US Small Business Lending in 2026: Who Gets Turned Down, What Online Lenders Charge, and Where a Startup Fits
Banks still make most small business loans, but young and riskier firms are approved far less often and fund the gap from the owner's pocket. What Census, Federal Reserve, FDIC, SBA and lender filings show about the US SME finance market, and where a new product can win.
If you are building a product for US small business finance, such as a lender, a credit marketplace, a cash-flow tool or underwriting software sold to banks, you are entering a market that looks crowded and solved. Banks have lent to small firms for a century, SBA programs guarantee tens of billions a year, and online lenders have been around for more than a decade. The useful question is narrower: which firms still can't get the money they ask for, and what does it cost to serve them? The public data answers that better than most pitch decks do.
How big the buyer base is
The Census Bureau counts 6,374,594 US employer firms with fewer than 500 employees (Census SUSB, 2022). Together they had about $17.8 trillion in annual receipts and $3.5 trillion in payroll that year (Census SUSB, 2022). The smallest are the most numerous: construction alone has 524,170 employer firms with fewer than 5 employees, and professional, scientific and technical services has 652,709 (Census SUSB, 2022).
Most of these firms already borrow. 71% of small employer firms with under $1M in revenue hold outstanding business debt (Federal Reserve Small Business Credit Survey, 2025). Money is tight across the board: 93–96% of firms in every size band from 1 to 499 employees reported at least one financial challenge in the prior 12 months (Federal Reserve SBCS, 2023–2025 waves). Only 43% of firms with 1–4 employees operated at a profit, against 73% of firms with 50–499 employees (Federal Reserve SBCS, 2025).
Who applies, and who gets turned down
The two main surveys measure applications differently, so read them separately.
Census Annual Business Survey (2023 reference year): 12.9% of all US employer firms applied for new credit. Of those applicants, 72.9% went to a bank and 22.9% to a finance company; CDFIs were 0.6% (Census ABS, 2023). Fintech and online lenders were the source for about 10–11% of applicants in firms aged 2–10 years, but only 4.9% for firms 16+ years old (Census ABS, 2023).
How often an applicant got the full amount depends heavily on the firm's age:
| Firm age | Full amount approved | Only part approved | Nothing approved |
|---|---|---|---|
| Under 2 years | 46.9% | 31.2% | 9.4% |
| 2–3 years | 49.6% | 31.3% | 8.2% |
| 6–10 years | 55.3% | 27.0% | 7.6% |
| 16+ years | 66.8% | 18.3% | 5.9% |
Source: Census ABS, 2023 reference year, US employer firms that applied for credit.
Federal Reserve Small Business Credit Survey (2025): the Fed sees more applications and a sharper gap. 35% of firms with 1–4 employees applied for new financing, and 47% of those applicants were approved for at least some of it. For firms with 50–499 employees, the approval rate was 72% (Federal Reserve SBCS, 2025).
The strongest split is by credit risk. 62% of high-credit-risk firms applied, compared with 39% of low-risk firms, but only 30% of high-risk applicants were approved for at least part of what they asked for, compared with 60% of low-risk applicants (Federal Reserve SBCS, 2025). The firms that need money most ask for it most and get it least.
When financing doesn't come, the owner fills the gap. 58% of firms with 1–4 employees used the owner's personal funds to deal with financial challenges, against 27% of firms with 50–499 employees. Among firms 0–5 years old the figure is 66% (Federal Reserve SBCS, 2025).
A share of demand never becomes an application at all. Among young firms that needed financing, 10.3% of firms under 2 years old chose not to apply, and 15–18% of firms that skipped needed financing said the application was too time-consuming (Census ABS, 2022 reference year).
What the incumbents look like
Banks are steady but slow to automate. At the start of 2026 a net 8.9% of US banks were tightening standards on commercial and industrial loans to small firms. Net demand from small firms was flat at 0.0% (Federal Reserve Senior Loan Officer Survey, Q1 2026). The FDIC's survey of banks shows how manual small business lending still is. Only 28% of large banks would auto-approve a $25,000 loan, 11% a $250,000 loan and 1% a $1M or $3M loan. Only 29% of small banks can approve a simple small business loan in one business day or less (FDIC Small Business Lending Survey, 2022). Banks themselves rate fintech lenders as ahead on customer-facing technology (56% of banks) and convenience (48%). Even so, 48% of banks had no plans to use fintech in small business lending (FDIC, 2022). Their stated blockers: cybersecurity risk (72%) and implementation cost (69%).
SBA programs are large and low-loss. The SBA guaranteed 77,600 7(a) loans in fiscal 2025, and $44.8 billion across 7(a) and 504 (SBA, FY2025). The outstanding 7(a) portfolio was $116.3 billion at the end of FY2024. Charge-offs were 0.55% of unpaid principal for 7(a) and 0.12% for 504 (SBA, FY2024). The typical guarantee is 75% of the loan (Live Oak Bancshares 10-K, 2025).
Online lenders charge for speed and access. Enova's small business book (OnDeck) earned an average annualized yield of 48% on installment loans and 49% on lines of credit in 2025. Net charge-offs were 4.6% of average balance in Q4 2025, 6.2% of the balance was more than 30 days delinquent, and net revenue margin after credit losses was 71.4% (Enova 10-K, 2025). Its typical borrower is no startup: median annual sales of $585,000 and an average of 11.4 years in operation (Enova 10-K, 2025). Square has facilitated $32.8 billion in loans and advances since 2014, underwritten on its own payment data (Block 10-K, 2025).
What this means for your product
1. The gap is in young and higher-risk firms, not in "small business" in general. An established firm with over $1M in revenue already gets credit: 64% of such applicants were approved for at least some financing (Federal Reserve SBCS, 2025). The underserved group is the firm under five years old, with 1–9 employees and a weak credit file, whose owner falls back on personal savings. A product that only re-prices credit for good borrowers is competing with banks and the SBA on their strongest ground.
2. Price is not the main objection. Friction and certainty are. Online lenders charge yields near 48% (Enova, 2025) and still grow: Enova's small business revenue rose 30.5% in 2025 (Enova 10-K). Banks themselves name speed as a core advantage: 75% cite decision speed (FDIC, 2022). Yet only a minority can decide in a day. For a startup, "approved or declined in hours, with a clear amount" is a sellable promise. So is reducing partial approvals, which hit about 31% of applicants in their first three years (Census ABS, 2023).
3. Selling to banks is a real channel, but a slow one. Half of banks say they have no plans to use fintech, and their objections are security and cost. If you sell underwriting or automation to banks, plan for long security reviews and a pricing model that doesn't need upfront integration budget. The 30% of small banks (under $10B in assets) already using fintech in small business lending (FDIC, 2022) are the warmest first segment.
4. Data is the moat, and incumbents built theirs from payments. Square lends to its own sellers, and BILL sees $371.3 billion in SMB payment volume a year (BILL 10-K, FY2026). A new lender without a data source starts with the losses: compare Enova's 4.6% quarterly net charge-off rate with 0.55% on SBA 7(a) (Enova 10-K, 2025; SBA, FY2024). Before you build the credit product, decide where your underwriting data comes from: bank feeds, accounting, payroll or payments.
What the data does not tell you
- The two surveys are not the same measure. Census ABS (about 13% of firms applied) is a large statistical sample of all employer firms. The Fed SBCS (35–46% applied, depending on size) covers firms that responded to the Fed's survey, weighted to the population. Use ABS for "how many firms", and SBCS for "how applicants differ by size, age and risk".
- Non-employer businesses are missing. Every number above covers firms with paid employees. Sole proprietors and gig sellers, who are the main audience for many fintech products, are outside SUSB, ABS and most of SBCS.
- Loan volumes are dated. The latest bank figure in our data is $754.5 billion in small business loans outstanding at commercial banks in 2021 (Federal Reserve, 2021). Treat it as order of magnitude, not as the current market.
- Online lender economics come from public filers. Enova and Block report under SEC rules, but they are two companies. Merchant cash advance providers and private lenders do not publish comparable yields or losses.
- Regulation is not covered here. Small business lending data collection and state disclosure rules affect cost to serve, but they are outside this article.
The short version: US small business credit is not an empty market. It is a split one. Banks and the SBA serve established, low-risk firms cheaply. Online lenders serve the rest at close to 50% yields. Young firms with thin credit files sit in the middle and pay with the owner's savings. That middle segment is where a startup has room, as long as it has a data advantage that brings loss rates down.
Explore the underlying data in the US Small Business Finance cell of the Atlas.
Keep reading
Stop reading. Start building.
Put this into practice with 100+ AI agents and proven frameworks — from idea to product-market fit.