Castlelake, Redwood Trust Launch $8B Jumbo Mortgage Platform

Castlelake and Redwood Trust are placing an $8 billion bet that wealthy homebuyers still need loans — even if traditional banks don't want to make them. The two investment firms announced a strategic joint venture on Tuesday to purchase prime jumbo mortgage loans, targeting borrowers whose financing needs exceed the conforming loan limits set by Fannie Mae and Freddie Mac.

It's a contrarian play. While most of the mortgage market has spent the past two years wrestling with rate volatility and shrinking volumes, jumbo loans — those above $766,550 in most markets — have become an orphaned asset class. Banks that once dominated this space have pulled back, squeezed by capital requirements and the end of easy securitization markets. That retreat created the opening.

The venture will source loans from Angel Oak Mortgage Solutions, a non-bank lender that's been expanding its origination footprint as traditional players exit. Castlelake will provide the bulk of the capital, while Redwood — a real estate investment trust with two decades of mortgage market experience — will handle asset management and contribute its balance sheet for additional firepower.

The structure is straightforward: buy the loans, hold them, earn the spread. No exotic derivatives, no leverage games. Just old-fashioned credit risk on borrowers who can afford $2 million homes but need financing to close.

Jumbo mortgages used to be a bank product. Wealthy borrowers worked with private bankers at JPMorgan, Wells Fargo, or regional institutions that kept the loans on their balance sheets. That model worked fine when capital was cheap and housing prices climbed steadily.

Then 2022 happened. The Federal Reserve's rate hikes didn't just kill refinancing activity — they turned bank balance sheets into liability puzzles. Deposit costs spiked while the value of existing mortgage portfolios cratered. Banks responded by tightening lending standards and raising rates on jumbo products, pricing themselves out of all but the most creditworthy deals.

At the same time, private credit funds and non-bank lenders saw an opening. They didn't have deposit bases to defend or regulators breathing down their necks about concentration risk. What they had was dry powder and a willingness to underwrite loans that cleared every credit hurdle but didn't fit neatly into a portfolio manager's risk model.

Angel Oak built its business in this gap. The Atlanta-based lender focuses on prime borrowers — FICO scores above 700, loan-to-value ratios below 80%, full documentation — but structures deals that might not pass a bank's automated underwriting system. Self-employed borrowers. Multi-property investors. High-earners with unconventional income streams.

Castlelake, a Minneapolis-based alternative investment firm managing over $23 billion, isn't new to structured credit. The firm has spent years buying aviation assets, specialty finance receivables, and other cash-flowing instruments that institutional investors overlook. Mortgages fit the playbook: predictable cash flows, collateralized by hard assets, originated by a partner with underwriting discipline. Castlelake's role here is primarily capital provider and co-investor.

Redwood Trust brings operational infrastructure. The Mill Valley, California-based REIT has been aggregating, securitizing, and managing residential mortgage loans since 1994. It pioneered the private-label jumbo securitization market before the financial crisis, collapsed alongside everyone else in 2008, and rebuilt itself as a mortgage credit investor with a focus on non-agency loans.

Redwood's expertise matters here because jumbo loans aren't plug-and-play. Unlike conforming loans that get sold to Fannie or Freddie with standardized terms, jumbo mortgages require hands-on servicing, active interest rate risk management, and a clear path to either securitization or whole-loan sale if the market turns. Redwood has done all of this before.

Angel Oak, for its part, gets a committed buyer for every loan it underwrites to the venture's standards. That's a significant advantage in a market where whole-loan buyers have become pickier and securitization windows open and close unpredictably.

The venture doesn't have a formal name yet — the press release just calls it a "joint venture." That's typical for these partnerships, which often function as special-purpose vehicles rather than branded platforms. Castlelake will contribute the majority of the equity capital. Redwood will invest its own balance sheet dollars and provide asset management services in exchange for fees and a smaller equity stake.

Eight billion dollars sounds like a lot of mortgages. It is — but context matters. The entire jumbo loan market originated roughly $250 billion in 2023, down from over $400 billion in 2021, according to industry estimates. If this venture hits its full $8 billion deployment target over the next few years, it'll represent about 3-4% of annual jumbo origination volume.

That's not market-moving size, but it's meaningful share for a single buyer. More importantly, it signals that institutional capital sees value in an asset class that retail banks have largely abandoned. If Castlelake and Redwood are right — if default rates stay low and prepayment speeds remain manageable — other credit funds will follow.

The unit economics depend entirely on the spread between what borrowers pay and what the venture's cost of capital is. Current jumbo mortgage rates hover around 7.0-7.5% for prime borrowers, depending on loan size and property location. If Castlelake's blended cost of capital (equity plus any leverage) sits around 5-6%, that leaves 150-250 basis points of spread to cover servicing costs, credit losses, and investor returns.

Tight, but workable — assuming credit performs. And that's the bet.

Prime jumbo borrowers default less than almost any other mortgage category. These are high-income households with substantial assets, low leverage ratios, and strong credit histories. The loans Redwood and Castlelake are targeting — those originated by Angel Oak to "prime" standards — should have loss rates well below 1% annually, even in a downturn.

Credit risk is the obvious one, but probably not the biggest. The real danger is interest rate risk and prepayment volatility. If rates drop sharply, borrowers refinance and the venture's assets disappear faster than planned. If rates spike further, the loans sit on the books longer, but new origination volume dries up and the partnership struggles to deploy capital.

There's also execution risk. The partnership depends on Angel Oak's ability to originate $8 billion in loans that meet Redwood's underwriting standards. If credit boxes tighten, if housing prices roll over, or if competition for prime borrowers heats up, the pipeline slows. Then the venture sits on committed capital earning nothing.

Timing is everything. The jumbo market is dislocated but not broken. Banks are out, but borrowers still exist. Origination volumes have stabilized after the 2022-2023 collapse, and credit performance has remained rock-solid even as the broader economy cooled.

For Castlelake, this is classic opportunistic credit investing. Find an asset class where fundamentals are sound but traditional capital providers have stepped back, and step in with patient capital at attractive yields. For Redwood, it's a return to its roots — aggregating jumbo loans for institutional buyers, just without the securitization piece upfront.

The deal also reflects a broader shift in mortgage finance. Non-bank lenders now originate more than 50% of all residential mortgages, and their share of the jumbo market is even higher. These lenders don't hold loans — they originate and sell. Partnerships like this one provide the exit liquidity that makes their business model work.

If this venture succeeds, expect more like it. Private credit funds are sitting on record amounts of dry powder, and residential whole loans offer yield, collateralization, and duration that's hard to find elsewhere. The jumbo market, once the exclusive domain of relationship bankers and portfolio lenders, is becoming institutionalized.

For the wealthy homebuyer applying for a $2 million mortgage, this deal changes nothing immediately. They'll still apply through a broker or directly with Angel Oak, submit the same documentation, and receive the same loan terms. Behind the scenes, though, the loan gets sold into a different capital structure — one backed by alternative asset managers instead of a bank's deposit base.

Over time, more competition for jumbo loans could mean better pricing. If institutional buyers keep entering the market, originators will have more exit options, which should compress spreads. That's good for borrowers, but it also compresses returns for investors like Castlelake and Redwood. The cycle continues.

This partnership is one data point in a larger reconfiguration of housing finance. The 30-year fixed-rate mortgage — the defining product of American homeownership — exists because of government guarantees. Fannie and Freddie absorb interest rate risk and credit risk, making it possible for lenders to offer long-term fixed rates at scale.

Jumbo loans don't get that subsidy. They're purely private credit, and the institutions willing to provide that credit have changed. Banks used to do it because they had cheap deposits and could hold loans at book value. Now, capital markets investors do it because they're willing to accept lower liquidity in exchange for yield.

The table above oversimplifies, but the trend is clear. The jumbo market has migrated from balance sheet lenders to capital markets buyers. Castlelake and Redwood are betting that migration continues — and that they can earn mid-teens returns buying loans that banks no longer want.

Whether that bet pays off depends on factors the partnership can't control: Fed policy, housing price trends, recession risk, employment stability among high earners. But the structure itself is sound, the collateral is strong, and the sponsors have done this before.

The venture is live. Angel Oak is already originating loans, and the first purchases have likely closed or will close within weeks. The $8 billion target is aspirational — a maximum capacity figure, not a committed deployment schedule. How quickly the partnership reaches that number depends on market conditions, origination volume, and whether the initial loan pools perform as modeled.

If performance is strong, Castlelake and Redwood will likely raise follow-on funds or expand the partnership. If credit deteriorates or rates move sharply, they'll slow purchases and wait. That optionality is one reason partnerships like this have proliferated — they allow institutional investors to access mortgage credit without building origination platforms or hiring servicing teams.

For the jumbo mortgage market, this deal is both a symptom and a catalyst. A symptom of banks' retreat from balance sheet lending. A catalyst for more private capital entering a space that was once exclusively institutional banking territory.

The next question: who's next? Other credit funds have watched Redwood rebuild its jumbo business over the past five years. If this partnership works, expect similar announcements from Apollo, Blackstone, Ares, or other alternative asset managers looking for yield in a low-default, hard-asset-backed credit market.

For now, though, Castlelake and Redwood have first-mover advantage in a market that's ripe for consolidation. Whether $8 billion is the right size or just the starting point remains to be seen.

Neither firm disclosed the expected returns on this partnership, the target loan-to-value ratios, or the geographic mix of properties they'll finance. Those details matter. A portfolio concentrated in California and New York has different risk characteristics than one spread across secondary markets. Loans at 75% LTV perform differently than those at 80%.

The partnership also didn't specify whether it plans to securitize these loans or hold them to maturity. Redwood has historically been a securitization shop, packaging loans into bonds and selling them to yield-hungry investors. But the private-label securitization market has been inconsistent since 2022, and holding loans whole might offer better risk-adjusted returns in this environment.

Finally, there's the question of scale. Is $8 billion enough to matter, or is this a test case for a much larger platform? If Castlelake and Redwood can deploy this capital efficiently and generate returns above their cost of equity, the logical next step is raising institutional money — pension funds, insurance companies, sovereign wealth funds — and scaling the strategy to $20 billion or more.

That would transform this from a partnership into a platform. And platforms, in private credit, tend to be where the real money gets made.