The Capital Didn’t Leave. The Doors Did.

SBA 7(a) approvals fell by a third this year and participating lenders hit a thirty-year low — while total dollar volume sat on its decade trend. What actually happened, and what quietly opened.

By The SUMMANTIS Strategic Advisory Team

Capital Education Series

Read the headline numbers from this fiscal year and the conclusion writes itself: small business credit has dried up.

FY2026, FIRST NINE MONTHS	CHANGE VS FY2025
7(a) loans approved, by count	Down 33.4%  (40,824 from 61,270)
7(a) dollars approved	Down 20.9%  ($21.8B from $27.6B)
Loans of $500,000 or less, by count	Down roughly 38%
Loans above $500,000, by count	Down roughly 15%
The $350,000 to $500,000 band	Down 64% — the hardest hit
Participating lenders	1,141 — a 30-year low, down 18.9%

Two hundred and sixty-six lenders stopped participating in the program. The band between $350,000 and $500,000 — the size a growing business most often asks for — contracted by nearly two-thirds. If your bank was one of the ones that left, the relationship you spent years building no longer moves you to the front of any queue.

And yet the conclusion is wrong.

The Dollars Are on Trend

Measured against fiscal 2024 rather than fiscal 2025, total 7(a) dollar volume this year is up roughly 2.3 percent — and closer to 10 percent higher once you adjust for the volume that was pulled forward into September 2025 ahead of a government shutdown. Total dollar volume is sitting on its decade-long trend line.

The Federal Reserve’s Senior Loan Officer Opinion Surveys in January and April of this year reported small-firm loan demand as essentially unchanged. Businesses did not stop wanting capital, and the system did not stop deploying it.

Three things converged to produce the drop in loan count. A forty-three day government shutdown froze new approvals through October and November of 2025.

Tighter underwriting took effect under the standard operating procedure revision that landed in mid-2025. And the comparison base itself was inflated, because fiscal 2025 was an unsustainable peak lifted by real demand and a pull-forward ahead of the tighter rules.

Strip those out and what remains is not a credit contraction. It is a concentration.

The same dollars are moving through fewer lenders to fewer, larger borrowers. Which means access is now far more a function of which door you approach and how the request is built than it was two years ago. The capital is there. The number of ways to reach it fell.

Three Rules That Opened While the Doors Narrowed

This is the part that has gone almost unreported, and it runs in the opposite direction from the headline.

EFFECTIVE	WHAT CHANGED	WHY IT MATTERS
March 1, 2026	The credit-score prescreen that auto-declined small-loan applicants below a set threshold was retired. Lenders now run their own credit analysis, with a debt service coverage ratio of at least 1.10.	A file that was rejected by a number before a human read it can now be underwritten on its cash flow.
July 4, 2026	The 7(a) and 504 limits were separated. An outstanding 7(a) balance no longer reduces the 504 debenture available.	A business carrying 7(a) debt now has up to $5 million in fresh 504 capacity — $10 million across both programs.
Current	The 7(a) Small Loan category, with lighter documentation and streamlined credit analysis, now runs up to $350,000 rather than $500,000.	It redraws the line between the streamlined track and the standard one. Where your request falls relative to $350,000 changes how it is processed.

The second of those is the one most owners should look at twice. Under the old rule, a business carrying a $4 million 7(a) balance had roughly $1 million of remaining SBA capacity — below the minimum size of most 504 projects, which is another way of saying it had none. That same business now has a separate $5 million lane available to it.

The Door Almost Nobody Is Looking At

While 7(a) lending fell by roughly a quarter year over year through the spring, 504 origination volume was down only about seven percent over the same period, and recovering steadily from the shutdown.

The 504 program also carries the cheapest fixed-rate money in the SBA toolkit. The CDC portion priced in the low-to-mid six percent range at the August debenture sale, fixed for the full ten, twenty, or twenty-five year term, against variable 7(a) pricing that runs roughly ten to thirteen percent. Credit performance backs it up: cumulative net charge-offs on 504 loans have stayed below one percent for every fiscal year since 2015.

So the program that held up best through the contraction, prices cheapest, locks its rate for the life of the loan, and just had its capacity decoupled from 7(a) is also the one that sits outside most owners’ field of view entirely.

Where This Breaks

Four limits, stated plainly, because the optimistic reading of this has real edges.

504 is not a substitute for what disappeared. It finances fixed assets — owner-occupied real estate and long-life equipment. The facilities that vanished from the market were largely small working capital loans, and no amount of 504 capacity fills that gap. A business that needs $300,000 to carry inventory is not helped by a program that will not fund inventory.

A separated limit is a ceiling, not an approval. Decoupling 7(a) from 504 raised how much you may borrow. It did not change whether you qualify, and a business already carrying $4 million of SBA debt faces a demanding conversation about global cash flow regardless of what the cap allows.

Retiring the credit prescreen does not mean credit stopped mattering. It means a number no longer declines the file before a person reads it. Lenders now apply their own analysis, and their own thresholds are frequently stricter than the one that was removed.

Program rules are revised on a schedule. A further revision to the standard operating procedure applies to applications issued an SBA loan number on or after October 1, 2026; files already in process before that date proceed under the prior rules. Anything written about program mechanics, including this article, is accurate as of a date and should be confirmed against the current SOP.

What This Changes About How You Approach It

In a market with a thousand-odd participating lenders instead of fourteen hundred, the shotgun approach stops working. Applying broadly and waiting was never efficient; it is now actively expensive, because each decline consumes weeks and leaves a credit inquiry behind.

What replaces it is narrower and slower to set up. Identify the handful of lenders still active in your size band and your industry, because the ones that exited are not coming back this fiscal year. Match the request to the right program before you make it, rather than defaulting to 7(a) because it is the one you have heard of. And structure the request so it reaches an underwriter already resembling the file they want to approve.

None of that is new advice. What changed is the cost of ignoring it. When fourteen hundred lenders were writing small loans, a mediocre approach eventually found a yes.

With a third fewer doors, it mostly finds silence.

The businesses getting funded this year do not have better numbers than the ones that did not. They asked a lender who was still lending, for a program that fit, in a shape that could be approved.

BEFORE YOU APPROACH A LENDER

In a market with fewer participating lenders, which door you knock on and how the request is structured carry more weight than they did two years ago. Summantis works with owners on both — matching the request to the program, and shaping it before it reaches an underwriter.

summantis.com  ·  (661) 213-9152

Educational content only. Not financial, investment, legal, or tax advice. Program figures reflect SBA loan-level data and program reporting for fiscal year 2026 as compiled by third-party analysts, and are accurate as of publication; rates, thresholds, and program rules change with SBA revisions and vary by lender, borrower, and transaction. Nothing here is an offer of credit or a quotation of terms, and no approval or funding amount is guaranteed.