Castellan Real Estate: The Book Looks Conservative, but 2026 Will Test the Funding Structure

Getting To Know The Company

Castellan Real Estate is not a typical Israeli real-estate issuer, and it is not a bank either. It is a bond-only public wrapper that was formed in March 2025 to hold, through subsidiaries, a portfolio of short-duration U.S. bridge loans, all secured by first mortgages on real-estate assets. Anyone reading it like a growth equity story or like a developer will miss the point. This is a real-estate backed credit vehicle, and the real question sits in bondholder protection and in the company’s ability to fund and refinance the book while it grows.

What is working right now is fairly clear. By the end of 2025, the portfolio stood at 85 loans with a net fair value of $471.1 million, average LTV of 49.7%, average coupon of 9.6%, and average remaining life of 13.2 months. The company also entered the public market with two bond series and now reports covenant capital of $149.0 million, net financial debt to net CAP of about 68%, and $343 million of unused funding capacity.

But this is still not a clean picture. The active bottleneck is not underwriting quality. It is funding quality. Almost all assets are already pledged, consolidated operating cash flow was negative $106.2 million because the book kept growing, and the parent company itself had only $163 thousand of unrestricted cash at year-end. On top of that, the capital that protects the bonds includes a $39.2 million subordinated shareholder loan, not only common equity.

That is also where a superficial read can go wrong. Consolidated net assets of $109.8 million may look like a sharp drop from $153.2 million a year earlier, but a large part of the decline came from a $50 million reclassification from equity into a subordinated shareholder loan. On the other hand, anyone focusing only on the covenant capital number of $149 million can miss the fact that this protection layer is not made entirely of permanent common equity sitting inside the issuer.

The economic map here is short and important:

For a credit reader, the real 2026 question is therefore not whether the book is large. It is whether the conservatism of that book continues to translate into liquidity, covenant room, and stable access to funding even when loans need extensions or refinancing.

Events And Triggers

The bond issuance created a new funding layer

July 2025 was the defining event. The company completed the issuance of two bond series for net proceeds of about $127.2 million. Series A was issued unsecured at a 6.85% nominal coupon, while Series B was issued secured at a 5.85% nominal coupon. That improved funding diversification, but it also introduced a fixed public-shekel liability layer on top of a dollar asset base.

Capital moved upward, then changed form

At the same time the assets were transferred into the company, $50 million was reclassified from equity into a subordinated shareholder loan. This is not an accounting footnote. It improves covenant protection for the bonds because the subordinated loan is counted for covenant-capital purposes, but it also changes the quality of the equity that remains as ordinary owner capital. That is why the balance-sheet equity decline does not tell the full protection story, and the covenant number does not by itself describe true shareholder-level capital quality.

This chart is one of the key ways to read Castellan. Capital did not fall because of credit losses. It moved between layers. That is better than a scenario of real economic impairment, but it also means bond protection depends not only on earnings but on sponsor capital discipline.

The March 2026 dividend does not remove protection, but it does show priorities

On March 25, 2026, the board approved an $8.495 million dividend to the shareholder. The shareholder said it intends to use the proceeds to increase the subordinated shareholder loan. On first look, that may seem neutral for bondholders because the money goes out and comes back in a junior form. But analytically, it sends an important message: profits are not automatically retained as common equity. They continue to move upstream and then return only if the sponsor decides to keep them inside the structure through a subordinated layer.

Extensions and repayments are the real 2026 trigger

According to the investment schedule, about $247.9 million of the portfolio contractually matures in 2026, another $178.7 million in 2027, and $44.5 million in 2028. At the same time, $129 million of loans had already been extended from time to time, about $10.8 million was extended during the fourth quarter, and after the balance-sheet date about $15 million was extended for the first time. That is not automatically a weakness, because extensions are part of bridge-loan economics, but it does mean 2026 will be a year of collections, renewals, and extensions rather than a simple year of volume growth.

When readers see an average duration of 13.2 months, it is easy to assume the book turns quickly and cleanly. But more than half the book matures contractually in 2026, and inside that number there is already a meaningful extension layer. So the next set of reports will need to show not only that loans still exist, but how they resolved in practice.

Efficiency, Profitability And Competition

Growth came from volume, not pricing

Interest income rose to $39.6 million in 2025, up from $35.7 million in 2024 and $34.5 million in 2023. That is a 10.9% increase versus 2024, but it did not come from sharper pricing. In fact, the average coupon charged to borrowers fell to 9.6% from 10.3% in 2024 and 10.8% in 2023. The story here is book growth, not price expansion.

The chart shows the move clearly. The company gave up some price and bought volume. As long as credit quality holds, that is a legitimate trade. But it also means future economics become more sensitive to funding costs because the asset-side pricing cushion is already less generous than it used to be.

The spread held, but it did not widen

Interest expense rose to $20.6 million from $15.9 million in 2024. At the same time, net investment income came in at $17.2 million, almost unchanged from $17.1 million in 2024 and slightly below $17.5 million in 2023. That is a story of operational stability, not of sharp margin expansion.

So 2025 was not a breakout year for bottom-line profitability. It was a year of building the public funding platform and expanding the balance sheet. That matters, because anyone looking for a fast-rising earnings engine will find that the company is still mainly proving it can scale the book without breaking its discipline. That is an achievement, but it is not the same thing as a structurally stronger earnings profile.

Competition helped, and it may also come back

The company describes a market in which banks and traditional lenders tightened credit after rate hikes, reducing funding availability and increasing the role of alternative bridge lenders. That is a supportive backdrop for a lender with speed, broker access, and underwriting experience. But the same report also notes that potential rate cuts in 2026 and 2027 may increase investment activity and support markets. That is not only a tailwind. It also opens the door to stronger bank competition.

That means Castellan’s advantage is not simply its cost of capital. It is primarily speed, disciplined underwriting, and distribution access. The group says it has reviewed more than 6,200 transactions over time and funded only about 272 loans, while also highlighting a no-principal-loss history across the platform. That is a real operating moat, but it will need to keep working even if markets become less stressed and borrowers regain more alternatives.

Reported profit includes some fair-value support, but this is still not the main story

In 2025, the company booked $1.174 million of unrealized investment gains after unrealized losses in 2023 and 2024. That is not the main earnings driver, which is why the year still looks mostly like a spread-income year. But it is important not to ignore the accounting model. Castellan operates as an investment company under U.S. GAAP, and the portfolio is marked at Level 3 fair value. The auditor identified fair value of investments as a key audit matter, and a 1% increase in the discount rate would cut portfolio value by about $6.1 million.

That does not threaten the company’s existence, but it is a useful reminder that the protection story is not only loan principal versus debt. Part of the balance sheet also rests on valuation judgment.

Cash Flow, Debt And Capital Structure

The right framework here is all-in cash flexibility

For Castellan, there is little value in discussing “normalized” cash generation the way one would for an industrial or real-estate operating company, because new loan originations are the business itself. The right cash lens is therefore all-in cash flexibility, meaning how much cash remains after the period’s actual uses.

On that basis, 2025 was a cash-consuming year. Consolidated operating cash flow was negative $106.2 million. The main reason was a $121.4 million net movement in investments, alongside a $9.2 million increase in restricted deposits. That was almost fully offset by $106.7 million generated from financing activities, including bond issuance proceeds, higher lender funding, and the subordinated shareholder loan.

That is not automatically a warning signal. It is how the model works. But the implication is clear: the book’s growth did not create self-funded flexibility. It was financed almost one-for-one by the debt market and senior lenders.

Unrestricted cash at the parent is small, and the cushion sits in the structure

At December 31, 2025, the consolidated company had $819 thousand of cash, $9.155 million of restricted deposits, and $2.739 million of interest receivable. Of those restricted deposits, about $4.155 million was already earmarked for the bond interest payment made on January 1, 2026, and another $5 million was posted as collateral for FX hedging. The parent company itself had only $163 thousand of cash.

That is a material point. Anyone looking only at covenant capital of $149 million may imagine a deep cash cushion. In reality, most of the strength sits inside the loan book, the collateral structure, the funding lines, and the subordinated sponsor loan. That is a big difference between “there is protection” and “there is free cash.”

The funding lines look comfortable, but 91% of assets are already pledged

On the positive side, the company reports five funding lines totaling $532 million, of which $189.2 million was drawn and $342.8 million remained undrawn. All of the lines are non-recourse to the company and are secured by ring-fenced loan pools. The report also emphasizes gradual paydown mechanisms that reduce refinancing risk if a line is not renewed.

The other side is that almost the entire balance sheet is already tied up. Assets pledged amounted to $428.4 million, or 91% of total assets, leaving only $42.7 million unencumbered. In other words, growth capacity still exists, but the layer of unpledged flexibility is already fairly small.

This chart makes the point quickly. Most of the remaining flexibility sits with Churchill, where the line is still barely used. That is a positive. It also means the next phase of book growth depends heavily on one large line that has not yet been meaningfully tested.

Covenant room is comfortable, but the cushion is more about avoiding breach than about funding independence

For bond purposes, the company reported covenant capital of $149.0 million, net financial debt to net CAP of about 68%, and loan-to-collateral ratio in the pledged entity of about 72% at year-end 2025. That leaves $76.5 million of room over the hard capital floor of $72.5 million, 9.5 percentage points below the 77.5% leverage ceiling, and 13 percentage points below the 85% collateral ceiling.

But there is also a tighter set of interest step-up thresholds, and the room there is smaller: only $54 million above the $95 million capital threshold, 4.5 percentage points below the 72.5% leverage threshold, and 10.5 percentage points below the 82.5% pledged-entity collateral threshold.

So the company is clearly far from default, but not so far that funding discipline can be ignored. That matters even more because the cushion is calculated with a capital definition that includes subordinated sponsor debt, not only common equity.

Outlook And Forward View

Four non-obvious findings should shape the 2026 read:

1. Bond protection is stronger than common equity alone. Covenant capital of $149 million sounds robust, but $39.2 million of that amount is a subordinated shareholder loan.
2. The short duration is less clean than it looks. About $129 million of loans have already been extended from time to time, and another $15 million was extended for the first time after the balance-sheet date.
3. The real cash cushion sits in facilities, not in the parent’s cash box. The parent had only $163 thousand of cash, while flexibility came mainly from undrawn lines.
4. The conservative screen also relies on estimates. Average LTV is based on property values at loan origination, and the portfolio is carried at Level 3 fair value.

The right label for 2026 is a funding proof year, not an underwriting proof year. There is no current sign that the company has lost control of the credit book. There is also no sign of principal losses. But several things still need to happen in the right way for the story to remain clean.

First test: refinance without simply hiding risk through extensions

About 52.6% of the portfolio matures contractually in 2026. That is large enough to move the liquidity picture quickly, but also large enough to hide behind extensions. So the next report will need to show not only that loans are still there, but what actually happened: how much was repaid, how much was extended, how much was sold, and on what terms.

The company already discloses that, as of the reporting date, one loan of about $8 million is past due on principal because it reached maturity, even though interest payments are still being made by the mezzanine lender. That is only 1.7% of the book, so it is far from a systemic stress signal. But it is still a reminder that Castellan’s short duration is not a frictionless chain of repayments. Some of the book moves forward in time.

Second test: expand the book without diluting protection quality

The company also had unfunded loan commitments of $137.8 million, nearly 29.3% of the current book. That is growth potential. But it also means that even if funding lines look open today, the company will need to prove it can keep the same standards on LTV, duration, and pricing in the next growth leg.

The challenge is two-sided. On one hand, tighter bank credit still supports alternative lenders. On the other hand, if U.S. rates continue to fall, competition can intensify again. So 2026 will test not only whether Castellan can grow, but whether it can remain selective while markets change.

Third test: protect profitability without leaning too hard on capital structure engineering

The average borrower coupon fell to 9.6%, while interest expense rose to $20.6 million. True, the average cost of funding sources fell to 7.78%, but the overall spread cushion is already less generous than in prior years. If the company keeps paying dividends and then reintroducing capital through subordinated sponsor loans, covenant capital may still hold up, but ordinary equity quality will not necessarily strengthen.

That is a meaningful difference between credit stability and capital quality. The bonds are better protected when there is a junior sponsor layer, but the balance sheet would be cleaner if more earnings simply remained as common equity instead of moving up and then returning in subordinated form.

Fourth test: prove that the cushion is not only LTV arithmetic

The presentation highlights about $961 million of pledged real-estate value and a $649 million “margin of safety” based on average LTV of about 49%. That sounds impressive, but it is important to understand what sits behind the number. The average LTV is based on property values at origination, not on a live market mark. At the same time, portfolio fair value is based on discounted cash flows with discount rates of 8% to 14%.

That is why 2026 will need to give the market more than LTV tables. It will need to show repayments, refinancings, and perhaps loan sales if stressed cases appear. Real protection is measured by the ability to turn collateral into cash, not only by the ability to show a comfortable ratio on paper.

Risks

Risk one: extensions are part of the model, but they can also become the slow way quality drifts

About $129 million of loans had already been extended from time to time, and another $15 million was extended for the first time after the balance-sheet date. That is not automatically a negative signal. Bridge loans move. But as the extended layer grows, average duration stops being a sufficient risk descriptor. What matters is how many such loans return to orderly repayment and how many start sliding toward principal delinquency, like the $8 million loan already past maturity.

Risk two: concentration is moderate rather than extreme, but it is very real

The company has no single borrower representing 10% or more of the portfolio. That is positive. Even so, the top ten borrowers accounted for 43.3% of the book and 37.4% of interest income at the end of 2025. In addition, one borrower with three loans represented about 6.8% of the portfolio, or $32.2 million of combined principal, all in California. That is not extreme dependence, but it is not full diversification either.

Risk three: dependence on a related-party manager and a structure with no employees

The issuer itself has no employees at all. It relies on a management company related to the controlling shareholder for underwriting, loan allocation, asset management, collections, and enforcement. Management-fee expense was small in 2025 at only $116 thousand, but that is not the main point. The main point is that the public credit wrapper depends almost entirely on a related-party platform for expertise, people, and infrastructure.

That works as long as incentives remain aligned. It also creates a built-in governance dependency because there is no truly stand-alone public operating platform here.

Risk four: with 91% of assets pledged, there is not much room for error

The company had only $42.7 million of unencumbered assets against $269.3 million of liabilities secured by charges over company assets. That still looks manageable, especially with $343 million of unused facilities. But it means the room for mistakes does not sit in a large pool of unpledged collateral. It sits in execution.

Risk five: the legal and tax structure is more complex than a standard domestic bond issuer

The company is a BVI entity, its owner is a U.S. REIT, and the issuer is supposed to remain a pass-through entity for U.S. federal tax purposes. The report also notes that, in an insolvency scenario, there could be legal frictions across Israel, the BVI, and the U.S. That is not a day-to-day operating risk, but it is a real structural layer that bondholders should keep in mind when they think about documentation and enforcement quality.

Conclusions

Castellan finished 2025 with what most bond issuers would like to show: a larger loan book, low average LTV, open funding lines, and covenant headroom. That is the supportive side of the story. The main friction is that the whole system still depends heavily on external funding, on assets that are already mostly pledged, and on a capital base that includes a subordinated sponsor layer rather than only common equity. That is what will drive the market read in 2026.

Current thesis in one line: Castellan currently looks more like a disciplined credit issuer with strong underwriting than like a pure credit-risk story, but 2026 will test whether that conservatism translates into funding flexibility and not only into a comfortable LTV profile.

What changed versus the simple read is not the quality of the loan book. It is the quality of the protection layer. After the bond issuance, funding is more diversified, but the capital structure is also more complex. The strongest counter-thesis is that the market is being too cautious because Castellan still shows 49.7% average LTV, $343 million of unused facilities, $43 million of unencumbered loans, and a no-principal-loss history across the platform. That is a fair argument. It becomes more convincing if the company proves in 2026 that maturities and extensions resolve cleanly, that dividends do not turn into a recurring capital-extraction loop, and that funding discipline stays intact while the book expands.

What could change the market reading in the short to medium term is the combination of three things: repayment and extension behavior inside the 2026 maturity wall, the ability to keep funding costs under control as rates change, and continued clean covenant compliance even if dividends continue. If the company shows that extended loans remain under control, that the past-due principal case is resolved without loss, and that covenant headroom is preserved without repeated capital engineering through the sponsor, the read improves. If the extension layer grows, funding usage rises faster than ordinary equity, or fair-value pressure starts to build, the market will become more cautious quickly.

Why this matters: Castellan is proving, at least for now, that a conservative loan book is not enough by itself. In this market, company quality is defined by whether conservative credit can be held inside an equally conservative funding structure.