Who Banks the Shadow Banks

Private credit has become the dominant growth engine in American lending. Firms like Blackstone, Apollo, ACORE, Blue Owl, Ares, Sixth Street, and Golub Capital now originate tens of billions annually, filling the vacuum left by banks that retreated after the SVB and Signature failures in 2023. But these firms do not fund their loans from equity and investor capital alone. Behind nearly every private credit loan sits a bank—providing warehouse lines, subscription facilities, and repo agreements that let funds amplify their returns.

This creates a two-layer credit structure that is largely invisible to regulators and market participants. Layer 1 is the loan: the private credit fund lends to a borrower. Layer 2 is the leverage: a bank extends a facility to the fund, secured by the fund’s loan portfolio or its LPs’ uncalled capital. Layer 2 lives in UCC filings buried in Delaware’s Division of Corporations and in collateral assignments recorded at county offices that almost no one systematically reads.

Executive Summary

These documents are titled “NOTICE OF COLLATERAL ASSIGNMENT OF MORTGAGES AND LOAN DOCUMENTS,” but ACRIS classifies them identically to routine mortgage transfers—indistinguishable from a CEMA refinancing. The only way to identify them is to read the instruments. Inside: advance rates, participation structures, multi-bank syndications, and warehouse relationships dating back a decade.

The Private Credit Boom

The reordering of CRE lending after March 2023 was swift and structural. Regulated depositories pulled back across construction, bridge, transitional, and permanent categories as examiner pressure mounted on banks with CRE concentrations above 300% of capital. Private credit filled the gap. Blackstone Real Estate Debt Strategies, Apollo, ACORE Capital, Blue Owl Capital, Madison Realty Capital, Oaktree, and Sculptor collectively deployed tens of billions into loans that banks no longer wanted on balance sheet.

But private credit is not self-funded. The economics depend on leverage. A fund deploying $1B in equity at a 10% coupon earns $100M gross. The same fund with $2B in bank leverage deploying $3B earns $300M gross, nets $160M after bank interest expense—a 16% return on the same equity. The bank that makes this possible does not appear in the borrower’s loan documents, does not record a mortgage, and shows up on no property-level database. But its capital is the load-bearing wall of the entire structure.

The Back-Leverage Architecture

Three standard structures dominate, each leaving a different documentary trail.

Warehouse lines — the most common — are revolving credit facilities secured by the fund’s CRE loan portfolio. As loans are originated, they’re pledged; as loans pay off or are securitized, pledges are released. Documented by a UCC-1 in Delaware and, in some cases, a county-level collateral assignment of mortgage.

Subscription facilities are secured not by loans but by the fund’s uncalled LP capital commitments. A $2B fund with $1.2B in undrawn commitments can borrow against that callable capital for working capital and bridge financing. Visible only through UCC filings; invisible at the property level.

Repurchase agreements function as secured lending structured as securities transactions. The fund “sells” a loan pool to the bank with an agreement to repurchase at a set price and date. Leave traces in UCC filings but rarely at the county level.

Two Complementary Data Sources

Two complementary sources create a complete picture.

UCC filings capture the fund-level relationship. A Delaware filing names the fund as debtor and the bank as secured party, revealing which banks are in the business and with whom—but not the specific properties or amounts.

Collateral assignments capture the property-level relationship. When a bank takes a security interest in a specific mortgage, it records an assignment at the county recorder’s office. In New York City these appear in ACRIS, naming the fund (assignor), the bank (assignee), the property address, and the loan amount.

UCC filings map the network at the institutional level. Collateral assignments ground-truth it at the property level.

The Bank League Table

Atrium assembled UCC filing data across Delaware and D.C. to construct a league table of bank back-leveragers, organized by breadth and depth of relationships.

Tier 1: Bulge Bracket

The five largest U.S. banks dominate back-leverage provision, each with a distinct strategy.

Goldman Sachs stands out for breadth—no other bank appears across as many distinct fund families. Goldman provides leverage to Apollo, BSP, Cerberus, Torchlight, and Madison Realty Capital, and participates in syndicated facilities for Blackstone. Its 10+ Apollo filings alone span multiple fund vintages.

JPMorgan Chase is the mirror image: depth over breadth. More than 40 UCC filings tie JPM to Sculptor Capital Management entities, suggesting JPMorgan is the primary if not sole warehouse provider for Sculptor’s CRE platform.

Bank of America shows the same pattern with TPG Real Estate: 30+ filings suggesting a strategic partnership rather than a transactional one.

Citibank’s most distinctive relationship runs through Athene Global Funding, Apollo’s insurance subsidiary. The chain runs four links: property borrower → Apollo origination → Athene insurance account → Citibank leverage. With 25+ filings, it represents the insurance-to-bank leverage pipeline.

Tier 2: Regional and Specialty Banks

Below the bulge bracket sits a group of regional and specialty banks that have built meaningful back-leverage businesses, often with narrower client sets.

Western Alliance Bank ($99B in assets) illustrates a different model entirely: NAV lending. NAV facilities are secured by fund-level LP interests—generating no ACRIS filing and no Delaware UCC lien against a specific property portfolio. Western Alliance serves as administrative agent for CL Forgotten Coast NAV Funding, a $3.5 billion NAV facility—the largest fund-level leverage relationship in this dataset—and also participates in Blue Owl’s $425M real estate term loan and a $450M Comvest Partners VII facility. Its near-invisibility in the ACRIS/UCC data reflects a structural choice to lend at the fund level rather than the asset level.

Exclusive relationships are more common among smaller banks. Stifel’s 7+ filings are all Oaktree; City National’s 8+ filings are all Kayne Anderson. PNC Financial has the most substantial traditional Tier 2 book, anchored by 18+ Blue Owl GP Stakes filings.

Axos Bank is the outlier. At $28 billion in assets it isn’t the smallest bank here, but its private credit back-leverage book is the most concentrated of any institution its size. The county-level evidence, detailed below, reveals a footprint far beyond what Delaware UCC data alone captures.

The County-Level Evidence

UCC filings reveal relationships. Collateral assignments reveal economics.

When a bank takes a collateral assignment on a specific mortgage, the county filing names the fund, the bank, the property, and the loan amount. Atrium systematically identified these in ACRIS—producing a property-level view of back-leverage that has not previously been assembled. Several examples illustrate the pattern:

BDT & MSD Partners to Axos Bank: A collateral assignment covering a $101 million mortgage on 250 Water Street in Manhattan. BDT & MSD Partners, the advisory and investment firm founded by Byron Trott and backed by Michael Dell, originated the loan. Axos Bank holds the security interest.

Court Square LIC Lender to Axos Bank: Two assignments totaling $275.8 million ($198 million and $77.8 million) on properties in Queens. The borrowing entity name suggests a Long Island City development. Axos Bank, a San Diego institution with $28 billion in assets, holds a security interest in nearly $276 million of New York City commercial real estate loans originated by a single borrower entity.

Emerald Creek Capital to Axos Bank: Three assignments totaling $51.9 million ($33.7 million, $12.7 million, and $5.5 million) at 660 Lexington Avenue in Midtown Manhattan. Emerald Creek Capital is a Los Angeles-based bridge lender; Axos holds the collateral interest.

Northwind Group to Axos Bank: Two assignments totaling $39 million ($27.7 million and $11.3 million) at 10 West 17th Street in Chelsea. Northwind Group is a New York-based private lender focused on transitional and bridge loans.

These assignments are public records. They have always been available. But no one has systematically connected them to the broader map of bank-to-fund leverage relationships revealed by UCC filings. The county-level data transforms the abstract question of “who finances private credit?” into a concrete, property-by-property answer.

The Axos Phenomenon

Axos Bank is a $28 billion federally chartered digital bank headquartered in San Diego, with no traditional branch network—and a private credit back-leverage book that puts it in a category of its own.

The Client Roster

Delaware UCC filings identify 21 distinct private credit and PE counterparties where Axos appears as secured party. At the top: The Carlyle Group ($426B AUM), TPG Angelo Gordon ($222B), Fortress Investment Group ($48B), and Blue Owl Capital ($235B). Below the mega-caps: ACORE Capital, Madison Realty Capital (6+ filings spanning Fund III–VI), Affinius Capital, MSD Partners, Claros Mortgage Trust / Mack Real Estate, White Oak Global Advisors, S3 Capital, CoreVest / Redwood Trust, and Scale Lending. Further into specialty PE: Victory Park Capital, Derby Copeland Capital, Catalyst Capital (industrial outdoor storage), and Real Capital Solutions (14+ filings tied to car wash and c-store portfolios).

ACRIS collateral assignments add nine more counterparties visible only at the property level, bringing the total to 30: BDT & MSD Partners, Emerald Creek Capital, Northwind Group / NEG Financing, Court SQ LIC Lender, NY Tower Capital, BCC Funding, G4 Funds, Crosswind Venture Fund, and Asset Preservation. Combined ACRIS assignment volume to Axos exceeds $500 million on NYC properties alone. The actual national book is larger.

Inside the Documents: How the Deals Work

Every ACRIS filing is classified as generic “ASSIGNMENT, MORTGAGE.” There is no special code for a collateral assignment—the only way to distinguish back-leverage from a routine transfer is to open the document. Each instrument reveals a different structuring mechanic.

Key Deal Case Studies

250 Water Street: The Warehouse Line

A nine-page instrument recorded February 19, 2026: assignors are MSD RCOF Partners CXXVI and CXXVII, LLC (c/o MSD Partners, 550 Madison Avenue); assignee is Axos Bank, San Diego. MSD—the investment firm founded by Michael Dell—originated a $120.1 million loan to 250 Water Street Owner LLC the same day the property sold for $143 million, structured across four tranches (Senior A-1 $90.3M, A-2 $10.7M, Project A-1 $17.1M, A-2 $2.0M).

The document then discloses that MSD borrowed $78,261,924 from Axos, secured by a collateral assignment of all four notes. The math is explicit: 65% advance rate against $120.1M in collateral. MSD retains $41.8M of its own equity in the loan position; Axos’s effective exposure is $78.3M on a $143M asset—approximately 55% LTV.

The document notes the assignment is “not subject to the requirements of Section 275 of the Real Property Law because it is an assignment within the secondary mortgage market.” Signed by Kenneth Gerold, VP of both MSD entities, notarized in Nassau County in December 2025.

10 West 17th Street: The Participation

An eight-page filing recorded April 28, 2026. Assignor: NW 10 West 17th Lender LLC (c/o Northwind Group, founded by Ran Eliasaf). Here the collateral assignment is not security for a warehouse loan but for a participation agreement—Axos is buying a participation interest in Northwind’s $27.7M bridge loan. The participation amount is not disclosed in the instrument. Signed by Eliasaf; Axos’s contact appears at Mail Code CLO-034 with Sheppard Mullin as counsel—the same law firm and desk code appear across multiple Axos assignments, evidencing a standardized documentation platform.

660 Lexington Avenue: The Long Relationship

Recorded February 10, 2026. Assignor: ECC SPE LLC (c/o Emerald Creek Capital, 575 Lexington Avenue). What makes this filing distinctive is the underlying facility date: “Master Loan and Security Agreement dated December 22, 2014.” Axos, then operating as BofI Federal Bank, has been warehousing Emerald Creek for over a decade.

Schedule A reveals the full origination chain: Maxim Credit Group originated the loans to 660 Lexington Avenue Development LLC (May 2023), assigned them to Emerald Creek Capital 3 (August 2024, consolidated via CEMA), which moved them into warehouse SPE ECC SPE LLC, which collaterally assigned to Axos. Four entities deep, each link documented in county records—invisible unless you pull every filing and read it.

85 Jay Street, DUMBO: The Multi-Bank Facility

A fourth pattern emerges at 85 Jay Street in DUMBO, Brooklyn, a large condo development with over 90 units. Between April and June 2026, ACRIS records show four separate agreement filings between 85 Jay Street (Brooklyn), LLC, Axos Bank, and JPMorgan Chase Bank, with recorded amounts ranging from $42.9 million to $78.6 million.

This is Axos co-lending alongside JPMorgan Chase on the same warehouse facility. A $28 billion San Diego bank and the largest bank in the United States, sharing the same deal. The multiple filings with varying amounts suggest a revolving construction warehouse line with periodic draws and modifications.

The Structuring Mechanics: What the Documents Reveal

Across these filings, several patterns emerge:

Advance rates are conservative. The one deal where both the warehouse loan and underlying collateral are disclosed (250 Water Street) shows a 65% advance rate. The fund retains 35% equity in the loan before the bank takes any loss. This is not reckless lending.

The relationships are durable. Emerald Creek’s master agreement dates to 2014. Madison Realty Capital’s UCC filings span Fund III through Fund VI. These are not one-off transactions but institutional lending platforms built over years.

ACRIS does not distinguish these from CEMAs. Every one of these collateral assignments is recorded as generic “ASSIGNMENT, MORTGAGE” (doc type ASST). In New York City, where the Consolidation, Extension, and Modification Agreement (CEMA) is used to save mortgage recording tax on refinancings, a large share of assignment filings are routine CEMA transfers. The collateral assignments that evidence back-leverage are hidden in the same category. The only way to identify them is to read the documents.

The Palmeraie: When the Loan Goes Bad

The four case studies above—250 Water Street, 10 West 17th Street, 660 Lexington Avenue, and 85 Jay Street—all involve performing loans where the back-leverage structure is invisible to the borrower and largely invisible to the market. The fifth case study is different. It is what happens when the underlying loan defaults, the borrower files for bankruptcy, and the back-leverage bank is pulled into litigation it never anticipated.

In May 2023, Madison Realty Capital originated a $585 million construction loan to Five Star Development, a Scottsdale-based developer controlled by founder Gerald Ayoub. The collateral: The Palmeraie, a 122-acre ultra-luxury master-planned development on the border of Paradise Valley and Scottsdale, Arizona—at 7000 East Lincoln Drive. The project was 80% complete at closing and includes a 215-room Ritz-Carlton Paradise Valley Resort, 80 branded Ritz-Carlton Villas (100% pre-sold at $585 million origination, representing roughly $250 million in commitments), 32 estate homes starting at $7 million, FENDI Private Residences, and a 29-acre luxury retail district. Total projected development value: approximately $2 billion. JLL Capital Markets brokered the deal.

MRC made the loan through a special-purpose vehicle, RC PV Lender I LLC (Ritz-Carlton Paradise Valley Lender), personally guaranteed by Ayoub. The loan was structured with a two-year term maturing June 1, 2025. Axos Bank entered as a direct participant in the loan at origination—holding a participation interest in the CLA with Five Star Development as the direct borrower. In September 2024—six months before MRC formally declared a default, and seven months before the loan’s maturity—Axos’s participation was converted to a loan-on-loan structure. In Axos’s own words, filed in the bankruptcy court (Dkt. 135): “Axos was a prior participant in the CLA with a Debtor as borrower, but as of September 2024, its interest was converted to a ‘loan on loan’ structure with Madison as borrower and Axos as Madison’s lender, and Axos no longer holds a participation in the CLA and Axos has no lending relationship with any Debtor.”

On April 9, 2025, MRC formally declared Five Star in default and stopped funding draws. Two days later, on May 2, a Texas state court issued a temporary restraining order blocking MRC’s foreclosure on El Paso collateral, finding the default was “artificially instigated…by [Madison’s] acts of manipulation and fraud.” On July 1, 2025, the same court granted a preliminary injunction blocking foreclosure—a higher evidentiary standard requiring a finding of probable right of recovery—and identified five categories of bona fide disputes about MRC’s conduct, accounting, and loan balance calculations. Madison abandoned its appeal of the injunction on October 15, 2025.

Two days after the Texas injunction, Madison filed notice of a trustee’s sale on Arizona collateral. On November 4, 2025, Five Star filed Chapter 11 in the Southern District of Texas, 34 debtor entities (Case No. 25-90607), automatically halting the Arizona foreclosure scheduled for November 12.

The bankruptcy immediately surfaced Axos. Five Star issued a Rule 2004 subpoena to Axos Bank on November 19, 2025. Axos filed a Motion to Quash on December 5, 2025 (Dkt. 135), arguing it “is not a creditor or party in interest in this case,” and estimated compliance costs of $219,000 to $494,000. The Official Committee of Unsecured Creditors opposed the motion. Their position, quoted in Axos’s own reply brief (Dkt. 193): “Axos repeatedly emphasizes that it is a ‘non-party’ to the bankruptcy, but that is not correct—it remains an economic party in interest through its loan-on-loan arrangement with Madison and was previously a direct participant in Madison’s Loan to the Debtors.” Axos acknowledged in a footnote to its own filing that “Axos holds a security interest in certain of the Debtors’ property.”

Five Star’s allegations against MRC are extraordinary: an employee forged founder Gerald Ayoub’s signature on construction cost approvals; MRC withheld payments to hotel contractors while directing proceeds toward villas (which generate near-term returns to MRC); MRC failed to credit $110 million in villa sale payments against the loan balance; MRC directed local officials to block certificates of occupancy to prevent villa closings. The Arizona suit invokes RICO. MRC’s counter: Five Star defaulted on a matured loan, and MRC funded an additional $79 million after the default to protect the project. The Texas court’s July 2025 preliminary injunction finding suggests Five Star’s version of events has substantial merit; the ultimate determination awaits trial.

On December 22, 2025, the bankruptcy court entered a Villa Sales Order protecting buyers. On February 27, 2026, Five Star secured $32 million in DIP financing from BH Capital Ventures LLC to advance construction. The foreclosure is still stayed. The Arizona lender liability trial against MRC was expected in early 2026. Axos’s subpoena dispute was largely resolved through voluntary document production negotiations. The Palmeraie remains open, unfinished, and contested—a live illustration of how back-leverage exposure travels from property to fund to bank, invisibly, until a bankruptcy filing forces it into the record.

The People Behind the Network

The back-leverage market is not anonymous. Each relationship runs through a team—often a small one—built by a specialist who saw the opportunity in private credit warehousing before it became institutional. The table below identifies the key individuals and their institutional mandates.

Key Team Leads by Bank

Several institutional dynamics emerge from this roster. Goldman Sachs and JPMorgan treat back-leverage as a product within their capital markets infrastructure—broadly available but managed within credit limits. Wells Fargo has created a dedicated “Specialty Real Estate Finance” function to house these relationships. And Axos—the outlier—has a standalone team dedicated exclusively to the product, reporting directly to executive leadership, with a mandate to serve a client set that the large banks underserve.

JPMorgan’s March 2026 pullback from private credit lending—after marking down software loans in private credit financing portfolios—is a significant data point. When the largest bank in the country reduces its appetite for back-leverage, the smaller specialists fill the gap. Axos, Northeast Bank, and Emigrant are positioned to absorb that deal flow. The implication is that concentration in back-leverage provision may be increasing, not decreasing, even as the private credit market grows.

The Scale Question

Forty-plus distinct private credit and PE counterparties. A $28 billion bank providing warehouse leverage to firms whose combined assets under management exceed $1 trillion. Carlyle alone manages 15 times Axos’s total assets. The diversity of the client base is as notable as its depth: from mega-cap PE (Carlyle, TPG) to New York City bridge lenders (Emerald Creek, S3 Capital, Northwind) to specialty verticals (CoreVest in single-family rental, Real Capital Solutions in car washes, Catalyst in industrial outdoor storage).

The question is one of correlation, not merely size. At $1.4 billion, the lender finance book is roughly 5% of Axos’s assets—but a single specialized team connects the bank to the credit performance of 40+ distinct private credit platforms simultaneously. This is second-order credit risk: the bank is underwriting the underwriters. The question is not “will this building cash flow?” but “will this fund remain solvent if multiple loans default at once?”

In June 2024, Hindenburg Research published a short report on Axos Financial, identifying $9.91 billion in CRE and multifamily exposure (53% of the loan book vs. a 16.5% peer average), 292% of tangible book value in CRE concentrations, and alleged extend-and-pretend tactics on NPLs. The report named no private credit back-leverage counterparties—the Palmeraie conversion happened three months later. But the thesis connects: if Axos’s direct CRE book already concentrated risk at the margins of institutional underwriting, the funds it warehouses—which themselves lend to borrowers who can’t get bank financing—represent a third-order concentration of the same risk.

Beyond Axos: Seven More Private Credit Lenders in the Back-Leverage Network

Axos is the most visible player in the ACRIS data, but a systematic search reveals a broader ecosystem. Seven more lenders, each with a distinct back-leverage architecture, emerge from the same public records.

Affinius Capital (formerly Square Mile Capital)

Affinius Capital (formerly Square Mile Capital) has built a warehouse infrastructure under the “SM Finance” brand: SM Finance II and III are internal SPVs with tranche designations (III-C, III-G, Warehouse 4, Warehouse 5) suggesting a programmatic facility. In ACRIS, Affinius assigns into these vehicles across major transactions: $228M at President Union LLC (Brooklyn, July 2025), $82M at Third Avenue Pavilion Residences (Manhattan, February 2026), and $67M at 570 Fulton Street (Brooklyn, May 2025). These are intra-platform assignments—the internal plumbing of a large institutional credit platform.

The external bank relationship surfaces at BG Junction Realty LLC, where Affinius assigned directly to Emigrant Bank across three tranches totaling ~$60M in October 2025. Emigrant, a NYC private bank (~$10B in assets), is also a repeat warehouse provider to Madison Realty Capital and Mesa West Capital—suggesting a focused CRE warehouse operation targeting mid-market private credit lenders.

Claros Mortgage Trust / Mack Real Estate Credit Strategies

Claros Mortgage Trust (managed by Mack Real Estate Credit Strategies) is a heavy user of Axos warehouse capacity. The clearest example: CMTG Lender 27 LLC assigned a multi-parcel position at 517 West 35th Street (Block 707, Midtown West) to Axos in August 2024. Mack originated three tranches against the block totaling $143.6M in face value (2018–2021), amended four times. Against that collateral, Axos extended a $95M warehouse loan at 66% advance rate. A credit amended four times over six years and still pledged to a warehouse line is one whose resolution has been deferred—Axos’s exposure depends on whether CMTG can refinance, sell, or enforce before the facility itself matures.

Arbor Realty Trust and the Named JPMorgan Warehouse

Arbor Realty Trust has constructed the most explicitly named back-leverage vehicles in the dataset: ARBOR JPM TERM FUNDING, LLC and its predecessor ARBOR JPM FUNDING, LLC. No ambiguity—JPMorgan Chase is Arbor’s warehouse provider. In March 2026, Arbor assigned a $61M multifamily loan at 22 Chapel Street (Brooklyn) to the Term Funding vehicle; in September 2023, a $35.4M loan at 331 Park Avenue South to the prior-generation vehicle. As one of the country’s largest multifamily bridge lenders, each individual ACRIS assignment is a fraction of the full pledged portfolio.

LoanCore Capital

LoanCore Capital, the CRE lending platform backed by GIC (Singapore’s sovereign wealth fund) and the Canada Pension Plan Investment Board, appears in the ACRIS data as assignor to its “LCC SS TRS HOLDCO LLC” warehouse vehicle across several large Manhattan transactions: a $156 million loan at 1 Whitehall Street (July 2025), a $60 million loan at the Suspenders Building in SoHo (multiple assignments over 2023–2026), and a $59 million loan at a Broadway multifamily (August 2024). The “SS” designation in the vehicle name suggests a Special Situations or Structured Strategies mandate.

LoanCore’s most notable transaction in the dataset is the 2024 sale of a $124 million loan at 105 West 28th Street in Manhattan to Reznik Paz Nevo Trusts Ltd., an Israeli institutional investment trust. This is not a back-leverage structure but a secondary market loan sale, and it illustrates a different risk pathway: institutional CRE loans finding their way into foreign portfolios through private market transactions that appear nowhere in standard databases.

Mesa West Capital (Morgan Stanley Real Estate)

Mesa West Capital, Morgan Stanley’s real estate debt platform, uses a programmatic warehouse structure under the “MW” prefix: MWREIF V URS SPE, LLC (for Mesa West Real Estate Income Fund V) and MWCLF RS SPE, LLC (for Mesa West Core Lending Fund). In October 2025, the MWREIF vehicle received a collateral assignment covering a $188.5 million loan at 140 Riverside Boulevard in Manhattan. In May 2026, the MWCLF vehicle received a $50.9 million assignment at 24 West 25th Street. Emigrant Bank appears as a direct assignee on smaller Mesa West loans, confirming it as a recurring warehouse lender for this platform as well.

G4 Funds: The Bridge Originator-Aggregator

The G4 Funds series (G4 18xxx, LLC) represents a distinctive business model that does not fit neatly into the originator-or-warehouse categories. G4 entities appear on both sides of the secondary market: buying performing CRE loans from established firms and selling or assigning them to warehouse providers. In April 2026, G4 18302 purchased $129.6 million in loans from Starwood Capital Group on a New Rochelle Industrial Development Agency credit. In prior transactions, G4 sold loans to Affinius Capital ($67.8M + $22M + $10.2M at 101 Fleet Street, Brooklyn) and to Northeast Bank.

G4’s warehouse relationship with Axos Bank is significant: seven distinct G4 fund entities (G4 18270, 18281, 18289, 18292, 18297, and others) appear in ACRIS assigning loans to Axos, with disclosed amounts ranging from a $100,000 small loan to a $49.5 million assignment at 102 Fleet Street, Brooklyn. The G4 model—buy loans from established platforms, warehouse with Axos and Northeast Bank, repeat—is one of the more unusual structures in the dataset.

NEG Financing: The Northwind / Eliasaf Group Infrastructure

The NEG Financing series reveals Northwind Group’s full infrastructure: NEG Financing 1, 3, 4, 7, 12 are systematic origination and warehousing vehicles (NEG = Northwind / Eliasaf Group). The entities dual-assign loans simultaneously to an external bank and an internal SPV—the bank takes the security interest, the internal entity retains servicing. In March 2026: NEG Financing 1 assigned a $19M loan at 5 East 59th Street to Axos; NEG 4 assigned $23.5M at 136 Prop Co to Axos; NEG 7 assigned $20M at 11 West 56th to Axos; NEG 12 assigned $22M at Gardner Purchaser LLC (Brooklyn) to Northeast Bank. ACRIS forces recording of both assignments, making the structure unusually transparent.

Northeast Bank: The New Tier 3 Player

Northeast Bank ($2.5B, Portland ME) appears as assignee in multiple NYC transactions: G4 18228 ($11.2M, Franklin 175), NEG Financing 12 ($22M, Gardner Purchaser LLC). Like Axos, it has built an NYC CRE bridge warehouse business disproportionately large relative to its asset base—and largely invisible in standard bank analysis.

The C&I Mirror: BDCs and the Same Banks

The back-leverage story does not end at real estate. The same banks providing CRE warehouse lines are also the primary leverage providers to BDCs and C&I private credit vehicles—with one key difference: BDC credit facilities are disclosed syndicated loans with full terms, syndicate rosters, and maturities. The scale is staggering.

Ares Capital (ARCC), the largest publicly traded BDC, maintains a $5.28B revolving facility (JPMorgan as admin agent), expanded in May 2026 to $4.15B revolver + $1.16B term loan, maturing 2031. A separate $800M facility through Ares Capital JB Funding runs through SMBC.

FS KKR Capital Corp carries a $4.67B revolving facility with SMBC as documentation agent (amended July 2025, matures 2030). Legacy FS Investment vehicles add another $700M+ across three facilities.

Blue Owl Capital is the most prolific borrower with 16 separate credit facilities totaling more than $20B—spanning its BDC, net lease REIT, tech lending, and corporate vehicles. Lender roster includes JPMorgan, Bank of America, Wells Fargo, State Street, MUFG, and Société Générale. Notably, PIMCO and Apollo Global Funding also appear as lenders, illustrating how the line between borrower and lender in private credit is increasingly blurred.

Sixth Street runs three major facilities totaling $5.5B+ across Sixth Street Lending Partners ($2.28B, JPMorgan/Wells Fargo), Sixth Street Specialty Lending ($1.68B), and Opportunities Partners VI ($1.65B). Golub Capital BDC carries $2.0B with JPMorgan as admin agent, syndicated to US Bank, Santander, MUFG, SMBC, Truist, and Wells Fargo. Blackstone Secured Lending (BXSL) maintains a $2.18B multi-currency facility across USD, CAD, GBP, EUR, and AUD tranches.

Main Street Capital: The Regional Bank Exception

Main Street Capital is the outlier. Where every other major BDC borrows from bulge bracket banks and Japanese megabanks, Main Street’s $1.15 billion revolving credit facility (Truist as admin agent) is syndicated to a roster of Texas and Southern community banks: Bank OZK, Frost Bank, Texas Capital, Veritex Community Credit Union, Woodforest National Bank, Cadence Bank, Comerica, and First Financial Bankshares. A second $240 million facility is 83% Truist, 10% First Financial, 6% Zions Bancorporation. This is a BDC that banks locally, an anomaly in a universe where cross-border megabanks dominate.

The Hidden Dominance of Japanese Megabanks

Perhaps the most surprising finding in the syndicated loan data is the dominance of Japanese banks. Sumitomo Mitsui Banking Corp (SMBC) appears across 11 deals spanning 9 borrower groups, making it the single lender with the broadest reach in BDC finance. MUFG Bank appears in 6 deals, Mizuho in 4, and Sumitomo Mitsui Trust in 2. Together, Japanese megabanks are the most active national group in back-leverage provision to U.S. private credit.

State Street Bank is the other quiet giant: 12 deals across 5 borrower groups, leveraging its custody and prime brokerage relationships to embed itself into credit facilities for Blackstone, Blue Owl, FS KKR, New Mountain, and Sixth Street.

Beyond Banks: Insurance Companies and International Lenders

The UCC data reveals two additional categories of back-leverage providers that existing regulatory frameworks are even less equipped to monitor.

Insurance companies appear as secured parties in Starwood Property Trust’s UCC filings. MetLife holds security interests in multiple Starwood Property Mortgage sub-entities—Sub-20-A, BC, and others—reflecting direct insurance company back-leverage to a mortgage REIT. Starwood simultaneously runs at least 12 distinct facility relationships: Bank of America (Sub-10-A), Wells Fargo (Sub-2, Sub-12), Goldman Sachs (Sub-31-A), JPMorgan Chase, Morgan Stanley, Barclays (Sub-22-A), Capital One (Sub-21, Sub-19), BMO Bank, MetLife, Citibank (Sub-6-P), and First Citizens Bank. The mortgage REIT model—in which a fund borrows from equity investors, securitizes to the agencies, and simultaneously runs 12 leverage facilities with every major bank plus an insurance company—is the most complex back-leverage architecture in the dataset.

International banks are also active in the back-leverage market. OCBC (Oversea-Chinese Banking Corporation, headquartered in Singapore) appears as secured party on Blue Owl Diversified Lending II. The Bank of Nova Scotia appears on PGIM Real Estate’s Agency Financing LLC and Nuveen’s CLO Issuance Fund III. Natixis (French) appeared on multiple Fortress Credit CLO vehicles before terminating in 2024. HSBC backs PGIM RE Borrower VI. These relationships are entirely outside U.S. regulatory visibility: a U.S. bank examiner reviewing PGIM, Blue Owl, or Fortress would have no visibility into the foreign bank facilities securing their loan portfolios.

Inside a Single Relationship: Blackstone BREP VII

To understand the granularity of back-leverage, consider Blackstone Real Estate Partners VII, a single flagship fund. In the UCC data, this one fund fragments into more than 20 distinct legal entities, each with its own filing:

Every single entity, Bank of America. All 40+ filings on a single date (December 5, 2025). This is what a subscription facility looks like in UCC filings: the bank is not lending against property, it is lending against the uncalled capital commitments of Blackstone’s limited partners across every onshore, offshore, tax-exempt, and non-qualifying investor vehicle in the fund structure.

The ABP Windward vehicle tells the opposite story. Where BREP VII uses subscription leverage from a single bank, ABP Windward (a Blackstone real estate operating entity in Hawaii) uses asset-level leverage from a five-bank syndicate: Citibank, Deutsche Bank, Goldman Sachs, Santander, and Wells Fargo.

Same firm. Two leverage structures. Entirely different risk profiles. The UCC data makes both visible.

What It Means

The back-leverage map reveals four systemic dynamics that merit scrutiny.

Concentration Risk Is Invisible to Existing Frameworks

No public database aggregates bank-to-private-credit leverage relationships. Call reports show total CRE exposure but don’t distinguish direct loans from warehouse lines to funds. The Fed’s SLOOS captures CRE lending standards, not fund finance. The OCC’s CRE concentration guidance applies to direct lending, not non-bank warehouse lines.

The result is a regulatory blind spot. A bank examiner reviewing Axos’s call report sees CRE exposure. But the systemic significance—that it connects Axos to Madison Realty Capital, ACORE, Blue Owl, Emerald Creek, Northwind, BDT & MSD, Claros/Mack, G4 Funds, NEG Financing, and dozens more—is invisible in any standardized disclosure. The Palmeraie demonstrates: Axos held a participation and then a loan-on-loan in a $585M construction loan on a high-profile luxury resort, and it first surfaced in public records only when a bankruptcy court issued a subpoena.

The Leverage Chain Creates Correlated Risk

When multiple private credit funds warehouse with the same bank, the bank’s exposure is correlated to the broader CRE market, not any single fund. Broad CRE deterioration generates correlated margin calls and covenant breaches across the entire warehouse book simultaneously.

This dynamic drove the MFS collapse (Market Financial Solutions, the UK bridging lender that entered administration in February 2026): HSBC, Barclays, and Santander discovered their exposures—treated as independent credits—were all tied to the same underlying collateral. Axos alone holds warehouse positions across 40+ distinct private credit counterparties; similar correlated stress exists domestically.

The Pullback Signal: JPMorgan’s March 2026 Retreat

In March 2026, JPMorgan reined in its lending to private credit firms after marking down software loans in private credit financing portfolios. The largest U.S. bank—with 42 UCC filings across Sculptor alone—reduced its appetite for fund-level warehouse and NAV lending.

When JPMorgan withdraws, displaced funds seek alternatives. The alternatives are smaller, more concentrated, and more exposed. Axos, Northeast Bank, and Emigrant Bank are positioned to absorb that deal flow—increasing back-leverage concentration in smaller institutions precisely when large-bank risk managers have concluded the product requires caution.

Small Banks in Big Games

Axos is the sharpest illustration, but Stifel, City National, Emigrant, Flagstar, and Northeast Bank all appear in the data with meaningful private credit relationships. The mismatch between bank size and fund size creates a structural vulnerability: Axos ($28B) is warehousing firms like Carlyle ($426B AUM). The 15:1 size ratio is the risk.

The broader ecosystem amplifies the picture. Mega-cap platforms (Morgan Stanley’s Mesa West, Goldman Sachs’s LoanCore) use internal warehouse SPVs that obscure the end bank. Mid-market lenders (CMTG, NEG Financing) warehouse directly with Axos. Bridge aggregators (G4 Funds) split between Axos and Northeast Bank. Affinius uses internal vehicles and Emigrant Bank. The back-leverage market is not a monoculture—it is a spectrum of arrangements calibrated to lender size, strategy, and banking relationship. It is larger, more complex, and more systemically interconnected than any existing regulatory framework was designed to see.