The mechanics of the sponsor model
A private equity fund is a limited partnership. The limited partners — pensions, endowments, insurers, increasingly individuals — put up nearly all the money and get liability capped at what they committed. The general partner, the sponsor, makes every decision. The classic economics are "2 and 20": roughly a 2% annual management fee on committed capital, and 20% of the profits once the LPs have their money back and cleared a preferred return, commonly around 8%. The fund has a finite life, typically about ten years, governed by a limited partnership agreement that sets , fees, the distribution waterfall and what LPs are allowed to know. In practice the sponsor is three entities stacked together: the fund partnership that owns the companies, the GP that has the authority, and a management company that employs the people and collects the fee.
The purchases are made mostly with debt. KKR's 1989 buyout of RJR Nabisco is the deal that fixed the image in the public mind — $109 a share, about $25bn, roughly $31bn including assumed debt, financed overwhelmingly by borrowing and serviced out of the target's own cash flow and asset sales. The company was broken up over the following years. The economics for KKR's investors were, by the post-mortems, dismal: an IRR reported well under 1%, with KKR finally selling out in 1995. The template survived the returns.
The 2006-07 vintage tested it at scale. Blackstone bought Hilton for $26bn in July 2007 and got its timing rewarded. KKR, TPG and Goldman Sachs closed the $45bn TXU buyout in October 2007 and got the largest LBO bust in history: shale gas collapsed wholesale power prices, the cash flow that was meant to carry the debt did not arrive, and roughly $8bn of sponsor equity went to zero. Both outcomes came from the same structure. Leverage does not have a view.
The exit drought and the backlog
The part of the model that returns money to LPs is realisation — sell the company, distribute the cash. That mechanism has been broken for several years, and the inventory is visible.
Global exits peaked in 2021 at $1,689bn across 4,482 deals. In 2025 they were $1,248bn. McKinsey's 2026 private markets work puts PE exits down 52% against 2021. First-half 2026 realisations came in around $293.7bn globally, with US exit value of $273.1bn and 353 US exits in the second quarter alone worth $102.6bn.
What that does to LPs shows up in the distribution yield: 14.8% of NAV as of the mid-2026 readings, against a long-run average of 23.4%. One count has distributions running 11% to 13% of starting NAV in each of the last three calendar years, roughly 10 to 15 percentage points below the 25-year average since the 2021 vintage.
Meanwhile the unsold pile grows. US inventory was about 12,000 PE-backed companies at the end of 2024 and 13,509 by the second quarter of 2026. Capital waiting to be spent is still enormous — PitchBook counted $4.63tn of dry powder across closed-end private capital funds globally at the end of Q2 2025 — but it is ageing. McKinsey found the share of dry powder two years or older hit a new peak of around 40% in June 2025. Sponsors are simultaneously unable to sell what they own and under pressure to deploy what they have raised.
Engineering liquidity
When the exit is closed, the sponsor's options narrow to two: borrow against the asset, or sell it to itself.
The borrowing route is the dividend recapitalisation. The portfolio company takes on new debt and pays the proceeds out to the fund as a dividend. The LPs get a distribution, the sponsor gets to show a return, and the leverage stays with the company. PitchBook counted $13.3bn of dividends funded by leveraged loans in 2026 through 31 July, against $23.8bn over the same period in 2025 and $19.1bn in 2024. That is a market running well below last year's pace, not a record. The channel is large and established; it is not currently expanding.
The selling-to-itself route is the continuation vehicle, and that one is booming. A sponsor moves one or more portfolio companies out of an ageing fund into a new vehicle it also manages, funded by secondaries buyers, with existing LPs given the choice to cash out or roll. Jefferies put continuation-vehicle volume at about $115bn in 2025 within a $240bn secondary market. Evercore's H1 2026 review counted about $121bn of total secondary volume, roughly $65bn of it GP-led, with single-asset continuation vehicles at about $34bn. GP-led transactions above $1bn rose to 29 in 2025 from 21 in 2024. Continuation vehicles now account for about 14% of sponsor-backed exits globally, up from 5% in 2021, with one count putting the range anywhere from one exit in eight to one in five. Pantheon co-led an €811m multi-asset continuation vehicle for Bregal Unternehmerkapital in July 2026, holding Safety21 and Onlineprinters. The conflict is structural and nobody pretends otherwise: the same firm is on both sides of a price. ILPA published proposed updated continuation vehicle guidance on 24 June 2026, with comments open to 5 August.
Then there is the collateral. After United Wholesale Mortgage listed, Mat Ishbia pledged most of his family's UWM equity against up to $1.8bn of JPMorgan loans used to help buy control of the Phoenix Suns and Mercury; a fifth loan in 2025 took the total principal to about $2.3bn, per a Michigan UCC filing. Filings also show the Suns holding entity pledging future dividends and distributions — and potential bankruptcy or insolvency proceeds — as collateral. UWM reported a $280m tax receivable agreement liability as of 30 June 2026, and rights to payments under that TRA were pledged as part of the collateral supporting a $2.05bn deal with Oaktree. held talks with Apollo in the summer of 2026 about a multibillion-dollar loan secured on his Los Angeles Lakers stake, and had separately proposed pledging his Guggenheim equity to raise short-term financing for at double-digit yields. He sold the Lakers stake instead. We wrote that up on 20 August.
Nobody is checking the homework
The SEC adopted Private Fund Adviser Rules in August 2023. On 5 June 2024 a three-judge panel of the Fifth Circuit vacated the entire set, unanimously. The SEC abandoned its review of the vacated rules that September. Whatever you think of the rules themselves, the practical position is that the disclosure regime governing continuation-vehicle pricing, fee offsets and side letters is the one the industry writes for itself in the LPA.
Why it matters to this crash
Sponsors are the demand side of . exists in its current size because sponsor-owned companies need debt that banks will not syndicate, and the load-bearing assumption of the whole asset class is that this debt can always be refinanced by the next fund with money to put out.
That assumption failed a test in August 2026. 's Sophos spent months trying to line up private credit for more than $2bn of loans. The effort faltered, and the company went back to its existing leveraged-loan holders offering a higher coupon, amortisation payments and a tighter package (Bloomberg). Sophos has a $2.1bn loan maturing in March 2027 and a $92.5m revolver due December 2026, and Moody's has cut it to B3 from B2. Thoma Bravo told lenders it would not inject fresh capital, rebuffing investors worried about what AI does to a cybersecurity business. We covered the reversal on 19 August and again later that day: the paper went back to the , the one place it gets daily. It was the third Thoma Bravo software credit in three months to need this treatment, after Medallia and Proofpoint.
That is the sponsor model's central asymmetry stated out loud. The sponsor's equity is the cushion under the debt, and nothing obliges the sponsor to top it up. When it declines, the loss stays with creditors, and the repricing happens through terms — amortisation, , coupon — rather than a default that would show up in a credit index.
The other reason sponsors matter here is that they are the beneficiaries of . SEC staff agreed with Latham & Watkins that some data-centre debt is not an asset-backed security and so escapes Dodd-Frank risk retention, the rule that makes a securitisation sponsor keep about 5% of what it sells. "It gives them the opportunity over time to push down the required equity in the deal," a securitisation lawyer at Orrick told CNBC. We wrote about it on 18 August. Less required equity, more leverage per dollar of building, and the sponsor keeps the fee either way.
What would make this dangerous
A continued, broad refusal by private credit to refinance sponsor-backed maturities. One Sophos is repricing and arguably healthy. A pattern across the 13,509-company backlog means the marginal borrower has nowhere to go, and losses resolve through defaults rather than negotiated terms.
Sponsors declining to inject equity as a general policy rather than a one-off. Thoma Bravo said no on Sophos. If that becomes the market convention, every workout starts from the position that the cushion is not there.
The funding side of private credit running while the asset side deteriorates. Investors in non-traded asked for 12.4% of NAV back in the second quarter of 2026 against caps of around 5%, leaving roughly $9.6bn queued — we set that out on 20 August. A rationing queue in the vehicles that fund sponsor debt is the same problem seen from the liability side.
Continuation vehicles marked at prices nobody independent tested. If a meaningful share of that 14% of exits turns out to have been transacted above what a third-party buyer would have paid, the reported distribution yield of 14.8% of NAV overstates how much real cash went back to LPs, and the correction arrives as marks rather than news.
Dividend recap volume turning back up sharply while exits stay frozen. Borrowing to pay yourself is defensible at the current pace; at 2021 exit-market prices with 2026 exit-market liquidity, it is a transfer from creditors to LPs with the sponsor taking a fee on the way past.