in the same band, or 9.15% to 11.15%. Junior capital and mezzanine, counting cash and payment-in-kind together, runs 13.00% to 16.00% (SPP Capital Partners, Market At A Glance, July 2026).

Size compresses all of it. The same unitranche is S+425 to 575 for a borrower above $25m of EBITDA, or 7.90% to 9.40%. That is 125 to 175 basis points inside the sub-$10m band at identical seniority, which on a $20m facility is $250,000 to $350,000 a year for nothing other than being larger and better packaged.

Senior stretch, the structure that sits between the two, is a single senior facility levered past what a bank credit committee would normally fund without a second tranche behind it. In the core middle market it prices at S+425 to 525 at $40m to $100m of EBITDA against unitranche at S+475 to 550 in the same band, roughly 25 to 50 basis points inside unitranche, with cash flow senior at S+400 to 500 (Lincoln International, Private Credit Snapshot, as of 1 May 2026).

The spot unitranche print for the market as a whole is 9.00% to 9.75% (VRC, Private Markets Trends Q2 2026, July 2026), against 8.9% all-in in the first quarter (Lincoln International, published 7 May 2026). For context on how far this has already travelled, the same Lincoln series showed S+550 to 625 over a 5.3% base rate for an 11.8% all-in cost in the first quarter of 2024. Spreads have tightened by roughly 75 basis points since then. Base rates have done all the rest of the work.

Is unitranche really more expensive than senior plus mezzanine?

Compared like for like, barely. The headline gap of 200 to 325 basis points between bank senior and unitranche is not a price comparison at all. It is a comparison of two different amounts of debt.

A commercial bank in this market funds roughly 2.5x to 3.0x. A first-lien lower-middle-market portfolio reports 2.8x weighted-average senior leverage on new platform deals (Capital Southwest, quarter ended 30 June 2026) and a comparable book shows median net senior debt to EBITDA of 2.5x on companies averaging $11.2m of EBITDA (Main Street Capital, as of 31 March 2026). Unitranche underwriting bands run 4.00x to 5.50x below $15m of EBITDA and 4.50x to 6.00x from $15m to $40m (Lincoln International, as of 1 May 2026). A borrower choosing unitranche over a bank facility is not buying the same loan at a higher rate. It is buying about two extra turns.

So hold the leverage constant. Take a borrower above $10m of EBITDA targeting 4.5x total, and build it the two-tranche way: 3.0x of bank senior at 6.65% to 7.40%, plus 1.5x of mezzanine at 12.00% to 14.00%. Weighted by tranche size, that blends to roughly 8.4% to 9.6%. Unitranche in the same size band is 8.65% to 10.15%. The convenience premium at equal leverage is therefore in the order of 20 to 55 basis points, not 300. Below $10m of EBITDA the same exercise gives a blend of roughly 9.1% to 10.6% against unitranche at 9.15% to 11.15%, which at the bottom of the range is close to a wash.

That arithmetic is ours, not a published figure. It uses the SPP ranges above at a stated 3.0x plus 1.5x structure, ignores the fee and closing-cost differences between one lender and two, and assumes both structures are available to the borrower, which for many founder-owned businesses is the assumption that fails first. Change the leverage split and the answer moves. The point is not the precise number. It is that the comparison every term sheet invites, coupon against coupon, answers a question nobody is actually asking.

What does the premium actually buy?

Four things, in descending order of what borrowers report caring about after the fact.

Leverage headroom, which is most of the price, as above. Loan-to-value on unitranche sits at 50% below $100m of EBITDA and 55% above it (Houlihan Lokey, as of 30 April 2026), against 29% weighted-average loan-to-value on that first-lien bank-style book (Capital Southwest, 30 June 2026).

One document and one counterparty. There is no intercreditor agreement to negotiate, no agreement among lenders, and no second creditor whose consent is needed when something has to change. That matters far more in year three than at signing, and it is systematically underweighted at signing.

Speed and certainty of close. A single credit decision compresses the financing timetable, which in a competitive sale process is worth something the coupon does not capture.

And a covenant package that is negotiated once. Equity cushions of 45% and above are now standard at $40m to $100m of EBITDA (Lincoln International, 1 May 2026), which is the lender's protection rather than the borrower's, but it is disclosed once rather than twice.

What is not in the coupon?

Original issue discount, first. New-issue unitranche and first lien below $20m of EBITDA prices at 98.0 to 99.0 (Houlihan Lokey, as of 30 April 2026), and lower-middle-market unitranche is quoted at 2% to 3% at funding (ABF Journal, 1 June 2026). Discount has been widening while spreads have not: lenders quoting 99 or tighter fell from 75% to 56% between the second quarter of 2025 and the second quarter of 2026, and quotes at 98.5 more than doubled from 18% to 42% (Houlihan Lokey, Q2 2026 Private Credit Survey). On a five-year facility, two points of discount is roughly 40 basis points of yield the term sheet does not show as interest.

Call protection, second, commonly 102 in year one, 101 in year two and par thereafter (ABF Journal, 1 June 2026). A borrower expecting to refinance inside two years is paying for an option it intends to exercise against itself.

Sector, third. Software carries 75 to 100 basis points above the standard matrix on one reading (Houlihan Lokey) and 150 to 300 basis points above comparable non-software credits on another (SPP Capital Partners, July 2026), with interest-only periods cut from three years to two or less and liquidity minimums now standard.

And the grids themselves disagree, which is worth knowing before either is used as a benchmark. SPP prints S+550 to 750 for sub-$10m EBITDA unitranche. Lincoln prints S+475 to 550 for sub-$15m. Both are current and both are correct, because they measure different populations: SPP surveys actual lower-middle-market placements, while Lincoln values a portfolio whose median company carries $60.6m of last-twelve-months adjusted EBITDA and skews heavily sponsor-backed. For a founder-owned business below $10m of EBITDA, SPP is the honest comparison, and quoting Lincoln at that size will produce a term sheet that disappoints.

What moves the number from here?

Base rates, not spreads, on the current evidence. Unitranche spreads were flat month over month at S+498 in April 2026 (KBRA DLD via Houlihan Lokey, as of 30 April 2026), while swap-adjusted unitranche yields rose roughly 47 basis points quarter over quarter to 9.47% to 10.22% (VRC, Q2 2026). The move came from the forward curve.

That curve prices 3-month SOFR at 3.90% at the end of 2026 and 4.04% at the end of 2027, against 3.76% today (Blue Gamma, 4 August 2026, and CME, 4 August 2026). It is pricing base rates higher at the end of next year than they are now, and higher than the Federal Reserve's own 2027 median projection. Three officials dissented in favour of a rate rise at the July 2026 meeting.

Spreads are expected to add a little on top at the smaller end: a further 25 to 50 basis points of widening in lower-middle-market unitranche over two quarters (ABF Journal, 1 June 2026), and 25 to 50 basis points generally on lower-quality or harder-to-finance credits (PitchBook, US Credit Markets Quarterly Wrap Q1 2026, cited 18 May 2026). No named source publishes a 2027 spread forecast for direct lending, and none is offered here.

The practical consequence for a borrower with a 2027 or 2028 maturity is that waiting is not a free option in the way it was for the previous eighteen months. Whatever the structure, the base rate underneath it is priced to be higher when the refinancing lands than it is while the decision is being made.