Top SBA 504 Lenders for Industrial Property

First-mortgage lenders on industrial 504 deals

An SBA 504 purchase is two loans. A bank or credit union writes the first mortgage, typically around half the project, and a Certified Development Company writes the SBA-backed second. The table below ranks the first-mortgage lenders, because that is the institution an industrial borrower is actually shopping.

Figures are first-mortgage dollars (SBA's ThirdPartyDollars field) across 4,658 industrial 504 loans approved in FY2021–FY2026, cancelled approvals removed. SBA 504 requires the borrower to occupy the building, so this sample is owner-user deals only: it excludes investor-owned industrial and every non-SBA execution. It is a window on the market, not the market.

Two things worth reading off this table. Concentration is low: 1,029 different institutions wrote a first mortgage on an industrial 504 deal in this window, and the largest wrote 184 loans across six fiscal years. No bank owns this market, so there is no single obvious call. And among these twenty, average loan size ranges from about $1.5M to $4.1M, a 2.7x spread and a better guide to whether a given lender will look at your deal than its rank is. Across all 1,029 lenders the spread is wider still, because the tail is mostly banks that wrote a single local deal.

The CDCs writing the SBA second

The other half of a 504 is the CDC. These are ranked by 504 debenture dollars on the same set of loans.

Industrial 504 approvals by fiscal year

Read FY2026 with care. This extract is as-of 30 June 2026 and a federal fiscal year closes on 30 September, so FY2026 holds at most three quarters here and is additionally subject to reporting lag. It is shown for completeness. No year-over-year conclusion should be drawn from it, and none is drawn here.

Where industrial 504 deals get done

Methodology, in full

Everything here is reproducible. The source is the U.S. Small Business Administration's loan-level 7(a) & 504 FOIA extract, published at data.sba.gov and updated quarterly. The file used is FOIA_504_FY2010_Present, as-of 30 June 2026.

• Industry filter. NaicsCode beginning with 3 (manufacturing, sectors 31–33) or in 484 (truck transportation), 488 (support activities for transportation), 492 (couriers) or 493 (warehousing and storage).
• Window. ApprovalFY FY2021–FY2026.
• Cancelled approvals excluded; pipeline included. Rows with LoanStatus of CANCLD are dropped: 507 in this window, none of which carries a disbursement date. Rows marked NOT FUNDED are kept. An earlier version of this analysis excluded them as "never funded" and that was wrong: NOT FUNDED is 0.0% of industrial 504 approvals in every fiscal year from 2013 through 2020 and then climbs with recency to 77% of FY2026, which is the signature of a not-yet-disbursed pipeline rather than a failure. Excluding it removed three quarters of the newest year and made the year-over-year table look like a collapse that the raw data does not show. The basis here is therefore approvals, and the not-yet-disbursed count is published per year below rather than dropped. 969 of the 4,658 approvals in this window are pipeline.
• Lender field. ThirdPartyLender_Name and ThirdPartyDollars. SBA's data dictionary defines BankName and CDC_Name as the party a loan is "currently assigned to", so those fields credit whoever holds a loan today rather than whoever wrote it. ThirdPartyLender_Name carries no such qualifier.
• Name consolidation. One institution can appear under several recorded spellings. Names are grouped after upper-casing, expanding "National Association" to N.A., and stripping punctuation and legal suffixes, which merges "Bank of America, National Association", "BANK OF AMERICA N.A." and "Bank of America, N.A." into one row. That rule does not catch bare brand names or holding-company forms, so five are mapped explicitly: BMO → BMO BANK, FIFTH THIRD BANCORP → FIFTH THIRD BANK, MANUFACTURERS AND TRADERS TRUST → M T BANK, TD BANK US HOLDING → TD BANK, US BANK → U S BANK. A further four need the lender's city because the name alone is ambiguous: CITIZENS FINANCIAL GROUP in Providence → CITIZENS BANK, FIRST AMERICAN BANK CORPORATION in Elk Grove Village → FIRST AMERICAN BANK, FIRST CITIZENS BANCSHARES in Raleigh → FIRST CITIZENS BANK TRUST, NBT BANCORP in Norwich → NBT BANK. That distinction is not academic. An earlier version merged on the name alone and folded First Citizens Bancshares of Dyersburg, Tennessee into First-Citizens Bank & Trust of Raleigh, North Carolina, which are unrelated banks. Deliberately not merged: "Bank of America California, N.A." (a separate charter); First Citizens Bank of Luverne, Alabama and of Mason City, Iowa; IBC Bancorp of Chicago versus IBC of Laredo; and the three unrelated "Dime" banks. Truncated source strings such as "Zions Bank, A Division of" are relabelled for display only and merged with nothing.
• The CDC table uses GrossApproval, which in the 504 file is the SBA debenture (the second), not total project cost.
• Reproducing this. The script that derives every figure on this page from the raw CSV, filter and alias map included, ships with the site as data-sources/build_sba504_dataset.py. Running it against the same extract reproduces this page's dataset exactly.

The one limitation that matters most. The SBA file contains no property-type field. NaicsCode describes the borrower's industry, not what the loan bought, so a manufacturer's 504 loan is highly likely to be secured by its facility but the file never says so outright. SBA 504 is a fixed-asset program: the proceeds buy real estate or long-lived equipment, which makes an industrial-NAICS filter a sound proxy for industrial property lending. It remains a proxy, and it is stated here rather than buried.

Source data: U.S. Small Business Administration, public domain. This analysis is not endorsed by or affiliated with the SBA.