The USDA Rural Capital Atlas

Research status: October 17, 2026. Index scores are analytical judgments, not USDA ratings. FY2026 terms are historical; confirm the terms in force at obligation. Source conflicts, unverified grant caps and missing precedents are disclosed in the data notes.

Executive summary and key findings

USDA Rural Development can finance most rural commercial asset classes, but as of October 2026 the binding constraints are credit quality, agency capacity and program continuity, not eligibility. This Atlas profiles 38 asset classes across ten research streams and scores each on a five-factor USDA Financeability Index. Four asset classes score 20 or more out of 25: food and meat processing (22), gas stations and convenience stores (21), grocery and food retail (21) and rural healthcare (20). Five score 10 or less: manufactured housing communities (6), EV charging hubs (8), merchant renewable energy (9), market-rate multifamily (10) and data centers (10).

Ten findings that should change how lenders and sponsors approach USDA-supported deals

1. Business & Industry (B&I) is the lead instrument for 26 of the 38 asset classes, and its FY2026 terms are the most generous in recent history. Loans under $5 million carry an 85% guarantee; loans of $5 million to $25 million carry 80%. The upfront fee is 3.0% of the guaranteed amount, the annual retention fee is 0.55%, and terms run to 40 years with full amortization and no balloons (91 FR 11272, effective October 1, 2025; USDA stakeholder announcement, March 27, 2026). The five-point step at $5 million creates a sizing cliff: a $4.99 million loan is structurally superior to a $5.0 million loan.
2. The eligibility line is control plus business, not owner-occupancy. Part 5001 sets no occupancy threshold for B&I. A project fails only where the borrower is a passive landlord with no control over tenants and services (7 CFR 5001.115(g)), or where the use is speculative real estate (5001.115(a)). Self-storage has been eligible since the December 2021 OneRD rule removed its exclusion, yet no documented USDA self-storage guarantee was found for 2021 to 2026.
3. Housing routes through Section 538, not B&I. B&I excludes independent housing and residential trailer parks (5001.118(a)). Section 538 is the only active USDA debt product for new rental housing: $400 million program level in FY2026, a 90% guarantee, 40-year fixed terms and Ginnie Mae execution. In September 2026 USDA raised the Option 3 loan-to-cost to 80% and launched a 1.11x DSCR pilot for the first 200 loans. For nonprofit and public sponsors, Community Facilities Direct loans price at 4.500% to 4.750% fixed for up to 40 years (Q1 FY2027).
4. Grants are now a take-out, not a source. The REAP final rule (91 FR 62600, effective October 16, 2026) converts the only grant open to for-profit commercial sponsors into a post-completion, performance-validated rebate capped at 25% of energy-system cost. For hotels, fuel retail, car washes, storage and general retail, the realistic grant share of total development cost is 0% to 3%. Material grant layers survive only for meat processing (15% to 40% via MPPEP and VAPG) and for nonprofit or public community facilities (20% to 60%).
5. Credit risk is repricing. The B&I credit subsidy rate rose from 0.20% (FY2025) to 1.09% (FY2026) and 1.93% (FY2027 estimate), with a 9.02% default assumption and 19.57% recovery assumption in the FY2027 Federal Credit Supplement. On May 12, 2026 USDA removed ten lenders holding roughly $620 million, about 47% of Rural Development delinquency. B&I approvals fell from 89.1% of submissions in FY2021 to 52.7% in FY2023.
6. Losses concentrate in special-use and first-of-kind assets. Biodigester guarantees were 27% to 28% delinquent and controlled environment agriculture 40% to 43% delinquent when USDA paused both through December 31, 2026. B&I has excluded wind and solar since August 2025. The REAP guarantee recovery assumption is 8.73%, the lowest of any Rural Development line.
7. Scale beyond $25 million lives outside B&I. B&I caps at $25 million per borrower ($40 million for value-added agricultural cooperatives). Section 9003 biorefinery guarantees reach $250 million, and Rural Utilities Service electric lending carried $7 billion of FY2026 authority, including a $3.3 billion loan to Oglethorpe Power closed October 14, 2026.
8. Eligibility is abundant; demand is scarce. The B&I rural definition (places under 50,000) covers roughly 98% of U.S. land area and, on the last published estimate, about 36% of population. Nonmetro growth since 2020 is entirely migration-led and concentrates in metro-adjacent and recreation counties. Market depth, not location eligibility, is the binding geographic variable.
9. Execution risk now sits beside credit risk. Rural Development lost 1,745 of 4,910 employees (36%) between January and June 2025. The 2018 Farm Bill extension expired September 30, 2026. The FY2027 budget request proposes winding B&I commitments down to $85 million. A guarantee attaches to the fiscal year of obligation, so timing is a structuring decision.
10. The feasibility study has become the decisive filter. An independent study is mandatory for B&I loans above $1 million to a new business (7 CFR 5001.306) and for CF guaranteed loans above $1 million to a new entity or activity (5001.304). With approval rates near 50% and loss claims reduced where origination was negligent (Enterprise National Bank v. Vilsack), the study is both the approval document and the lender's guarantee-defense record.

How to use this Atlas. Credit officers should start with the Financeability Index heat map and the asset-class profile for the deal in front of them. Sponsors should read the capital stack playbooks and the feasibility requirements section before engaging a lender. Investors underwriting rural platforms should weight the credit forensics and geography sections, which explain where USDA leverage is cheap and where it is expensive.

Methodology and evidence base

The Atlas is built from ten independent research streams, each run to the same standard and the same template, and consolidated into a single scoring framework. Status is as of October 17, 2026.

Research architecture. Six streams profile asset classes (hospitality and leisure; roadside and fuel retail; storage, industrial and agribusiness; housing, senior living and community facilities; Main Street, services and acquisitions; frontier assets). Four streams are cross-cutting: capital stack rules and lender landscape, the grant layer, credit forensics, and eligible geography and market depth. The program compendium consolidates the program baseline that each stream verified independently against the Federal Register and the eCFR. Every asset class is profiled on an identical ten-part template: eligibility verdict with citation, lead and layerable instruments, structural terms, comparative advantage against SBA and conventional debt, underwriting sensitivities, an illustrative capital stack, documented deal evidence, risk flags, Financeability scoring inputs and a practitioner FAQ.

Evidence hierarchy. Every material claim is tagged by tier. [REG] denotes statute, the eCFR or the Federal Register. [USDA] denotes agency instructions, notices, award announcements, OIG and GAO work, and the Federal Credit Supplement. [STAT] denotes federal statistical agencies. [MKT] denotes lender, association and trade-press material, used only to evidence market practice. Where sources conflict, the Federal Register is treated as authoritative and the conflict is disclosed. Where evidence is absent, the Atlas says so rather than extrapolating.

What is primary in this research. Three components are original: the cross-asset Financeability Index, the reconciliation of program rules against asset-class economics at the level of specific CFR provisions, and the dataset design that combines USAspending.gov loan-level records with USDA's Lender Lens portfolio data (launched January 19, 2026) to produce the first lender-attributed, performance-linked view of rural commercial credit. The regulatory and program review is desk research and is presented as such.

The USDA Financeability Index. Each asset class is scored 1 to 5 on five inputs and summed to a composite out of 25.

A sixth input, loss severity given default, is scored separately in the credit forensics section on the eight asset groups where public loss evidence exists. Scores are analytical judgments grounded in the cited evidence; they are not USDA ratings.

Limits. No public OIG or GAO data isolate default rates for hospitality, fuel retail or storage. The most recent USDA-specific estimate of rural eligibility population share dates from the 2000 Census. Several demand datasets (truck parking, food access, childcare gaps) predate 2023. These gaps are flagged in the relevant sections and listed in the data notes.

Program compendium: the USDA rulebook as of 2026

Regulatory companion: USDA Feasibility Study Framework, the 37 factors of Part 5001, and USDA program guidance.

Fifteen USDA instruments are relevant to rural commercial finance, and for a for-profit sponsor only three are directly accessible as debt: the B&I guarantee, the REAP guaranteed loan and Section 538. Everything else is either restricted to public, nonprofit, cooperative or producer sponsors, or has been converted into a post-completion rebate.

Equity and collateral rules common to the OneRD guarantee programs (7 CFR 5001). Existing businesses need 10% balance-sheet equity (or debt-to-equity of 9:1 or better). New businesses need 20% balance-sheet equity or 25% of project cost, rising to 25% balance-sheet equity where the guarantee is issued before construction is complete (5001.105(d)). Grants and subordinated debt under a life-of-loan standstill count toward equity; appraisal surplus and bargain-purchase gains do not. Discounted collateral must at least equal the loan amount, and any going-concern or business-enterprise value in an appraisal must be deducted before discounting (5001.203). Personal guarantees are normally required from owners of 20% or more (5001.204).

Ineligible purposes register (7 CFR 5001.115, 5001.118, 5001.127). Lines of credit and floorplan; owner-occupied housing and independent rental housing; residential trailer parks and timeshares; golf courses and golf infrastructure; racetracks; for-profit zoos and safaris; publicly owned or nonprofit amusement and water parks; speculative real estate investment; commercial rental where the borrower has no control over tenants and services; projects using technology that is not commercially available; businesses deriving more than 15% of gross revenue from gambling, including machine lease income; lending, investment and insurance companies; wind and solar generation under B&I (since August 19, 2025); and any project that would violate the environmental requirements of 7 CFR Part 1970.

What changed in 2025 and 2026

Eligible geography and rural market depth

Location eligibility is near-universal outside metropolitan cores, so geography enters the underwriting through market depth rather than through the eligibility map. The business-program rural definition captured about 98% of U.S. land area and roughly 36% of population on the last USDA estimate, and nonmetro counties (1,958 of 3,144) hold 46.2 million people on 72% of the land.

Three definitions govern commercial credit.

Sites are geocoded against the Rural Development eligibility map for the relevant program. Eligibility is a gate, not a score: rewarding it would reward remoteness.

Rural fundamentals, 2020 to 2026. The nonmetro population has grown every year since 2020, but only through migration. Between 2020 and 2024 net migration added 974,379 people to nonmetro counties while natural change subtracted 563,550, and 76% of nonmetro counties recorded natural decrease in 2023 to 2024. Growth concentrates in metro-adjacent and recreation counties. Nonmetro median household income was $60,459 in 2023 against $81,270 in metro areas, a gap of about $20,800 that has been stable since 2021. In 2023, 21% of the nonmetro population was 65 or older against 17% in metro areas, and health care and social assistance is the largest nonmetro employer at 14.1% of employment. Rural housing is tight: rental vacancy outside metropolitan areas was 5.8% in the second quarter of 2026, below suburbs (6.9%) and principal cities (8.0%).

What the fundamentals mean by asset class. The age profile favors senior living and healthcare and constrains discretionary retail. Migration-led growth favors workforce housing, grocery and fuel retail in adjacent counties. Recreation counties support hospitality and outdoor assets. Persistent natural decrease in remote counties limits the depth of any asset class that depends on local population rather than through-traffic or visitors.

Demand corridor indicators used in the Atlas

USDA activity geography. Guaranteed volume is rising while delivery capacity falls. B&I program level rose from $1.829 billion (FY2024) to $2.112 billion (FY2025), and Section 502 single-family guarantees from $6.130 billion to $8.351 billion. Volume concentrates in the South: Louisiana, Texas, Tennessee, Missouri and Alabama lead single-family guarantees, and Texas, Louisiana and Florida lead B&I. The FY2027 Explanatory Notes project permanent field positions falling to 2,879 in FY2026.

Provisional state tiers (FY2025 USDA volume, nonmetro fundamentals and documented complementary programs; per-capita intensity could not be computed from sourced data)

Rural Market Depth score (the index input). Each county in the eligible nonmetro universe is scored on a core demand index (50% weight: three-year net migration 30%, employment growth 20%, cost-adjusted median income 15%, metro adjacency 15%, inverse vacancy 10%, inverse household debt-to-income 10%), an asset-class demand module (35%) and an execution overlay (15%: Rural Development obligations per nonmetro resident 40%, state office staffing change 30%, state complementary programs 30%). Indicators are standardized as percentiles within the 1,958 nonmetro counties, not nationally, so rural counties are compared with rural peers. The composite maps to quintiles: 5 for the top 20%, 1 for the bottom 20%. Two overrides apply: any county with negative three-year net migration and natural decrease is capped at 3, and any county whose asset-specific driver ranks in the bottom quintile is capped at 2. Counties whose driving dataset predates 2023 are flagged, not penalized.

USDA Financeability Index: cross-asset heat map

Food and meat processing is the most financeable rural asset class under USDA programs in 2026, scoring 22 of 25; manufactured housing communities are the least, at 6. Composite scores of 18 or more mark core USDA asset classes with clear eligibility, active lenders and leverage uplift. Scores of 14 to 17 mark workable classes where structure, sizing or precedent requires attention. Scores of 13 or less mark classes that are structure-dependent, paused or effectively closed.

Reading the map. Three patterns stand out. First, the classes that score highest combine a named eligible purpose in the regulation, policy priority and documented deals: processing, fuel retail and grocery each have all three. Second, grant-layering is the weakest input almost everywhere; only processing, healthcare, childcare and cooperative energy score above 2, which is the direct consequence of the October 2026 REAP rule. Third, the bottom of the map is defined by rule, not by market: manufactured housing, merchant renewables and EV charging are excluded or starved of program support regardless of demand.

Loss severity, the sixth input. Public loss evidence supports a separate severity score for eight asset groups, reported in the credit forensics section: energy and biofuels 5, controlled environment agriculture 5, food and meat processing 4, healthcare 4, hospitality 3, retail and fuel 3, manufacturing 3, multifamily 2. Read against the composite, processing is the clearest case of high financeability paired with high severity: it is the asset class where feasibility quality matters most.

Atlas: hospitality and leisure

Hospitality is B&I's largest sector, taking 32% to 37% of core obligations in each measured year and nearly $4 billion over FY2015 to FY2024, and hotels and RV parks are the only two asset classes in this group named in the regulation. 7 CFR 5001.105(b)(8) makes eligible tourist and recreation facilities, including hotels, motels, bed and breakfasts, and resort trailer parks and campgrounds operated as commercial enterprises. Glamping, marinas, event venues and restaurants rely on the general commercial-enterprise purposes and must clear three exclusions: timeshares and independent housing (5001.118(a)), golf courses, racetracks and public or nonprofit recreation (5001.115(b)), and principal places of worship (5001.115).

What credit committees test on a hospitality B&I loan. The regulation sets structure rather than ratios. It imposes no minimum DSCR on new money; a 1.1x historical DSCR applies only where refinancing is the majority purpose (5001.102(d)). What it does impose: balance-sheet equity tiers of 10%, 20% or 25% (5001.105(d)); quarterly projected cash flow for seasonal borrowers (5001.202(b)(6)(v)); removal of going-concern value from the appraisal before discounting (5001.203), which hits hotels hardest because hotel appraisals routinely carry business enterprise value; an independent feasibility study above $1 million for a new business (5001.306); and full amortization with interest-only capped at three years (5001.402). Lenders in practice size at about 1.25x DSCR and 75% to 80% loan-to-value. Documented USDA hotel deals show loan-to-cost of 74% to 78%.

Hotels and motels, flagged and independent

Hotel feasibility studies

Verdict: Eligible, High confidence. Named in 5001.105(b)(8); the regulation does not distinguish flagged from independent. Condo-hotel and timeshare structures fall under the 5001.118(a) exclusion. Composite 18.

Structure. B&I up to $25 million; 85% guarantee under $5 million, 80% above. Term to 40 years, with market practice at 25 to 30 years. Working capital, franchise fees and pre-opening costs are eligible uses (5001.105(b)(4)). New Markets Tax Credits are expressly compatible (5001.141). Against SBA 7(a), B&I offers five times the loan size and permits passive and institutional ownership; against 504, B&I finances working capital and PIP, while 504 requires 15% to 20% equity on special-purpose property.

What the study must evidence. Occupancy, ADR and RevPAR against a defined competitive set; supply pipeline; demand segmentation; seasonality; franchise fee load; PIP cost and timing; and management track record. The lender must review the franchise and management agreements (5001.202(a)).

Deal evidence. Hotel Ava LLC, Paso Robles, CA: $25,000,000 for a 115-room hotel. Hallmark Inns & Resorts, Newport, OR: $25,000,000 across two notes for refinance and a 26-room addition. Chelsea Hospitality Group, MI: $4,900,000 to refinance and renovate for a flag change to Hampton Inn, a documented PIP use. Best Hospitality Group, Okemos, MI: $6,810,000 acquisition at 74% loan-to-cost. Inn at Bellefield, Hyde Park, NY: B&I covered nearly half of a project above $50 million.

Risk flags. Condo-hotel or timeshare structures; an attached golf course; owner quarters not deducted pro rata; tax-exempt financing on the same component; projects above $1 million that relocate jobs and add more than 50 employees, which trigger Department of Labor review; metro-fringe sites where rural status must be confirmed on the eligibility map.

Sizing note. An acquisition structured at $4.9 million rather than $5.1 million moves the guarantee from 80% to 85% and cuts the lender's unguaranteed exposure from $1.02 million to $0.735 million.

RV parks and campgrounds

RV park feasibility studies

Verdict: Eligible, High confidence, for parks operated as commercial enterprises; residential trailer parks are ineligible (5001.118(a)). The dividing line is transient versus residential occupancy. No numeric threshold exists in regulation; the 50% revenue test consultants apply is borrowed from the mixed-use rule. Composite 18.

Structure. Most acquisitions fall in the sub-$5 million, 85% band. Interest-only periods to three years and annual payment frequency (5001.402) allow seasonal matching. Quarterly projected cash flow is mandatory (5001.202(b)(6)(v)). The guarantee offsets the heavy discount conventional lenders apply to seasonal, single-purpose land.

What the study must evidence. Site mix (transient, seasonal, monthly) and length of stay; rate surveys of competing parks; demand anchors; shoulder-season fill; utility capacity per pad; and the residential-tenant share, to defend against the exclusion.

Deal evidence. Camp Table Rock LLC, Branson, MO: $1,400,000 to acquire a 57-site park with cabins, boat storage and restaurant. Ogden Industries, Archie, MO: $865,000 to refinance a 50-site park and add cottages and solar. REAP grants predating the 2026 rule: Settlin Inn, AZ, $250,000 for solar shade canopies; Cedar Valley Management, AR, $49,192.

Risk flags. Majority long-term tenancy; membership campgrounds; flood-zone sites requiring NEPA mitigation; single-season markets with under six months of operation, where the DSCR rarely closes.

Glamping, cabin resorts and clustered short-term rentals

Glamping feasibility studies

Verdict: Conditional, Medium confidence. The regulation is silent on glamping; operator-run transient lodging is the defensible framing under 5001.105(b)(8). Individual unit sales or rental pools approach the timeshare and independent-housing exclusion. Composite 13.

Structure and evidence. Ground-up resorts are new businesses: 20% balance-sheet equity or 25% of cost, and a feasibility study above $1 million. Soft-sided structures carry useful-life limits that cap the term. The study must show comparable ADR and occupancy from booking platforms, OTA channel dependence, seasonal closure risk, replacement capex for soft goods, and septic, water and zoning approvals. The strongest precedent is Pinecrest Smoky Mountain Holdings, Sevierville, TN: $7,850,000 to convert 35 long-term residential units into short-term vacation units. No glamping-specific B&I deal was found in primary sources.

Risk flags. Fractional or condo sales; owner residence without deduction; local short-term-rental bans or permit caps; structure life shorter than the loan term.

Marinas

Agritourism and recreation feasibility

Verdict: Eligible for for-profit commercial marinas, Medium confidence, under the general commercial-enterprise purpose. Public or nonprofit recreational marinas are at risk under 5001.115(b). Composite 13.

Structure and evidence. Docks, lifts and dry-stack buildings have distinct useful lives that cap the term. Submerged-land leases must run at least as long as the loan with no reverter clauses (5001.105(b)(14)). The study must show slip and dry-storage occupancy and waitlists, the seasonal versus transient revenue split, fuel margin and underground storage tank liability, dredging and storm capex, and boat-registration trends. Precedent: Eastgate Harbor LLC, Branson, MO, $4,165,000 to acquire Branson Bay Marina and build dry storage.

Risk flags. Public ownership; unresolved fuel contamination; short submerged-land lease tenor; hurricane exposure and insurance cost; dockominium slip sales.

Event and wedding venues

Wedding venue feasibility

Verdict: Conditional, Medium confidence. A for-profit venue fits 5001.105(b)(1), but chapels used as a principal place of worship are ineligible, and the agency's treatment of commercial wedding chapels is undocumented. Composite 11.

Structure and evidence. Most venues are new businesses and carry the higher equity tier and mandatory study. The study must show booked backlog and deposits, weekend versus midweek utilization, average event revenue, regional wedding counts, catering model, alcohol licensing and septic and parking capacity. No standalone venue B&I deal was found; adjacent precedents are the Sagebrush hotel in Taos, NM, with a 500-seat venue, and Commongrounds Cooperative, Traverse City, MI, an $8 million mixed-use B&I with a performing-arts venue. Pairing a venue with lodging materially strengthens the eligibility narrative.

Full-service restaurants

Verdict: Eligible, High confidence, as a commercial enterprise under 5001.105(b)(1) to (4). Composite 14.

Structure and evidence. Equipment terms are limited to useful life; leasehold improvements require a lease at least as long as the loan. The study must benchmark food and labor cost ratios against industry standards, which 5001.202(b)(6)(iii) requires, and stress DSCR on labor and food-cost inflation. Goodwill can be acquired but intangibles cannot serve as primary collateral. Precedents: Hawaii Sea Spirits, Maui, $10,000,000 covering distillery, retail and café expansion; Daddio's Soda Shop, Lake of the Ozarks, MO, $410,000. Under $5 million with an owner-operator, SBA 7(a) is usually the simpler route; B&I suits larger or passive-investor deals.

Hospitality FAQ. USDA does not require franchise approval by regulation; the lender must review the agreements. PIP costs are financeable as renovation under 5001.105(b)(1). Refinancing is eligible where the loan extinguishes the debt and, if refinancing is the majority purpose, historical DSCR is at least 1.1x; refinancing the applicant lender's own debt is capped at 50% of uses. The guarantee can attach during construction for a 0.50% fee with a fixed-price contract, bonding and monthly reporting (5001.205(e)(2)). Personal guarantees apply to owners of 20% or more but not to passive investors (5001.204).

Atlas: roadside and fuel retail

Five of these six asset classes are eligible rural businesses under 7 CFR 5001.126(d)(2), and the binding constraints are sizing and collateral rather than eligibility. Three disqualifiers matter: gambling revenue above 15% of gross, including machine lease income (5001.127(b)); projects that cannot clear the environmental review under 7 CFR Part 1970 (5001.115(i)); and passive commercial rental where the borrower has no control over tenants and services (5001.115(g)). The collateral rule at 5001.203, which strips going-concern value from the appraisal before discounting, bears hardest on single-purpose fuel and car wash assets, where most of the value sits in the business enterprise. Real-estate-plus-equipment coverage will typically fall short of the loan amount after discounting, so the feasibility study must evidence repayment capacity with enough margin to justify the shortfall, and personal guarantees and additional collateral belong on the term sheet.

Environmental diligence is a gating item. The B&I and REAP lender checklists require a Phase I Environmental Site Assessment and Part 1970 information, and Rural Development warns that starting development before the environmental process completes may jeopardize the guarantee. State underground storage tank fund status is a credit variable: 36 states maintain financial assurance funds, six have funds that no longer provide financial responsibility, and owners of tanks 30 years or older report difficulty renewing private pollution insurance.

Gas stations and convenience stores

Gas station feasibility studies

Verdict: Eligible, High confidence. Composite 21, the highest in the roadside group, driven by market depth (5) and lender appetite (4).

Structure. B&I for land, building, canopy, tank systems, equipment, acquisition and permanent working capital; 85% under $5 million. A REAP guaranteed loan can layer for LED, refrigeration and HVAC; REAP grants now apply only to existing sites after 12 months of operating data. Acquisitions including goodwill are eligible, but intangibles cannot be primary collateral.

What the study must evidence. Fuel gallons with cents-per-gallon margin separated from volume; inside sales and gross margin by category; the fuel supply agreement, including term, volume commitments and incentive repayment on debranding; tank age, compliance testing, state fund eligibility and deductible; competitive supply and traffic counts, including planned realignments.

Deal evidence. Merced River Development, Snelling, CA: $2,358,797 to refinance and improve a station and store. REAP precedents at fueling sites: Mission Construction, St. Paul, KS, $97,500 for a 64.99 kW solar system; Millston Kwik Stop, Rockland, WI, $31,704.

Risk flags. Unresolved releases; single-walled or aging tanks without a financial responsibility mechanism; unamortized branding incentives payable on change of control; video gaming revenue approaching 15%; absentee investor structures.

Truck stops and travel centers

Truck stop financing research

Verdict: Conditional on the gaming test, High confidence. Scale typically pushes loans into the $5 million to $25 million tier at 80%. Composite 16, with the highest leverage uplift in the group (5).

Structure and evidence. The standard $25 million ceiling applies; the cooperative exception does not. New-business construction with a pre-completion guarantee requires 25% equity. The study must evidence diesel gallons by fleet versus independent mix and fleet card discount exposure, truck parking demand and supply, restaurant franchise terms, shop and tire revenue, interstate access, competing chains within roughly 60 miles of corridor and gaming revenue tested against the 15% line. Ordinary lottery commissions are not clearly excluded from that test; a written State Office determination should be obtained. Precedent: Ceres Enterprises, CA, $9,736,000 to build a fuel stop with truck diesel, store and restaurant.

Risk flags. Gaming above 15% including leased-machine income; fuel-only interchanges; multi-parcel structures with uncontrolled tenants; diesel demand concentrated in one carrier contract.

Express tunnel car washes

Verdict: Eligible, High confidence, as a service business; not a recreation use under 5001.115(b). Composite 16.

Structure. B&I for land, tunnel, equipment and working capital, with the 85% band strongly favoring loans under $5 million. Lender practice sets stabilized DSCR floors at about 1.25x to 1.35x. Membership value is not collateral; going-concern value is deducted. Under SBA, car washes are special-purpose and carry 15% to 20% 504 equity, which is where B&I's balance-sheet equity test is most useful.

What the study must evidence. Membership penetration, monthly churn and revenue per member; conversion from single washes; the ramp curve to stabilization; competing tunnels in the drive-time trade area; weather-adjusted retail volume; equipment replacement reserves. The 2025 to 2026 industry center for monthly churn is about 7% to 8%; stress at 10% or more.

Deal evidence. Kalkat Inc., Wheatland, CA: $750,000 for real estate and equipment. OWC Holdings (Hollister Wash Co.), Hollister, MO: $1,768,500.

Risk flags. Churn above 8% to 10%; a membership book built on promotional pricing; new builds in oversupplied trade areas; equipment-heavy collateral with steep liquidation discounts; unsecured water rights or discharge permits.

EV charging hubs

Verdict: Conditional, Medium confidence. Composite 8. REAP expressly excludes chargers, charging stations and any system that dispenses energy at retail. B&I permits charging in principle, but no documented USDA-financed charging transaction exists.

Structure and evidence. Underwrite rural charging as ancillary capex inside a fuel, QSR or travel-center credit, financed with B&I and sponsor equity, with the halo effect on inside sales as the core thesis. Federal support is contracting: NEVI obligations reached $1.4 billion of $4.2 billion by December 31, 2025; $503.8 million was transferred to other highway uses in February 2026; CFI has no new NOFO; the FY2027 request proposes cancelling $4.2 billion; and a proposed 100% Buy America content rule is one no known charger meets. Treat any NEVI award as upside, never as a source required to close. The study must evidence utilization per port, demand charges and tariff structure, make-ready cost, uptime obligations and clawbacks.

Quick-service and franchise restaurants

Verdict: Eligible, Medium-High confidence. Part 5001 contains no franchise-specific provision, and no documented B&I loan to a named QSR franchisee was identified, although USDA has financed franchised lodging (Mayur Hospitality, Milledgeville, GA, $5,322,000) and independent restaurants (T's Redneck Steakhouse, Lebanon, MO, $1,255,000). Composite 17.

Structure and evidence. B&I for real estate, building, FF&E, franchise fee and working capital. The capitalized franchise fee is an eligible start-up cost but an intangible that cannot be primary collateral. Ground leases must run at least the loan term with no reverter clauses (5001.105(b)(14)). The study must evidence FDD Item 19 unit economics, brand average unit volume in comparable rural trade areas, royalty and advertising load, remaining franchise term versus loan term, transfer and cross-default provisions, and franchisor rights of first refusal that affect foreclosure. Franchisor estoppels are lender practice, not regulation. B&I beats 7(a) above $5 million, for multi-unit packages, or where the 85% guarantee materially improves pricing.

Auto service and repair

Verdict: Eligible, High confidence, with precedent. Composite 17.

Structure and evidence. B&I for acquisition, real estate, lifts, diagnostic equipment and refinancing; acquisitions qualify where jobs are created or saved. Lenders treat auto repair as environmentally sensitive, so a Phase I is required and the study must evidence waste oil, solvent and in-ground hydraulic history alongside technician capacity, bay utilization, labor-hour rates and insurer concentration for collision shops. Precedent: OHS Auto Body and CVR Holdings, MT, $4,397,537 to refinance a high-interest SBA loan across four shops with 43 employees, saving about $8,000 a month.

Risk flags. Historical tanks without closure documentation; floor drains and dry wells; key-technician dependency; insurer program concentration.

Roadside FAQ. A $5.2 million car wash or store loan is usually better resized to $4.9 million, with the gap filled by equity or a REAP guaranteed loan, to capture the extra five points of guarantee. REAP can still fund LED, refrigeration, HVAC and on-site solar at existing stores, washes and restaurants, but it cannot fund chargers, flex-fuel pumps or new-build efficiency, and it should never appear as a construction-period source. A guarantee percentage attaches to the fiscal year of obligation, not of application.

Atlas: storage, industrial and agribusiness

Processing and manufacturing are the strongest USDA credits in this group, storage is eligible but unproven, and energy is the most constrained asset class in the entire Atlas. The pivotal question for storage and leased industrial product is whether the borrower runs an operating business or holds passive real estate, and the regulation answers it with control and jobs, not occupancy.

The active-versus-passive line. Part 5001 sets no owner-occupancy threshold for B&I; the 25% leased-space cap applies only to Community Facilities. Five provisions do the work. The borrower must be engaged in a business that provides employment (5001.126(d)(2)) and the project must primarily create or save rural jobs (5001.126(d)(4)). Speculative real estate investment is barred (5001.115(a)). Commercial rental is barred only where the borrower has no control over tenants and services (5001.115(g)). Constructing or equipping facilities for lease to commercial or industrial businesses is an express eligible purpose (5001.105(b)(12)), and mixed-use property qualifies where at least 50% of projected revenue comes from business use (5001.105(b)(13)). Lending, investment and insurance companies are ineligible borrowers (5001.127(g)).

OpCo/PropCo and master-lease structures are handled through co-borrowing: where affiliates are effectively one business or rely on each other for repayment, they must be co-borrowers unless the agency waives it in writing (5001.126(a)(2)). A third-party manager is permitted; the borrower remains responsible and must keep ownership and control (5001.126(a)(1)).

The February 2026 underwriting reset. The RBCS Administrator's letter of February 18, 2026 told lenders that guarantors should be the norm, offtake agreements should align with the loan term, reserves should be borrower-funded, and interest-only periods must be justified with monthly cash-flow projections, with principal and interest beginning at the break-even month. Feasibility studies in this group now carry the ramp-up and offtake burden directly.

Self-storage

Self-storage feasibility studies

Verdict: Conditional, Medium-High confidence: eligible as an operator-controlled business since the December 10, 2021 OneRD rule (86 FR 70349) removed the exclusion, but with no documented USDA guarantee to a self-storage facility for 2021 to 2026. Some lender pages still quote the superseded exclusion. Composite 15.

Structure. Standard B&I terms; 25% equity on a ground-up build where the guarantee is issued before completion, 10% for an existing operator. Grant layering is negligible. The real estate and rental sector's share of B&I dollars fell from 8.25% in FY2019 to 3.28% in FY2024, so expect slower state-office review and close scrutiny of job creation.

What the study must evidence. Supply per capita in a three- to five-mile trade area, where a single new competitor can flip the market; the lease-up curve and break-even month; rate seasonality; and documentation of operator control over rent-setting, tenant selection and services.

Risk flags. Absentee, fully outsourced operations with no evidence of control; build-to-flip plans; low job counts in competitive funding rounds; lender checklists that still quote the pre-2021 rule.

RV and boat storage

Verdict: Conditional, Medium confidence: eligible by extension, since no storage-specific bar remained after 2021, and the least proven asset in the group. Composite 12.

Structure and evidence. The collateral test is the binding constraint. The regulation requires discounted collateral at least equal to the loan amount, and open-lot product with low improvement value will cap loan size. Covered or enclosed storage with on-site services (dump stations, wash bays, launch concierge) strengthens the operating-business case; a campground component brings 5001.105(b)(8) into play, while residential or long-term RV dwelling triggers 5001.118(a). A pure ground lease to an unrelated operator risks 5001.115(g). The study must evidence drive-time demand from lake and recreation corridors and from HOA communities that ban on-site RV parking, covered versus open pricing spreads and seasonal occupancy.

Warehousing, distribution and cold storage

Cold storage financing research

Verdict: Eligible, High confidence, for owner-operated, 3PL and cold-chain operators; Conditional, Medium, for multi-tenant buildings leased to third parties. Composite 17.

Structure. The main structural advantage of contracted cold storage is equity: a new business with creditworthy sales contracts at least as long as the loan needs only 10% equity (5001.105(d)(2)) against 20% to 25% otherwise. REAP guaranteed loans layer for refrigeration efficiency. C-PACE can sit alongside only with the senior-lien adjustments the February 2026 letter requires.

What the study must evidence. Contracted versus spot pallet positions; customer concentration; contract tenor against loan term; refrigeration engineering and energy cost; product mix; and, for leased buildings, the borrower's control over tenant selection and services.

Deal evidence. FlexCold, FL: $8,183,000 Food Supply Chain guarantee for mobile racking, 35 jobs. Food Group II, Goshen, NY: cold storage adjacent to processing, with C-PACE equity. Bottomland Prime, Amarillo, TX: $4,950,000. These are Food Supply Chain Guaranteed Loan precedents; that program has no visible 2025 to 2026 window, and B&I is now the instrument for the same projects.

Light manufacturing

Industrial feasibility studies

Verdict: Eligible, High confidence. Section 5001.126(d)(2) names manufacturing first among eligible activities, and land, buildings, machinery and permanent working capital are all eligible uses. Composite 18.

Structure. 10% equity for existing manufacturers, with reductions possible where guarantees are obtained and ratios meet the RMA median quartile (5001.105(d)(5)(ii)). Equipment terms run to useful life, about 15 years; working capital about seven. NMTC (5001.141) and ESOP or worker-cooperative conversion financing layer. Priority points reward five or more jobs above 150% of the federal minimum wage, employer health plans paying at least half of premiums, and international sales of 20% or more (5001.318(d)).

Deal evidence. Shenandoah Valley Organic, Harrisonburg, VA: $25,000,000 B&I to expand processing capacity. Jackson Purchase Distillery, Fulton, KY: multiple B&I loans in 2023 and 2024. Secure Semiconductor Manufacturing, Neodesha, KS: B&I-financed equipment. Lone Star Bakery, China Grove, TX: $40,000,000 Food Supply Chain guarantee. Ranger Scientific, Montgomery, WV: $7.5 million to convert a former high school into an ammunition plant.

Risk flags. Projects over $1 million adding more than 50 jobs are screened for job transfer from another area and for market overcapacity (5001.118(b)); process technology must be commercially available (5001.115(h)); related-party construction mark-ups (5001.122(a)).

Food and meat processing

Food processing feasibility studies

Verdict: Eligible, High confidence, and the highest-conviction asset class in the Atlas, backed by policy, precedent and the only material grant layer open to for-profits. Composite 22.

Structure. B&I is the lead, with cooperatives doing value-added processing able to borrow up to $40 million (5001.105(b)(18)). Integrated agricultural production linked to processing, such as poultry houses feeding a plant, is eligible up to the lesser of 50% of the loan or $5 million (5001.105(b)(7)). Grants count toward equity. Layers: MPPEP Phase 4 ($60 million, $50,000 to $2 million per project at 50% cost share, closed August 7, 2026, limited to facilities primarily processing cattle); VAPG working capital up to $200,000; the SPUR program of up to $500 million announced August 31, 2026, with a guaranteed loan component pending terms. USDA's October 2025 beef plan directs RBCS to prioritize beef processor applications for B&I.

What the study must evidence. Livestock or commodity supply agreements (head per week, producer commitments, tenor); offtake agreements aligned to loan term; the FSIS or state inspection pathway; rendering and by-product outlets; wastewater capacity; labor availability; and a monthly ramp-up curve to the break-even month.

Deal evidence. Shenandoah Valley Organic, VA: $25,000,000. Yonah Valley Farms, GA: $10,730,000 with Stone Bank to acquire and automate an idle plant. FPL Food, GA: $24,187,066 for high-pressure processing equipment. Centrillium Protein, OK: $15,100,000. Pure Prairie Farms, Charles City, IA: $38,720,000 for a start-up chicken plant, now the reference failure for ramp-up risk (see credit forensics).

Risk flags. Supply risk in a contracting beef herd; start-up ramp-up; episodic grant windows and a FY2027 request that reportedly eliminates VAPG appropriations; agricultural production above the 50% or $5 million cap.

Renewable and agri-energy

Renewable energy feasibility studies

Verdict: Ineligible for wind and solar under B&I (policy announcement of August 18, 2025); Conditional for REAP guaranteed loans; unavailable for biodigesters through December 31, 2026. Composite 9.

Structure and evidence. The REAP guaranteed loan is the only remaining route for facility-sized systems: 80% guarantee, 1.0% fee, 0.25% retention, up to 75% of project cost, with at least 25% matching funds. The borrower must own the project, control the site and control revenues. The October 2026 REAP rule bars ground-mount solar and wind on cropland and foreign-adversary components, and under the 2025 tax law any wind or solar not under construction by July 4, 2026 must be placed in service by December 31, 2027 to retain the 45Y/48E credits. The study must evidence historical energy usage (right-sizing is required), site control, component origin and the placed-in-service schedule; for digesters, feedstock contracts, RNG offtake and environmental-attribute pricing. USDA paused digester and controlled environment agriculture guarantees citing 27% to 28% and 40% to 43% delinquency respectively.

Storage and industrial FAQ. Self-storage is no longer ineligible; pages saying otherwise quote the superseded rule. A landlord can finance a multi-tenant industrial building under 5001.105(b)(12) if it controls tenants and services. A new cold-storage operator with creditworthy contracts at least as long as the loan needs 10% equity. Company-owned poultry houses can be financed with the plant up to the lesser of 50% of the loan or $5 million. VAPG pays for planning and working capital only, never equipment. MPPEP Phase 4 is closed; SPUR terms are not yet published.

Atlas: housing, senior living and community facilities

The asset's primary use and the sponsor's tax status decide the program; lender preference does not. B&I excludes apartments and other residential housing where the primary purpose is independent housing, and residential trailer parks (7 CFR 5001.118(a)), with only three narrow exceptions: mixed-use property with at least 50% of projected revenue from business use (5001.105(b)(13)), owner-occupied lodging with living quarters deducted pro rata (5001.115(n)), and combined assisted and independent living campuses where medical services are optional for independent residents or most residents need assisted care (5001.105(b)(20)). Section 538 is the only active USDA debt product for new rental housing. Community Facilities lends only to public, nonprofit and tribal borrowers.

Section 538 in 2026. The program level is $400 million, handled first-come, first-served until funds deplete. The value is a 90% guarantee, a 25- to 40-year fixed term and Ginnie Mae execution; no interest credit is available. Maximum loan is 90% of the lesser of total development cost or value for for-profits and 97% for nonprofits, and Option 3 loan-to-cost rose from 70% to 80% on September 25, 2026 (91 FR 60930). Household income at initial occupancy must be at or below 115% of AMI, unit rents at or below 30% of 115% AMI, and the project average at or below 30% of 100% AMI. Fees reset September 30, 2026 to 0.65% upfront and 0.35% annual, with 0.60% and 0.25% for workforce (80% to 115% AMI), preservation and Opportunity Zone deals. The pilot running October 9, 2026 to September 25, 2028 gives the first 200 loans a 1.11x DSCR, caps developer fee at 15% of cost where no LIHTC funds it, and accepts HUD environmental reviews and HFA-approved capital needs assessments. Lenders approved by Fannie Mae, Freddie Mac, Ginnie Mae, HUD, an FHLB or an HFA are automatically eligible. Refinancing is ineligible except for existing guaranteed loans (3565.206(g)). Because Section 538 sits under Part 3565, the OneRD tax-exempt bar does not apply, and the program layers with 4% LIHTC and bonds, 9% LIHTC, HOME and hard second mortgages.

Community Facilities in 2026. CF Direct carries $1.25 billion of authority at fixed rates of 4.500% (poverty), 4.625% (intermediate) and 4.750% (market) for October to December 2026, for up to 40 years with no prepayment penalty, in communities of 20,000 or fewer. CF Guaranteed carries $650 million of authority at an 80% guarantee, 1.25% upfront and 0.50% annual, in communities of 50,000 or fewer, and is heavily underused: about $89 million obligated in FY2025. CF grants cover 15% to 75% of eligible cost by population and income tier, but the competitive pool is $4 million nationally against $659.2 million of congressionally directed spending; applications are considered loan-first. Do not rely on CF grants.

Decision tree. If the primary use is independent residential rental and rents fit the 538 caps, an existing Section 515 property goes to a 515 transfer or MPR with a 538 overlay under the 2026 pilot, and new construction or acquisition-rehab goes to 538 with LIHTC or HOME; outside the caps there is no USDA route. If the use is licensed care or an essential facility and the sponsor is public, nonprofit or tribal, a community of 20,000 or fewer goes to CF Direct with a CF grant where the tier qualifies, and a community of 20,001 to 50,000 or a sponsor wanting bank execution goes to CF Guaranteed. A for-profit sponsor in a community of 50,000 or fewer goes to B&I, with assisted living required to predominate on senior campuses. Mixed use goes to B&I only where business revenue is at least 50%. Manufactured housing communities and employer-only housing have no USDA route.

Multifamily, affordable and market-rate

Multifamily feasibility studies

Verdict: Affordable: Eligible, High confidence, composite 18. Market-rate: Conditional, High confidence, composite 10; it works only if the deal accepts the 538 rent and income caps.

Structure. Section 538 Option 1 or Option 3, layered with LIHTC, HOME, assumed Section 515 debt with Section 521 rental assistance and the transfer pilot. Leverage is comparable to Fannie Mae's 90% LTV executions but with longer fixed terms, better rural reach and the ability to carry hard seconds; it is about 15 to 25 points above bank debt. Borrower operating capital of at least 2% of the loan is expected. Lenders underwrite to 93% occupancy, or 95% with rental assistance.

What the study must evidence. Unit-by-unit compliance with the rent caps under current USDA income limits; comparable rents at or below the caps; absorption to 90% occupancy for 90 days; capture and penetration rates by AMI band; rural eligibility under 42 U.S.C. 1490; a capital needs assessment.

Illustrative stack, 60 units with 4% LIHTC, $15.0 million cost. Section 538 Option 3 loan $6.0 million (40%); LIHTC equity $6.6 million (44%); HOME or state soft debt $1.5 million (10%); deferred developer fee $0.9 million (6%).

Deal evidence. Greystone, rural North Carolina, February 2026: $52,552,000 of Section 538 financing across ten properties and 640 units, with $28.4 million of LIHTC equity and assumed Section 515 loans. Churchill Stateside Group, Castle Hayne, NC: $1,432,200 Option 1 permanent loan for 68 senior units with 9% LIHTC. CSG, Inverness, FL, October 15, 2026: $4.4 million forward commitment for 96 senior units to be securitized through Ginnie Mae.

Risk flags. Rents above the caps; sites associated with urban areas; construction-only financing; refinancing of non-guaranteed debt; the 200-loan pilot cap.

Workforce housing

Verdict: Conditional under Section 538, Medium-High confidence, with a dedicated fee tier for rents between 80% and 115% of AMI (0.60% and 0.25%); Ineligible under B&I for employer-sponsored housing, High confidence. Composite 16. No closed 538 deal publicly labeled as workforce housing was found. Units restricted to one employer's staff raise unresolved fair-housing and public-occupancy questions on which USDA has issued no guidance; employer participation is better structured as equity, a master lease, land or TIF than as occupancy restriction.

Manufactured housing communities

Verdict: Ineligible under B&I, High confidence, as residential trailer parks; Conditional under Sections 538 and 515, Low confidence, since GAO reports both programs are eligible but have not been used for manufactured housing. Composite 6, the lowest in the Atlas. Underwrite MHCs without USDA, through CDFI debt such as ROC USA Capital, state housing finance agencies or conventional execution.

Senior living: independent, assisted and memory care

Assisted living market research

Verdict: Independent living: Eligible via Section 538, High confidence; Ineligible via B&I, which states that independent living facilities are not eligible (5001.105(b)(20)). Assisted living and memory care: Eligible under B&I, High confidence, for for-profits where constant medical care is available on site; Eligible under CF, High confidence, for nonprofit, public and tribal operators of licensed care. Composite 16 for both.

Structure. For-profit assisted living: B&I to $25 million, with market practice at up to 30-year amortization, 1.20x to 1.30x DSCR and 80% LTV, plus C-PACE and NMTC. Nonprofit: CF Direct at 4.50% to 4.75% fixed for 40 years, below the market cost of debt. Against HUD 232, B&I reaches small rural assets and new construction faster; against Fannie Mae and Freddie Mac, it does not require a stabilized, scaled operator. One conflict to resolve before filing: B&I bars charitable organizations, yet USDA has guaranteed loans to a nonprofit assisted living operator.

What the study must evidence. Licensure category and the share of residents needing assisted-level care; demand from residents aged 75 and over and 85 and over, adjusted for penetration; private-pay affordability against median income and home values; Medicaid share; staffing availability; lease-up period and working-capital reserve; operator track record.

Illustrative stack, 60-unit for-profit assisted living and memory care, $18.0 million cost. B&I loan at 80% guarantee $11.7 million (65%); sponsor equity $4.5 million (25%); seller note or C-PACE $1.8 million (10%). Uses include a $1.5 million lease-up reserve and about $0.28 million of B&I fees.

Deal evidence. LifeQuest Assisted Living, Bucks County, PA: $39.9 million CF Direct for 123 assisted living units plus memory care. Bethany Home: two guaranteed loans totaling more than $7.8 million for a 16-bed memory unit and eight assisted units. Lument: $19.33 million CF Direct for a 70-bed replacement skilled nursing facility with assisted and dementia units.

Risk flags. Campuses dominated by independent living; unlicensed care models; tax-exempt tranches on the same component (5001.102(c)); owner-operator splits without control (5001.105(b)(12)).

Childcare centers

Childcare feasibility studies

Verdict: Eligible under CF, High confidence, for public and nonprofit owners; Eligible under B&I, Medium-High confidence, for for-profit operators. Composite 14.

Structure and evidence. CF Direct, grant and guaranteed, plus REDLG revolving funds, state CCDF capital, earmarks and CACFP for operations. Documented awards: Appletree, Rensselaer, IN, $133,200 CF grant for 72 children; Parent Child Center of Rutland County, VT, $833,900 CF Direct plus a $367,600 disaster grant; Coon Valley Cooperative Telephone, Menlo, IA, $300,000 REDLG-type grant to a revolving fund for a childcare center (October 12, 2026); NBRC $1 million to Tree Street Youth. No 2022 to 2026 CF Direct childcare loan with a stated amount was found. For-profit operators must not apply to CF; leases must run at least the loan term.

Rural healthcare: clinics, critical access hospitals, skilled nursing

Healthcare feasibility studies

Verdict: Eligible under CF, High confidence, for hospitals, with refinancing possible where it reduces debt service (5001.102(d)(5)); Eligible under B&I, High confidence, for for-profit skilled nursing (5001.105(b)(20)) and clinics. Composite 20, the highest in this group.

Structure. CF Direct, often combined with CF Guaranteed, plus earmarks and foundation funds; B&I plus C-PACE for for-profits. ERS finds CF recipients were 88% to 94% less likely to close, which mitigates default but not loss severity: CF Direct cohorts from 2019 to 2022 carry about $622 million of upward subsidy re-estimates.

Deal evidence. Community Hospital Association, McCook, NE: $75.7 million CF Direct (October 8, 2026). St. Croix Regional Medical Center, WI: $105 million ($95 million direct plus $10 million guaranteed, November 26, 2025). Bethany, MO: $47 million for a 14-bed critical access hospital. Centre Care, Bellefonte, PA: $35.6 million CF loan via Lument for a 240-bed skilled nursing facility on a 36-year term. X-Caliber Rural Capital: $25 million B&I plus $7.2 million C-PACE for a 150-bed facility (2021). No 2022 to 2026 B&I skilled nursing deal was found.

Risk flags. For-profit hospitals applying to CF; population limits; Medicaid rate exposure; refinancing that does not reduce debt service; more than 25% of space leased to ineligible users.

Housing and community facilities FAQ. A market-rate deal can use 538 only within the 115% AMI rent and income caps. Interest credit is not available. Maximum 538 leverage is 90% for-profit or 97% nonprofit, with Option 3 capped at 80% of cost. A 538 loan cannot refinance a bank loan. B&I cannot finance plant employee housing unless the project passes the 50% business-revenue test. A mixed assisted and independent campus is B&I-eligible if medical services are optional for independent residents or most residents need assisted care. CF cannot fund private, commercial or business undertakings.

Atlas: Main Street, services and business acquisitions

Asset eligibility is rarely the problem on Main Street; transaction structure is. Retailing and services are named directly in 7 CFR 5001.126(d)(2), and business acquisitions are eligible where jobs will be created or saved (5001.105(b)(2)). The binding rules are structural: no lines of credit or floorplan (5001.115(e)); partial buyouts only under a 24-month test; mixed-use only where at least 50% of projected revenue comes from business use; and going-concern value stripped from appraisals before collateral discounting (5001.203). Documented deal evidence is thin and most named examples date from FY2022.

Baseline terms. Maximum $25 million; 85% guarantee under $5 million, 80% above; 3.0% upfront on the guaranteed portion and 0.55% annual retention, locked for the life of the loan; negotiated fixed or variable rates, with variable rates adjusting no more than quarterly; tenor to 40 years, with lenders still applying the legacy caps of 30 years on real estate, 15 on equipment and seven on working capital; guaranteed and unguaranteed portions amortize together with no balloons. Lender practice: up to 80% loan-to-value on real estate, 60% loan-to-cost on equipment, working capital and refinance, and about 1.25x DSCR. Unlike SBA 7(a), B&I imposes no ten-year cap on acquisition amortization.

Grocery stores and food retail

Grocery feasibility studies

Verdict: Eligible, High confidence, with the strongest grant layer on Main Street. Projects that expand staple food availability in underserved areas may be located in non-rural areas and earn priority points (5001.105(b)(15)). Groceries are ineligible under Community Facilities as inherently commercial. Composite 21.

Structure. B&I for real estate, equipment, inventory as a permanent working capital term loan, and goodwill. Layers: HFFI Food Access and Retail Expansion grants ($16.5 million across 62 projects in 2024 to 2025, implementation grants typically up to $250,000); the HFFI Partnerships Program announced June 29, 2026 ($20 million, grants of $200,000 to $3 million); REAP for refrigeration efficiency; IRP for smaller deals. Grants count toward the 10% project-investment equity alternative. Against SBA 504, B&I funds inventory and business value in the same loan.

What the study must evidence. Trade area and leakage; the competitive threat from dollar stores and supercenters; gross margin (typically 20% to 30%) and shrink; SNAP share of sales; equipment age.

Illustrative stack, $3.0 million acquisition. B&I $2.55 million; buyer cash $0.3 million; HFFI grant $0.15 million.

Deal evidence. PM Ventures II, Whatcom County, WA: $4,003,000 to acquire a gas station and grocery at 5% fixed over 25 years. The Mercantile Benefit, Torrey, UT: $2,249,128 for a food co-op, restaurant, laundromat and liquor store. East Dakotah Natural Foods Cooperative, Sioux Falls, SD: $961,000 through Black Hills Federal Credit Union.

Mixed-use Main Street buildings

Verdict: Conditional, Medium confidence. Mixed use is authorized where at least 50% of projected revenue comes from business use (5001.105(b)(13)); the test is measured on revenue, not square footage. A passive landlord is excluded. Composite 15.

Structure and evidence. SBA 504 requires 51% owner occupancy for existing buildings and 60% for new construction, so it does not fit investor-owned mixed-use; B&I's revenue test does. A separate real-estate entity is permitted with the operating company as co-borrower (5001.126(a)(2)). Whether loan proceeds may fund residential build-out within a building that passes the 50% test is undocumented and should be confirmed with the state office. Precedents: Fort Henry LLC, Wheeling, WV, $5.2 million to restore a historic building with a $1.9 million borrower contribution; SLK Investment Property, rural King County, WA, $3.6 million to acquire a ten-tenant property at 4.5% fixed for five years on a 25-year term. The study must forecast the revenue split, borrower control over tenants, historic-rehabilitation costs and a rural-in-character determination where the site touches an urbanized area.

Medical and professional office

Verdict: Eligible for an owner-operator; Conditional for an investor-owned multi-tenant building, Medium-High confidence, under the 5001.105(b)(12) lease authority read with the 5001.115(g) control test. Composite 17.

Structure and evidence. Above $5 million, B&I beats SBA 7(a); against 504, it wins on investor and tenant flexibility rather than cost. Community Facilities is available to nonprofit clinics. Precedents: North Canyon Medical Center, Jerome, ID, $3,757,000 to buy a medical office building by exercising a lease option; Clarke Avenue Realty (Milford Wellness Village), Milford, DE, $25 million refinance. The study must evidence provider recruitment and retention, payer mix and lease terms. Risk flags: single-provider concentration; lender equity holdings that create conflicts (5001.127(f)).

Family entertainment, fitness and recreation

Verdict: Conditional, Medium confidence. For-profit tourism and recreation such as fitness centers, bowling alleys and family entertainment centers fit 5001.105(b)(8). Golf courses and golf infrastructure including par-3, racetracks, for-profit zoos and safaris, and public or nonprofit amusement and water parks are ineligible (5001.115(b)); the treatment of a for-profit amusement park is ambiguous and should not be inferred. Composite 12.

Structure and evidence. Expect heavy collateral discounts on special-use real estate and agency increases to the equity requirement for higher-risk projects (5001.105(d)(5)(i)). Precedents: Gold's Gym, Hollister, CA, $249,000 refinance; Silver Creek Plunge, Garden Valley, ID, $1.6 million campground resort acquisition; Frame 10 bowling alley, Kona, HI, $5 million refinance. A miniature golf component is not addressed in the regulation and carries risk under the golf-infrastructure language; obtain a preliminary eligibility review (5001.302).

Consumer services: laundromats, funeral homes, veterinary clinics

Laundromat feasibility studies

Verdict: Eligible, High confidence, as service businesses; animal boarding and training are expressly eligible (5001.105(b)(7)(ii)(E)). Composite 17.

Structure and evidence. B&I for acquisitions and equipment; for small deals, IRP loans of up to $400,000 and RMAP loans of up to $50,000. For sub-$1 million acquisitions, SBA 7(a) is often quicker; B&I's edge is the absence of a ten-year acquisition amortization cap. Precedents: Alan Rimer, Edina, MO, $385,200 to acquire a two-location funeral home; Big Kahuna Laundry, Licking County, OH, $660,021 for equipment; West River Veterinary Clinic, ND, $2 million REDLG loan through Slope Electric Cooperative. The study must evidence equipment age, preneed trust compliance and veterinarian succession.

Illustrative stack, $900,000 funeral home acquisition. B&I $765,000; buyer cash $90,000; seller note $45,000 subordinated with a life-of-loan standstill so it counts as equity.

Equipment dealerships and farm and ranch supply

Verdict: Eligible, High confidence, for real estate, acquisition and inventory term loans; floorplan and revolving lines are Ineligible (5001.115(e)). Composite 16.

Structure and evidence. B&I term debt with floorplan held outside the guarantee by manufacturer captives or conventional lenders, which forces intercreditor work on lien priority. An existing line can be refinanced into term debt (5001.102(d)(4)(ii)). Precedents: Hokanson Properties, Fair Grove, MO, $410,000 to acquire a farm store; Foth Agricultural Services, Ord, NE, $247,750 through Great Western Bank. The study must evidence dealer agreements and transfer consent, inventory turns and farm-income cycles.

Illustrative stack, $5.0 million dealership acquisition. B&I $3.5 million for real estate, goodwill and permanent working capital; buyer equity $0.75 million; manufacturer floorplan $0.75 million outside the guarantee.

Transaction types under B&I

Acquiring a rural business with B&I: goodwill and equity. Goodwill is not an ineligible use, so the full purchase price can be funded, but goodwill is an intangible (5001.3) and going-concern value is deducted from the appraisal before discounting, so discounted hard collateral must still cover the full loan. A goodwill-heavy service acquisition therefore needs outside collateral or a larger equity contribution. An existing business acquired as a going concern generally qualifies under the existing-business equity rules (10%), but whether a newly formed acquisition entity is treated as new or existing turns on the 5001.3 definition and should be confirmed in writing. Seller notes count as equity on a life-of-loan standstill and must not be paid from guaranteed funds to a seller who keeps any interest; the seller may stay on as an employee during a reasonable transition period. Earnouts are not addressed in the regulation and need state-office review. By contrast, SBA 7(a) under SOP 50 10 8 requires a 10% equity injection on a change of ownership, with a seller note counting only on full standby and only for up to half of the injection; B&I has no equivalent standby rule.

Atlas: frontier asset classes

Scale sits in the Rural Utilities Service and Section 9003, not in B&I, and B&I does not underwrite technology risk. B&I is capped at $25 million per borrower and counts both the guaranteed and unguaranteed portions and all outstanding B&I balances (7 CFR 5001.406(c)); the only exception, $40 million for value-added agricultural cooperatives, requires the Secretary's non-delegable approval and may not exceed 10% of fiscal-year funds. Three rules constrain frontier projects: non-commercially-available technology is ineligible (5001.115(h)), passive commercial rental is ineligible (5001.115(g)), and agricultural production may not exceed the lesser of 50% of the loan or $5 million (5001.105). Sponsors exceed $25 million by stacking programs, not by exception: AppHarvest's Somerset, KY greenhouse carried $50 million of 80%-guaranteed debt in July 2022 by pairing a B&I guarantee with a REAP guarantee, each capped at $25 million. The same project ended in a liquidating Chapter 11 in July 2023, and its lender was among the ten removed from the OneRD program on May 12, 2026.

Instruments beyond B&I. Section 9003 (7 CFR 4279 Subpart C) guarantees up to $250 million per borrower for biorefineries, renewable chemicals and bio-based product manufacturing, accepts technology demonstrated to have technical and economic potential for commercial application, requires an independent engineer's technical assessment and an independent feasibility study, and limits total federal participation to 80% of eligible cost. RUS electric lending carried $7 billion of FY2026 authority, has hit its cap for several years running and has notified Congress for three consecutive years that it will lend up to 25% above authority; its $3.3 billion loan to Oglethorpe Power closed October 14, 2026. The IRA cooperative programs were re-scoped rather than closed: New ERA obligated about $9 billion of $9.7 billion before awards were released for voluntary revision in March 2025; PART opened August 3, 2026 with $410 million of partially forgivable loans for hydro, geothermal, biomass and storage; and an FY2026 ARC notice followed on September 16, 2026. The 2025 tax law moves storage and bio-manufacturing up relative to solar and wind: storage keeps its credits on a phase-down beginning in 2034, while wind and solar lose them unless under construction by July 4, 2026 or in service by December 31, 2027, and foreign-entity-of-concern restrictions now apply across 45Y, 48E, 45X, 45Q, 45Z and 45U.

Controlled environment agriculture

Verdict: Conditional, Medium confidence, and paused: USDA suspended new CEA guarantees through December 31, 2026 citing 40% to 43% delinquency on $311.9 million of exposure. Composite 13.

Structure and evidence. A standalone production greenhouse is capped at $5 million of agricultural production financing; larger facilities qualify only where the financed activity is structured as processing, packing or distribution. Unproven robotic or vertical systems fail 5001.115(h). AppHarvest's loans ran 23 years at an initial 6.45% (FHLB five-year rate plus 340 basis points). The sector's failure record since 2022 includes AppHarvest, Kalera, AeroFarms, Fifth Season, Bowery and Plenty; AppHarvest is the only one with documented USDA guarantees and no public loss figure exists. The technical study must evidence crop-specific yield curves, the HVAC and lighting energy model and grower track record; the market study must evidence contracted offtake with a grocer or marketer.

Battery energy storage and microgrids

Verdict: Eligible when cooperative-owned or utility-contracted, under RUS Electric and PART, Medium-High confidence; Conditional under B&I and REAP for behind-the-meter projects. Composite 20, the highest in the frontier group.

Structure and evidence. PART explicitly finances storage that makes intermittent renewables more reliable, with a $1 million minimum and letters of interest accepted September 8 to October 9, 2026. RUS loans price off the cost of Treasury. No B&I storage transaction was identified, and no USDA guidance addresses merchant storage; assume a tolling, capacity or resource adequacy contract is required. The technical study must evidence cell supplier, warranty, augmentation and degradation schedule, integrator track record and FEOC material-assistance compliance. The August 2025 solar policy targets solar on farmland, not standalone storage.

Data centers and edge computing

Data center feasibility studies

Verdict: Conditional, Low-Medium confidence. Powered-shell and wholesale colocation face high ineligibility risk under 5001.115(g) and (a); an owner-operated enterprise or edge facility is not excluded by text. Composite 10.

Structure and evidence. A hyperscale campus cannot be meaningfully financed by B&I at $25 million. The federal capital pool for data-center-adjacent infrastructure is RUS, lending to the serving cooperative at cooperative balance-sheet credit, not project credit: the Oglethorpe plant serves 38 member cooperatives and is described by its owner as not dedicated to a data center. No USDA-financed data center was identified. Underwriting turns on large-load tariffs, contribution-in-aid, minimum bills and exit fees that protect the cooperative from stranded assets. No explicit rule on crypto mining was found.

Biorefineries and bio-based product manufacturing

Section 9003 research

Verdict: Eligible, High confidence, under Section 9003. Composite 17, with the lowest lender appetite in the Atlas (2): USDA's own 2011 rulemaking recorded applications held back because no lender of record could be found.

Structure. Guarantee of up to $125 million at 80%, or 90% where federal participation is at most 60% of eligible cost and feedstock and offtake agreements of at least one year are in place; 80% from $125 million to $150 million; 70% to $200 million; 60% to $250 million. The plant need not be rural, only in a state. Pilot data suffices under the technical-and-economic-potential prong if a qualified independent engineer supports it. A 2026 rule added a post-loss exclusion and a write-down bar. Precedent is cautionary: Range Fuels, the first 9003 guarantee in March 2010 at $80 million, shut in 2011 and its plant sold for about $5.1 million against $37 million outstanding. FY2025 to FY2026 funding is unconfirmed.

Advanced manufacturing and supply-chain reshoring

Verdict: Eligible, High confidence, subject to size. Composite 16. B&I should be sized as a tranche, not the senior facility, on any project above $25 million. No reshoring-specific Rural Development initiative exists, and 2025 to 2026 Section 232 tariff proclamations raise input costs. Underwriting turns on customer concentration, labor availability, utility capacity and tariff-driven cost shifts; the job-transfer and overcapacity screens at 5001.118(b) apply to projects over $1 million adding more than 50 jobs.

Broadband and digital infrastructure

ReConnect feasibility research

Verdict: Eligible, High confidence, under ReConnect, Telecommunications Infrastructure Loans, Community Connect and DLT. Composite 17. ReConnect uses a 20,000-population rural test and in Round 5 required at least 90% of households in a service area to lack 25/3 Mbps, excluded unlicensed fixed wireless, and offered $700 million with loans of up to $50 million. The program has spent more than $5.5 billion since 2018; FY2026 appropriation is $50.75 million; Round 6 was signaled in April 2026 but is unconfirmed, and grants are proposed to end in FY2027. Underwriting turns on subscriber take-rates, overbuild of BEAD and RDOF areas and construction cost.

Technology risk: how USDA underwrites first-of-kind projects. B&I requires technology to be proven upstream and will not finance a process that is not commercially available. Only Section 9003 contemplates early-commercial technology, and it does so through a qualified independent engineer who must show the project will operate as specified over its useful life, with pilot or demonstration results, integration data and the scale of proof, plus feedstock and offtake contracts. The documented losses in biorefineries, biodigesters and controlled environment agriculture, where recovery assumptions fall below 10%, are the evidence base for that posture. For every frontier asset, the feasibility study should separate technical feasibility from market feasibility and state plainly where no commercial-scale operating comparables exist.

Capital stack playbooks and grant layering

USDA instruments stack through program-specific caps, not a single federal-share ceiling, and for a for-profit commercial sponsor the realistic grant contribution is 0% to 3% of project cost, received after completion. Neither 7 CFR 5001 nor the REAP or REDLG rules impose a cross-program limit on combined federal assistance; each cap is cleared independently. B&I can carry up to 100% of eligible capital needs subject to the 10% to 25% equity tests, REAP grant plus REAP guaranteed loan is capped at 75% of project cost, REDLG funds up to 80%, and Section 538 lends up to 90% (for-profit) or 97% (nonprofit) of cost. Three structural rules govern every stack: B&I equity is a balance-sheet test rather than a cash-injection test, subordinated debt counts toward it only under a life-of-loan standstill, and the entire B&I loan must be secured pari passu by the same collateral (5001.450(b)(1)), so any subordinate lender sits behind the full loan, not merely the unguaranteed piece.

Pairing rules by instrument

Five worked capital stacks (illustrative, FY2026 fee notice applied)

A. Branded select-service hotel, new single-purpose entity, ground-up, $20.0 million. B&I guaranteed senior loan $14.0 million (70%; 80% guarantee, $11.2 million guaranteed); subordinated sponsor or state gap note on full-life standstill $1.0 million (5%); sponsor cash equity $5.0 million (25%, meeting the new-business test and the pre-completion guarantee test). Upfront fee $336,000; first-year retention $61,600; a further 0.50% if the guarantee is issued before completion. Federally guaranteed exposure is 56% of cost. A REAP grant on a $600,000 solar package would return about $150,000 twelve to 24 months after completion, worth about $118,500 after tax at 21% and about $89,000 in present value: 0.6% of cost. A $1.0 million REDLG pass-through at 0% is worth roughly $400,000 of nominal interest saving against a 7.5% market rate.

B. Gas station and convenience store, existing operator, new site, $4.5 million. B&I senior loan $3.6 million (80%; 85% guarantee, $3,060,000 guaranteed); REAP guaranteed loan for refrigeration, LED and canopy efficiency $450,000 (10%); operator equity $450,000 (10%). B&I upfront fee $91,800; first-year retention $16,830. The equivalent 7(a) loan would carry a 75% guaranty and an FY2026 upfront fee of $98,750: higher fee, lower coverage. Under 504, the site is single-purpose property requiring at least 15% equity. Phase I and often Phase II tank work is the principal schedule risk.

C. Section 538 new-construction family rental with LIHTC, 48 units, $12.0 million. Section 538 permanent loan $3.6 million (30%; 90% guarantee, $3,240,000 guaranteed; fixed for the life of the guarantee); LIHTC equity $7.2 million (60%); state HFA or HOME soft loan $800,000 (6.7%); deferred developer fee $400,000 (3.3%). Fees $21,060 upfront and about $11,340 annually, or $19,440 and $8,100 in a Qualified Opportunity Zone, workforce or preservation deal.

D. Cattle processing expansion, independent processor, $6.0 million. MPPEP Phase 4 grant $2.0 million (33%, the maximum award at 50% cost share); senior term loan $2.7 million (45%); sponsor equity $1.3 million (22%); VAPG working capital $200,000 off-budget if producer-majority; REAP efficiency grant later on refrigeration. Grants can reach 33% to 37% of cost in this niche, but the grant is taxable to a for-profit (about $420,000 at 21%) and award timing makes the senior loan contingent.

E. Childcare center, nonprofit owner with a for-profit operator, $2.1 million. NBRC Catalyst grant $1.0 million (48%); REDLG revolving fund loan $300,000 (14%); CF Direct loan or CF grant $500,000 (24%); capital campaign or state grant $300,000 (14%). The grant share is 48% to 70% depending on the CF tier, but only because ownership sits with a nonprofit. A for-profit operator captures the value through below-market occupancy cost; lenders should underwrite the lease, not the grant.

Grant accessibility for a for-profit sponsor (D = documented award; P = plausible under the rules, no example found; I = ineligible)

Timing and bridge rules. Grants are reimbursement-based. VAPG awards run about five months from deadline; RBDG about two; REDLG about two after each quarterly deadline; MPPEP six to nine months or more; regional commissions six to nine months. Under the new REAP model the full energy system must be financed with debt and equity first, and the grant acts as a 15- to 30-month post-completion equity return. The structuring rule is absolute: no lender should count a USDA grant toward construction sources unless an obligated Financial Assistance Agreement exists and reimbursement mechanics are bridged, and under REAP even that is impossible before completion.

Compliance and tax. Grants are generally taxable income for for-profit recipients; a 21% corporate rate cuts a $150,000 grant to about $118,500. Build America, Buy America applies to public and nonprofit recipients under the RBDG NOFO; for-profit direct recipients of REAP, VAPG and MPPEP sit outside that definition, but REAP now imposes its own bar on components from foreign-adversary countries. Davis-Bacon applicability to these business grants was not confirmed and must be verified program by program. RBDG withholds the last three months of reimbursements until final reports are approved; REAP requires a final financial report 120 days after award approval.

USDA versus SBA versus conventional debt

B&I wins on size, coverage and tenor; SBA 504 wins on fee cost for owner-occupied real estate below the debenture cap; SBA 7(a) wins on speed and lender depth. For loans under $5 million, B&I's 85% coverage now exceeds 7(a)'s 75% at a comparable upfront fee. For loans of $5 million to $25 million, B&I has no SBA substitute. For owner-occupied real estate within the $5.0 million to $5.5 million debenture limit, 504's 0.50% upfront fee and below-market debenture rate (4.37% assumed in the FY2026 Federal Credit Supplement) are structurally cheaper than either guarantee program.

Where B&I loses. The 3% fee plus 0.55% retention is a cost conventional loans do not carry; on a $4 million loan with an 85% guarantee, omitting the retention fee overstates annual cash available for debt service by $18,700. Processing depends on state-office capacity that has fallen by a third. And SBA has answered in B&I's lower size band: the FY2027 7(a) fee notice, effective October 1, 2026, sets a 0% upfront fee on loans of $700,000 or less to rural businesses.

Where B&I wins decisively. Hotels and other passive-ownership structures, which SBA excludes; acquisitions, where B&I imposes no ten-year amortization cap and no seller-note standby rule; projects between $5 million and $25 million; and any deal where 40-year amortization changes the DSCR. The guaranteed portion carries a 20% risk weight (12 CFR 3.32(a)(1)), can be excluded from the legal lending limit where the guarantee is unconditional and payable within 60 days (12 CFR 32.3(c)(4)(ii)), and trades in the secondary market.

On the risk comparison. The widely quoted claim that B&I is about half as risky as 7(a) (roughly 3.7% lifetime loss against 7.5%) divides FY2015 to FY2024 loss payments, mostly on pre-2015 vintages, by FY2015 to FY2024 obligations. At the cohort level, OMB's B&I gross default assumptions of 7.9% to 10.2% with recoveries of 19.6% to 30.2% put B&I much closer to seasoned 7(a) charge-offs (6.44%) than to SBA 504 (1.49%). B&I's low recovery assumption for a nominally collateral-secured program is the special-use collateral problem made visible.

Lender landscape and portfolio performance

Lender concentration was the B&I program's growth engine and is now its principal risk: ten of more than 775 participating lenders held about 47% of Rural Development delinquency when USDA removed them on May 12, 2026, and the program's credit subsidy has risen roughly fivefold in two years. For a sponsor, lender selection is now an underwriting variable on par with the deal itself.

The lender universe. The February 18, 2026 RBCS letter reported participation from over 775 lenders on more than $12 billion of active OneRD guarantees; after the removals USDA cites over 750. The last static league table (FY2020) ranked Live Oak Bank, Greater Nevada Credit Union, North Avenue Capital, Crestmark (now Pathward), BOM Bank, United Community and Byline among the leaders; a current ranking must be built from Lender Lens. Nonregulated lenders, including nondepository CDFIs, community development corporations, insurance companies and mortgage companies with commercial lending experience, may be approved for up to five years with renewal (7 CFR 5001.130). Participating-lender counts have declined even as volume grew, creating geographic coverage risk.

The May 2026 removals. USDA press release 0065.26 named Bank of Montgomery (BOM Bank), Byline Bank, Celtic Bank, Community Bank & Trust of West Georgia, Genisys Credit Union, Greater Nevada Credit Union, North Avenue Capital, Optus Bank, U.S. Eagle Federal Credit Union and ReadyCap Commercial, holding approximately $620 million of delinquent loans. Removed lenders cannot submit new guarantee requests; the effect on existing guarantees was not stated. Four of the ten appear on the FY2020 leader list. Genisys and Greater Nevada publicly dispute the action, ICBA and America's Credit Unions challenged the process on June 26, 2026, and the figures are USDA assertions rather than adjudicated findings.

Secondary market and regulatory capital. Guaranteed portions trade through broker-dealers on Form RD 5001-6, carry the full faith and credit of the United States, and leave servicing with the originator; sale to a holder effectively perfects the guarantee because holder rights are not contestable for lender servicing failures. Farmer Mac offers an alternative channel in which the lender earns an ongoing spread rather than a one-time premium. Premiums run from the OCC's illustrative 6% with a 1% servicing spread to the 7% average Live Oak reported on $322 million of guaranteed sales in the second quarter of 2025, with broker illustrations of 15% on long, floating, prepayment-protected paper as an upper bound. The guaranteed portion typically takes a 20% risk weight and may be excluded from the legal lending limit where the guarantee is unconditional and payable within 60 days.

Process and cost benchmarks

Approval rates and claims. OCC data show submissions, approvals and claims of 414 / 369 / 19 in FY2021, 590 / 367 / 15 in FY2022 and 596 / 314 / 38 in FY2023, implying approval rates of 89.1%, 62.2% and 52.7%. Only one partial denial of a guarantee claim occurred across the three years and there were no full denials: properly serviced guarantees are honored.

Credit subsidy rates, FY2027 Federal Credit Supplement (FCRA basis)

The B&I history runs 5.11% (FY2015), 2.32% (FY2019), 0.00% (FY2021), about 2.3% (FY2024) and 0.20% (FY2025); the FY2025 rate should be read as a cyclical low. B&I's recovery assumption of 20.72% sits well below 7(a)'s 31.18% despite similar default assumptions, so loss-given-default is priced as worse. Section 538 and CF guarantees remain the only fee-positive lines, while CF Direct cohorts from 2019 to 2022, booked at negative subsidies, now carry a combined $621.7 million of upward re-estimates. The FY2027 B&I commitment placeholder of $85.2 million reflects a proposed program reduction, not a market forecast, and is the single largest policy risk to the program.

B&I volume, FY2020 to FY2025

The loan count halved while dollar volume nearly doubled: the program has migrated toward larger loans originated by fewer specialists. USDA paid about $501.6 million in core B&I losses over FY2015 to FY2024 against $13.4 billion of obligations, with annual losses above $70 million in 2016 and 2018 and just under $30 million a year in 2021 to 2024; the roughly $300 million of repurchases and losses in the twelve months to February 2026 is a run rate about ten times the recent average, although it covers all OneRD programs and includes repurchases.

The Atlas dataset. Transaction-level public data are now available for the first time. USAspending.gov carries loan-level face value and original subsidy cost by Assistance Listing (B&I 10.768, Community Facilities 10.766, REAP 10.868, Section 538 10.438, Section 515 10.415, REDLG 10.854, RBDG 10.351), with recipient, location, action date and a transaction description that separates 3% standard loans from 1% reduced-fee loans; it carries no reliable sector code and no lender identity. Lender Lens, launched January 19, 2026 with monthly refresh, carries borrower, lender, state, close year, original amount, guarantee percentage, note rate, maturity, unpaid principal and delinquency bucket for the OneRD programs. The Rural Data Gateway adds a rolling ten years of project-level investments with six-digit NAICS for most programs. Joined on normalized recipient name, state and obligation year, and reconciled to the Federal Credit Supplement and the Summit/NRLA obligation series, these produce the first lender-attributed, performance-linked rural commercial credit dataset. Two cautions govern its use: USAspending's obligation field for a loan is the subsidy cost, never the face value, and face values reflect initial lender reporting that is generally not updated for later modifications.

Credit forensics: why USDA-financed deals fail

USDA commercial loan failures come from origination, not servicing, and they cluster in special-use and first-of-kind assets where recovery is lowest. The documented pattern has not changed from GAO's 1999 review of B&I losses, which found USDA did not follow its own requirements on loans to 18 of 24 loss borrowers and that 11 lacked required feasibility studies, through the 2026 pauses on biodigesters and controlled environment agriculture. A weak feasibility study hurts twice: it raises default probability, and it becomes the record USDA uses to argue that the lender's origination was negligent and to cut the loss claim.

Root causes

Case studies

How USDA cuts loss claims. USDA's main lever against lenders is reduction, not denial, and it holds up in court. The reduction is measured by the loss occasioned by the lender's failure: unverified equity, unbuilt improvements, unperfected liens, negligent servicing. A guarantee obtained by fraud the lender participated in or knew of is contestable outright. Few claims are contested; the defense is built at origination, in the file.

Early-warning dashboard for USDA-guaranteed credits

The regulation itself expects loan agreements to carry annual guarantor statements, limits on dividends and owner compensation, limits on asset purchases and sales, no merger or sale without lender concurrence, limits on management changes, maximum debt-to-net-worth and minimum DSCR where required, and inspection rights (7 CFR 5001.451(b)(3)).

Loss severity by asset group (1 = lowest expected loss given default; 5 = highest)

Feasibility study requirements under USDA programs

See USDA feasibility study requirements, feasibility study vs business plan, and our independent USDA feasibility study practice.

An independent feasibility study is mandatory only at defined triggers, but it has become the decisive filter: with B&I approvals at 52.7% of submissions and loss claims reduced wherever origination was negligent, the study is both the document that gets a file obligated and the lender's evidence that it was not negligent. Feasibility standards should accordingly move from market support for projections to loss-severity and guarantee-defense work.

When a study is mandatory

The $1 million trigger is measured on the whole guaranteed loan, not the guaranteed portion. Whether a newly formed acquisition entity is a new business turns on the 5001.3 definition and should be confirmed with the state office in writing.

Required components. The operative template remains the legacy B&I standard at 7 CFR 4279.150: an evaluation of economic, market, technical, financial and management feasibility, with an executive summary reaching an overall conclusion on the business's chance of success. The income approach of an appraisal is not an acceptable feasibility study. The biorefinery standard adds recommendations for implementation, competitive threats, site suitability and verification that project agreements will be in place by closing. The study feeds the lender's credit evaluation under 5001.202, the agency's priority scoring and the conditional commitment, which frequently carries borrower-performance conditions for new businesses and construction loans; unmet conditions can result in USDA declining to issue the guarantee after the loan has closed.

Recurring deficiencies. Substituting an appraisal income approach for a study; omitting the 3.0% guarantee fee and 0.55% retention fee from projections; omitting the management and technical components; analysis that is not site-specific; and undocumented author independence or qualifications. GAO's historical finding, that 11 of 24 loss borrowers lacked studies and three had significantly flawed ones, remains the reference.

What a defensible USDA feasibility study contains in 2026

1. Independence and scope. A third-party analysis covering market, technical, financial and management feasibility, not a sponsor business plan, with the author's qualifications documented.
2. Sector-calibrated downside cases. Stress cases anchored to realized data rather than generic haircuts: OMB's 9.02% B&I default assumption, the 27% to 43% delinquency in biodigesters and controlled environment agriculture, and SBA seasoned charge-off benchmarks by industry. The project should pass 1.0x DSCR under a downside that reflects its sector's documented failure mode.
3. Ramp-up and liquidity test. Monthly cash flow through stabilization, including interest-only periods, with principal and interest beginning at the break-even month as the February 2026 RBCS letter expects. Working capital and reserves sized from equity, not loan proceeds. Months of runway stated under a 20% revenue shortfall.
4. Equity and sponsor verification. The source, timing and verification of the cash contribution; sponsor liquidity and contingent liabilities; prior bankruptcies or closures. This is where claim reductions originate.
5. Collateral on a liquidation basis. For special-use assets, the going-concern value, the liquidation value after the 5001.203 deduction, and the implied recovery against guaranteed exposure, compared with OMB's program recovery assumption.
6. Technology readiness. For first-of-kind processes, independent engineering review of commercial-scale operating data, stating plainly where no comparables exist.
7. Construction-cost realism. Hard-cost inflation at the high end of current indices (about 9% or more for metal-intensive buildings in 2026) and the construction standard assumed.
8. Capital-stack fragility. Grants treated as upside unless an obligated Financial Assistance Agreement exists and is reimbursing; reimbursement delay modeled.
9. Counterparty and program checks. The originating lender's standing on Lender Lens and against the May 12, 2026 removal list; any sector pause; the fee and guarantee terms in force at obligation.
10. Guarantee-defense traceability. Each conditional-commitment covenant traceable to a study assumption, so the study is the lender's evidence that origination met the standard.

Program-specific evidence the study must carry. For seasonal borrowers, quarterly projected cash flow (5001.202(b)(6)(v)). For restaurants and retailers, cost ratios benchmarked against industry standards such as RMA (5001.202(b)(6)(iii)). For processing and cold storage, supply and offtake agreements aligned to loan term. For storage and leased industrial product, documentation of borrower control over tenants and services. For hospitality, occupancy, ADR and RevPAR against a competitive set and the franchise and management agreements. For Section 538 multifamily, unit-by-unit rent-cap compliance, absorption to 90% occupancy for 90 days and capture rates by AMI band. For senior living, licensure category and the share of residents needing assisted-level care.

Frequently asked questions

Sources and data notes

All figures are as of October 17, 2026 and are drawn from the ten underlying research briefs, each of which carries its own full source list. The primary sources below are those the briefs opened and relied on most; the Federal Register is treated as authoritative where sources conflict.

Regulation and statute

• 7 CFR Part 5001, OneRD Guaranteed Loan Regulation (eCFR): eligibility, equity, collateral, feasibility and servicing rules for B&I, CF, REAP and WWD guarantees
• 7 CFR Part 4279 Subpart B and C (eCFR): legacy B&I feasibility standard and Section 9003 biorefinery guarantees
• OneRD Annual Notice of guarantee fee rates, 91 FR 11272, March 9, 2026: FY2026 fee and guarantee schedule
• REAP final rule, 91 FR 62600, October 1, 2026: post-completion grant model and exclusions
• Section 515 and 538 transfer and guarantee action, September 25, 2026: Option 3 loan-to-cost and the 2026 pilot
• OneRD Guarantee Loan rule, September 30, 2024 and confirmation, December 12, 2025: lender oversight and refinancing amendments
• REDLG FY2026 Notice of Funding Opportunity, September 15, 2025

USDA program documents and data

• B&I Guaranteed Loan program page and OneRD Guarantee portal: current terms and funding status
• OneRD Lender Guide
• RBCS Unnumbered Letter on strong stewardship, August 19, 2025: wind and solar removed from B&I; servicing of delinquent loans
• Stakeholder announcements: 85% guarantee (March 27, 2026), REAP pause (March 31, 2026), biodigester and CEA pause (April 2, 2026)
• New and Better Markets project chart, May 24, 2022 and REAP award chart, June 26, 2024: deal evidence
• Rural Development open data dashboard methodology: Lender Lens and Investment Data Download
• FY2027 Explanatory Notes, Rural Development: program levels, staffing and proposed terminations
• Community Facilities Direct Loan and Grant program page and REAP guaranteed loan page
• USDA press release 0065.26, May 12, 2026: removal of ten OneRD lenders (as reported at Feedstuffs)

Oversight, budget and comparators

• OMB Federal Credit Supplement, FY2027 Budget, April 2026: subsidy rates, default and recovery assumptions
• GAO RCED-99-249, Rural Development: B&I loan losses, August 25, 1999
• USDA OIG audit 34-001-03-HQ and 34601-5-At: B&I loanmaking deficiencies; OIG staffing report OAI 25-064-01, December 17, 2025
• OCC Community Developments Insights, B&I Guaranteed Loan Program, June 2025: bank-regulator view of terms, approvals and claims
• Summit LLC economic assessment for the National Rural Lenders Association, House Agriculture Committee record, September 18, 2025: B&I obligations, sector mix and losses FY2015 to FY2024 (advocacy-sourced, built on USDA data)
• CRS R48564: FY2026 agriculture appropriations including congressionally directed CF spending
• One Big Beautiful Bill Act energy credit changes (Kirkland alert, August 2025)
• SAM.gov Assistance Listings: program numbers for the dataset build

Data notes and evidence gaps

• No public OIG or GAO data isolate default rates for hospitality, fuel retail, storage or Main Street assets; the loss-severity scores for those groups carry Low confidence until Lender Lens NAICS data can be tabulated.
• The most recent USDA estimate of rural eligibility population share (36%) dates from the 2000 Census; no equivalent estimate exists on 2020 geography.
• FY2024 and FY2025 B&I obligation figures conflict across the Summit/NRLA series, USDA Explanatory Notes and the Federal Credit Supplement; both figures are shown where they differ.
• The $620 million and 47% delinquency figures attached to the May 2026 lender removals are USDA assertions, disputed by two of the lenders, and not adjudicated.
• The REAP grant caps of $500,000 and $250,000 and the 25% share appear in market summaries of the October 2026 rule and were not verified in the rule text.
• Reports that the FY2027 budget request eliminates appropriated B&I authority and VAPG appropriations come from secondary sources and are treated as risks, not facts.
• The truck-parking inventory (2019), ERS Food Access Research Atlas (2019) and childcare gap analysis (2020) predate 2023 and are flagged wherever they inform a market-depth score.
• USAspending's obligation field for a loan is the subsidy cost, not the face value; face values reflect initial lender reporting and are generally not updated for later modifications.