Not All Permanent Capital Is Permanent

The private-credit redemption scare of 2026 is doing the alternative asset managers a favour: it is teaching the market to price the difference between capital that is locked and capital that merely looks locked.

For most of the last decade the bull case for the listed alternative asset managers reduced to a single word: permanent. Fee-paying assets that did not have to be re-raised every few years would compound; fee-related earnings would become subscription-like; the multiple would re-rate from private-equity-cyclical toward asset-manager-durable. The thesis was right often enough that the entire group leaned into it, marketing perpetual-capital ratios the way a software company markets net revenue retention.

The early months of 2026 have tested that word, and the test is clarifying. When Blue Owl moved to limit withdrawals at two of its retail credit vehicles after redemption requests reached roughly a fifth and two-fifths of shares respectively, the read-through was not that perpetual capital is a myth. It is that the label has been applied to two structurally different things, and the market had been paying for both as if they were the same. We think the distinction between them is now the most important variable in the group, and most of what follows is an attempt to make it legible.

Two kinds of “permanent”

Begin with the accounting fiction the word papers over. A perpetual, non-traded credit fund sold through wirehouses carries no fixed term, so it is reported as permanent capital. But it offers periodic liquidity — quarterly tenders, typically capped at five per cent of net asset value — and that liquidity feature is precisely what made it sellable to private wealth in the first place. The capital is permanent only so long as redemptions stay below the cap. Cross the cap, and the manager gates; gate, and fundraising into the vehicle stops; stop fundraising, and the fee base that was supposed to compound instead shrinks. The permanence was conditional all along.

Now contrast an annuity liability sitting on an insurance balance sheet. The policyholder cannot demand the money back at par on a Tuesday because they have read a worrying headline; surrender is contractual, penalised, and spread over years. The asset manager earns a spread on the float for the life of the liability regardless of sentiment. This is permanent capital in the strong sense — it is permanent because someone else has contractually given up the right to redeem it. Apollo built its franchise on exactly this insight with Athene; KKR followed by taking Global Atlantic to full ownership; Ares is building Aspida off a smaller base.

The 2026 sell-off priced this difference in real time. The names most exposed to retail semi-liquid credit fell hardest, while the insurance-levered platforms, though not spared the sympathy move, held their fee narrative intact. The market, in other words, did the work of separating the two kinds of permanence that the income statements had blurred.

Measuring durability twice

A label that conceals this much should be replaced by measurement. We score fee durability along two axes that, read together, are harder to game than any single perpetual-capital ratio. The accompanying Alt-AM Signal Dashboard renders both interactively; here we set out the logic.

Score one — where the capital is locked

The first axis is the familiar one: perpetual and permanent capital as a share of fee-paying assets. A higher share means less of the fee base resets each fundraising cycle, and on its own it is a reasonable proxy for resilience. Plotted against the multiple the market pays on fee-related earnings, it should slope upward — durability earning a premium. The instructive exception is the name that breaks the slope. Blue Owl carries the highest perpetual share in the group and trades at the lowest fee multiple, because the dashboard tags its perpetual base as redeemable retail credit rather than insurance liability. The score and the share price disagree, and the disagreement is the whole point: the ratio is necessary but not sufficient.

Score two — how much of the earnings is recurring

The second axis is the complement, and it is where the realisation cycle becomes visible. Distributable earnings exceed fee-related earnings by the amount of carried interest and realised performance the manager books in a period. That gap — distributable earnings minus fee-related earnings — is the most cyclical dollar in the business: it depends on exits, which depend on markets, which do not cooperate on a schedule. A manager whose distributable earnings lean heavily on that gap is, by construction, cheaper to own through a cycle and should be priced accordingly. We fold capital durability, earnings recurrence and fee-related margin into a single composite quality score so the two axes can be read against valuation at once.

*OWL scores highest on the mechanical composite precisely because the score cannot yet see redemption risk in the perpetual base — the limitation the dashboard’s capital-type tagging is designed to correct. Figures FY2025 reported; HLI omitted (pure advisory, no AUM). Carry-dependence and composite are analytical estimates.

What the FY2025 numbers actually said

It is worth stressing that the redemption scare arrived on top of a genuinely strong fundamental year. This was not a group missing estimates. Fee-related earnings grew across nearly every name in FY2025 — Blackstone, Apollo and Blue Owl in the mid-to-high teens, Ares and TPG faster, Carlyle to a record on a record fee-related margin. The growth was real; what changed in 2026 was the market’s willingness to capitalise it at the old multiple, once the quality of the capital behind it came into question.

That is the right way to hold the two facts together. The earnings were good. The re-rating is about durability, not about a stumble in the numbers. An investor who conflates the two — who reads the sell-off as a verdict on FY2025 results — will misjudge both the names that deserve it and the names that do not.

Recent trends — what moved this quarter

This is the section we expect to carry forward each period, because the texture of the cycle changes faster than the structure. Four developments defined the last few months, and each one feeds the durability question rather than displacing it.

Request is not redemption

The single most useful distinction in the BDC data is between what investors asked for and what they received. The five-per-cent quarterly cap means a fund can face an enormous request and honour only a sliver — and the gap between the two now separates the franchises from the gates. Blue Owl’s tech-focused OTIC received requests equal to roughly 41% of shares and fulfilled the 5% cap, paying out something like twelve to fourteen cents on each dollar tendered; OCIC saw 22% and paid about a fifth of what was asked. Apollo’s Debt Solutions fund honoured close to 45 cents on the dollar; Blackstone’s BCRED did the rare thing of upsizing its cap to 7% and meeting requests in full, helped by the firm and its employees injecting roughly $400m of their own capital. Goldman’s vehicle came in at 4.999%, a hair below the cap — the only major non-traded BDC that did not have to ration at all. The Cliffwater direct-lending index shows the industry redemption rate jumping to about 4.8% in the fourth quarter of 2025 from 1.6% a quarter earlier. Who could pay, and who had to gate, is the cleanest read on franchise quality the cycle has offered.

The software question — and the disclosure gap

Underlying the redemptions is a sector worry: software, which Morgan Stanley estimates at roughly a quarter of all BDC exposure, and where the firm warns direct-lending defaults could climb toward 8% against a 2–2.5% historical norm. The sharper point, surfaced by a Wall Street Journal analysis of four flagship funds, is a reporting gap. Funds classified about 19% of their books as software on average; the Journal put the true figure nearer 25%. Blackstone’s BCRED carries the highest absolute exposure — reported around 26%, estimated above 30%. Blue Owl’s OCIC shows the widest discrepancy, reported near 12% against an estimated figure roughly double that. Apollo’s fund sits lowest, in the mid-teens on either measure. The instinct that Blue Owl looks most exposed is half right: it is not the largest absolute software book, but it is the largest gap between what was disclosed and what the loans actually are, and in a confidence-driven channel the gap is the risk.

Direct lending is losing its premium

The yield backdrop reinforces the same story from a different angle. Through 2025 the premium of direct-lending takeout yields over broadly syndicated loans compressed — from roughly 244 basis points toward the high-100s on LSEG data — as competition intensified and borrowers refinanced direct loans back into the cheaper syndicated market. Direct lending still out-yields syndicated loans, high-yield and investment grade in absolute terms (a Cliffwater takeout yield in the low elevens against syndicated loans in the low nines), but the marginal edge that justified the asset class’s fundraising boom has been narrowing, and only began re-widening as credit stress returned in 2026. A shrinking illiquidity premium and a rising redemption rate are an awkward pair: investors are being paid less to hold the least liquid version of the same credit.

The fundraising mix has tilted to real assets

Where new capital is being raised matters as much as where it is leaving. Capital formation was the weak link of an otherwise strong 2025 — the softest fundraising year since 2020, with US commingled buyout vehicles down sharply. But the mix shifted decisively: real-estate debt and opportunistic strategies, infrastructure, and secondaries all grew, while traditional private-capital buyout funds shrank. For a group that spent the prior cycle marketing private credit, the forward growth areas are increasingly real assets and credit-adjacent structures — which is exactly where the recent deal activity points too. GCP International went to Ares for real estate and digital infrastructure; Peppertree took TPG into digital infrastructure; BlackRock’s GIP and Preqin deals were about infrastructure scale and data. Meanwhile the realisation engine is restarting: global exit value rose more than 80% year-on-year through the first nine months of 2025, third-quarter buyout value was the strongest since the 2021 peak, and a reopening IPO window should feed the carry line through 2026. That recovery is the cyclical tailwind sitting above the fee base — good for distributable earnings, but a reminder that the carry-dependent names are levered to a cycle that has only just turned.

Where that leaves the group

On a quality-adjusted basis the ranking is unusually clean. The insurance-linked compounders sit at the top: their permanent capital is the contractual kind, their spread earnings behave like an annuity, and — in Apollo’s case — the market is paying the cheapest large-cap fee multiple in the group for the privilege. We would rather own durability that is mispriced than durability that is fully priced, and that is the case for the spread platforms over the premium-multiple growth names.

Blue Owl is the sharpest disagreement between our score and the tape, and the honest answer is that both are partly right. The franchise quality is real; the perpetual share is genuine; the fee margin is healthy. But none of that resolves until the retail credit base stabilises, and until it does the cheapness is a function of redemption risk rather than an entry point. It is a falling knife with good fundamentals, which is a more dangerous object than a falling knife with bad ones, because the fundamentals invite you to catch it.

At the other end, Carlyle and StepStone screen lowest on durability — the former on classic private-equity carry dependence even after a record fee year, the latter on a thin perpetual base. Neither is a name we would reach for on valuation alone; cheap and low-quality is the value-trap signature, not the value signal.