The exit from private credit evergreens now has a public price

Public sources retrieved between October 1 and 4, 2026

September put a public price on leaving a private credit evergreen fund. A third-party vehicle, Cox Capital Retail Secondaries Fund I, bid for Class I shares of Blackstone Private Credit Fund, Apollo Debt Solutions BDC and Ares Strategic Income Fund at 12.5% to 14.6% below their August 31 NAV. Blackstone and Apollo had published prorated tenders in August, weeks before the September bids, and Ares Strategic Income Fund filed its own proration the day the offers opened. Busy month in evergreen territory.

On Apollo, the discount halved in 2 months, from 30% to 14.6%, while the fund’s own queue barely shortened. We think the bid prices the wait, and the managers’ letters say the wait is getting shorter. Across the 3 funds, the September bids pay c. 5 to 6 points of NAV per quarter of waiting, and on Apollo the price of a quarter fell by 44% while the queue shortened by 12% (section 4). The view fails if Cox’s offers buy little by November 14, or if final Q3 results clear the queues while the discount holds.

None of August’s 4 expectations fell due; 2 received new facts (section 8).

No adviser published a new primary estimate in September, so the H1 range of $117bn (UBS) to $124bn (Lazard) still sets the scale. The month itself comes from the registries.

Formation picked up at the edges of the market. Prime Infra Secondaries II filed a Form D showing $748.7m sold to 11 investors, an average ticket of c. $68m, with Evercore as agent and no announcement we could find by October 1. Westwood Energy Secondaries VIII and BlueEarth Impact Secondaries II ($60m sold) add an energy and an impact sleeve. Our view is that secondaries capital now forms strategy by strategy, and that the specialist sleeves are multiplying faster than the flagships, even in the month CVC closed $10bn.

The amounts tell a narrower story than the count. The 14 vehicles declare c. $852m sold, against c. $539m for August’s 9, but 88% of it sits in Prime Infra alone and 9 vehicles report nothing sold yet. Most of the new sleeves are still shells waiting for a first close, and the Form D amendments of the next months will show which ones fill up.

And it goes without saying that CVs keep their momentum when it comes to new capital (see §3 below).

These counts come from selected name and text searches, which miss vehicles that avoid the search criteria and catch some unrelated entities. They are applied identically every month, which is what makes the comparison with August valid.

No adviser published a new GP-led / LP-led split in September; the H1 figures from Evercore ($65bn GP-led against $56bn LP-led) and Lazard ($61bn against $63bn) remain the reference.

The Cox offers fit neither box. A third party buys LP interests in a non-traded fund, without the manager’s consent, at a stated price, through a public offer at retail ticket size. Mechanically it is an LP-led trade, run as a tender, on shares whose holders mostly bought through a wealth platform. We expect this flow to grow and to land in the “other” columns of adviser reports, which is precisely what our August expectation on Campbell Lutyens’ unclassified share was about.

The listed vehicles are building their own exits too. Partners Group Private Equity put realisation shares to a vote, with a full realisation of the portfolio if more than 40% of holders opt in (EGM on October 7). HVPE announced a $400m tender expected at c. 10% below NAV, and NBPE exited Solace in full and Osaic in part through continuation vehicles, a GP-led used to return cash to a listed LP.

The managers also changed the meaning of their own demand numbers. Apollo estimates that “the vast majority” of Q3 repurchase requests at ADS are re-tenders of unfilled requests from prior quarters, and Blackstone says a significant portion of BCRED’s unserved Q2 requests came back in Q3. A queue counted in holders would look shorter than a queue counted in requests, and neither fund publishes the first.

7 continuation vehicles appear in September’s selected tape with a primary source: 2 hold loans, and 5 hold a single company.

September’s 2 credit vehicles come on top of August’s 2 (BlackRock TCP and Willow Tree), which makes 4 public credit continuation vehicles with a named lead in 2 months. Pantheon led 3 of them (BlackRock TCP, Bridgepoint Credit, PennantPark; HarbourVest led Willow Tree). That’s a huge market share for Pantheon, and that makes one buyer the reference price-setter for an entire segment. Credit continuation vehicles are being described as a market, but a market with one dominant lead is closer to a relationship, and we would be careful with any “market price” quoted for them in the next adviser reports.

Carlyle AlpInvest published a paper on credit secondaries on September 21 that describes this segment from the buyer’s seat: a small market with few scaled buyers, capital concentrated in a handful of visible, intermediated deals, and buyers “comfortably clubbed up” on many of them. It still concludes that the market is not too crowded. The public tape supports the description and sits less comfortably with the conclusion, since Pantheon led the 2 largest of the 4 vehicles (€1.2bn and $745m) and PGIM came in alongside it on PennantPark.

The paper’s diligence list is also a useful mirror for what these releases leave out. It asks buyers to underwrite the effective discount, which can differ from the headline price, and to check whether the strongest credits stay with the manager while the weaker ones go to the vehicle. Of the 4 public credit vehicles, 1 published a price (BlackRock TCP, c. 95% of fair value, because a listed BDC has to file an 8-K) and none published which loans stayed behind.

Both September portfolios are described as senior secured or mature (Bridgepoint, PennantPark), which is where the paper says risk is best taken at this point of the cycle. Buyers are at least doing what they write.

The registries keep running ahead of the announcements:

The month produced 3 price regimes for the same question, how much it costs to leave early.

One comparison in this table deserves a second look. A listed PE share that can be sold any day trades at 22% to 29% below NAV, while a gated credit share draws a bid at 12.5% to 14.6%. Daily liquidity is priced below a place in a queue, which tells us the listed discount carries a doubt about NAV and duration that the credit bid does not carry. This does make sense as a principle, and we’ll keep monitoring this gap.

The Cox offers in detail:

The 3 offers let us price the wait itself. If each quarter keeps serving the same share of requests as Q3, a holder needs requested / served quarters to get out. Dividing each discount by that wait gives what the bid pays per quarter skipped.

Across 3 funds run by 3 managers, the September bids pay c. 5 to 6 points of NAV per quarter of waiting. On ADS, the queue shortened by 12% between the 2 offers while the price of each quarter fell by 44%. A discount that lines up this closely with the length of each line is hard to explain by credit quality, since the 3 portfolios differ, and it is what a tariff on time would look like.

On BCRED, the 12.5% discount equals c. 16 months of the fund’s 9.2% annualised Class I distribution, to skip a queue the board says pays c. 75% of requested capital at NAV within about 90 days. The board’s 8-K also writes the arbitrage down: a buyer who acquires shares below NAV and tenders them back at NAV keeps the discount. That makes the Cox bid a bet on the duration of the queue, and the narrowing on Apollo, from 30% to 14.6% in 2 months, says the bidder now expects that queue to clear faster.

The headline caps understate the bid. Cox reserves the right to buy up to an additional 2% of each fund’s Class I shares without extending the offers, c. $191m for ADS and c. $141m for ASIF, which takes the 2 offers to c. $357m, or c. 14x the $25m in the release. Last month, the PR size of a credit continuation vehicle overstated its equity cheque by a factor of 3, and the Cox release errs the other way on capacity, which only the offer document shows.

The price is also set at the very edge of the line: the headline caps come to c. 1% of what each fund left unserved in the quarter.

Event ticketing ran the same experiment long ago. When the issuer fixes the price and rations the quantity, a secondary market forms to price the ration, above face value when seats are scarce. Here the scarce thing is the exit, so the secondary price sits below NAV, and it will move with the length of the line.

The same paper gives an example that the public record happens to illustrate: a portfolio bought at 95 can be a better investment than one bought at 88, depending on cash yield, fees and marks. Both prices now exist in public for senior direct lending books. BlackRock TCP’s continuation vehicle changed hands at c. 95% of fair value in August, with a lead buyer and a process, and Cox bids c. 85 to 87.5 for retail-held shares of 3 funds that own the same kind of loans.

The retail holder is quoted c. 8 to 10 points less than the listed BDC obtained. The loans are comparable, so we attribute most of that gap to the wrapper, the ticket size and the absence of a process, and we expect it to narrow as more bidders show up. Valuation dates, leverage and fees differ between the 2 observations.

A correction on our side: the August edition said we found no public bid for shares of the prorated credit vehicles. Cox’s release shows its July offers, at 30% below NAV for ADS and 15% for ASIF, were open until August 24. We missed them. One bidder with a $25m headline remains a thin basis for a market price, although a thin price still carries more information than the NAV-or-wait choice holders had in June.

Selected announced capital in September:

Credit secondaries capital keeps forming in the registries: ASPF III - Credit Secondaries, the AlpInvest sidecar, filed alongside AlpInvest Secondaries Fund IX on September 29; Dorchester Credit Secondaries VI reported $29.7m sold since September 9 (a new notice for an entity formed in 2022); IEQ - FoxPath Credit Secondaries, an iCapital feeder, and PBN III Equity Holdings, linked to Coller’s credit secondaries manager, filed with nothing sold yet. HarbourVest hired Tony Ranaldi to lead its private credit evergreen strategy, credit secondaries included. The 4 public credit continuation vehicles of the last 2 months give that capital a deal flow, and the prorated evergreens may give it another one.

AlpInvest’s own timing is worth a note: the credit secondaries sidecar was filed 8 days after the same paper. The paper names where it expects supply to come from, which is liquidity pressure at BDCs and evergreen vehicles, and puts redemption requests at the 12 largest non-traded BDCs at c. 12% of NAV in Q1 and c. 13% in Q2 (its figures, drawn from third-party data). The 3 Q3 prints in this month’s filings, c. 10% at BCRED, 13.1% at ASIF and c. 14.7% at ADS, average 12.6%. The buyers’ research and the funds’ own 8-Ks now describe the same sellers, with the same number.

The paper also argues that information in credit secondaries is limited and that proprietary data is the edge. For the retail end of this market, September says otherwise: the length of each queue, the managers’ estimates of re-tenders, the distribution rate and the third-party bid were all filed or published within the month. What stays scarce is someone putting them side by side.

The wealth channel was sold the entry and the exit in the same month. Blackstone launched BXPM, ICG and Amundi completed their partnership with a first joint evergreen in PE secondaries, and Ardian launched an infrastructure ELTIF, while the credit evergreens next door rationed redemptions and drew a discounted bid. Another signal that retail is here for the long run, and that its exit terms are now part of the product.

The intermediaries kept hiring for this flow: Houlihan Lokey (GP-led), Union Square Advisors (secondaries and GP solutions), Debevoise and A&O Shearman (secondaries teams), Adams Street (venture secondaries) and Golub, which named a head for Golub Equity Continuation Partners.

Obviously an exciting time to be in this space.

A public slice of September: quiet processes, and deals reported only behind paywalls, sit outside it.

Private credit funds that prorated, or announced a proration, in September:

Acceptance = shares accepted / shares tendered; demand = shares tendered / shares accepted. Unserved requests at Ares Strategic Income Fund come to c. $0.83bn (amount paid × (tendered / accepted - 1), a floor); at BCRED, the letter implies c. $2.15bn.

Compared with Q2, the queues moved little: BCRED from 48.5% to c. 50%, Apollo from 29.8% to c. 34%. That is the gap the Cox bid is pricing, and the gap that narrowed by 15 points on Apollo’s discount over the same weeks.

The managers’ own statements let us rebuild the line quarter by quarter. BCRED’s c. 75% of requested capital within about 90 days is 2 rounds at 50% (1 - 0.5 x 0.5). Apollo’s c. 75% over 3 quarters, with 29.8% accepted in Q2 and c. 34% in Q3, implies c. 46% in Q1: the ADS queue lengthened sharply in Q2 and has eased since, which is the path Cox’s discount followed.

Apollo’s update also gives the other side of the ledger. Its 5% served is c. $0.7bn, so the c. 14.7% requested is c. $2.1bn, against c. $0.2bn of gross subscriptions in the quarter. New money covered c. 10% of the requests and c. 29% of what was paid out.

Rationing is a fund-level fact. Fidelity Private Credit Fund, New Mountain Private Credit Fund, Oaktree Strategic Credit Fund, Antares Private Credit Fund and Goldman Sachs Private Credit Corp. served every request or reported demand under the cap in the same quarter. Morgan Stanley’s 2 North Haven funds, with the same strategy, accepted 43.8% and 73.3% of shares tendered. The filings do not say why; the distribution channel is the obvious suspect, and it is the variable none of these documents report.

PE evergreens show no queue at all. All 9 results were served in full or drew no tenders, and Hamilton Lane Private Assets Fund received requests worth c. $57m of NAV against a cap of c. $323m, 17.7% of its capacity.

The strongest objection is the double count. When holders expect proration, they tender more than they intend to sell, and unserved shares come back the next quarter, which both managers now confirm. The multiples overstate the number of people waiting. They still show a line, and Cox’s pricing confirms someone is willing to pay to stand in it.

Last month’s calls. None fell due by October 4. 2 received new facts:

New expectations.

Also on the hot list for October: HVPE’s tender terms with its interim results; the outcome of Partners Group Private Equity’s EGM on October 7; the expiry of the Cox offers on ADS and ASIF on November 14; the Fidelity Yoshikawa Data Center Continuation Fund, still unannounced a month after its GLEIF registration.

This Monitor uses public sources only, retrieved between October 1 and 4, 2026: SEC EDGAR filings (Form D, SC TO-I, SC TO-I/A, 8-K), GLEIF records, regulatory announcements of listed vehicles, issuer, buyer, seller and law-firm announcements, public adviser reports and other public documents gathered from Axorion’s research framework. Paid sources are excluded as per Axorion’s no-paid-source policy; transactions reported only by paid outlets stay out of the tape until a public primary source confirms them, and paywalls are not considered as sources.

Registry counts rely on name and full-text searches and are proxies. The segment of each evergreen fund cited follows the investment objective stated in its latest annual, quarterly or semi-annual report. BCRED and ADS Q3 figures are the managers’ estimates ahead of final results. Cox’s offers are offers, and their outcome is unknown. Listed-vehicle discounts are those each vehicle publishes, at different dates. Assets of the unannounced continuation vehicles are inferred from their names. The expected wait in section 4 is requested / served and assumes the Q3 acceptance rate holds; it ignores the distributions earned while waiting and the early repurchase deduction BCRED applies to shares outstanding for less than 1 year. ADS’s unserved requests are derived from Apollo’s own figures. Carlyle AlpInvest’s figures are its own, drawn from third-party data it cites, and are used for comparison only. Figures computed by Axorion show their formula. Quiet processes stay outside the record, and so does the remaining capacity inside any buyer’s mandate.

Axorion Liquidity Monitor · September 2026 · source publication dates apply