Private equity is the fallback exit. Its bid for software fell two thirds in a quarter.

Why does this route exist at all?

Because most venture-backed companies do not reach the outcome their capital structure was built for, and the alternative to a sponsor sale is frequently no transaction. A company with real revenue, real customers and a growth rate that no longer supports another priced round has a business worth owning. It does not have a business worth the preferences sitting on top of it, and those are two different statements.

Private equity has historically been the buyer that closes that gap, because it underwrites cash generation rather than terminal market share. What has changed in 2026 is the price at which it is willing to do so in the sector where most of these companies sit. US sponsor deal value in software fell to $10.7bn in the second quarter of 2026, down 65.7% year on year, cutting software to roughly 6% of private equity deal value from 13.3% in the first quarter (PitchBook data, reported via Blue River Financial Group, July 2026).

The financing market moved first and further. Software's share of broadly syndicated loan issuance fell to 8.6% in 2026 to date from 17.6% in 2025, its lowest since 2013 (PitchBook LCD, as at 30 June 2026). Credit commentary treats this as structural rather than cyclical, on the view that AI-driven disruption of legacy software business models is a credit story rather than a cyclical one (Sikich, Q2 2026 Credit Market Update, 13 July 2026). A founder waiting for the software bid to revert is waiting on a call that the lenders have already made the other way.

What changes when the valuation basis changes?

The question the buyer is answering. A venture investor priced the company on what the revenue could become. A sponsor prices it on what the earnings are, and then on how much debt those earnings will carry. That is not a haircut applied to the same calculation. It is a different calculation, and it frequently produces a number below the last preferred round regardless of how the business has performed.

The marks in sponsor portfolios have moved in the same direction, which tells a seller what the comparison set looks like. Software valuations held in private equity portfolios declined about 8% in the first quarter of 2026, and 8.9% in the United States (Bain, Private Equity Midyear Report 2026, 8 June 2026, on MSCI data as at 31 March 2026). Private marks lag public ones, so the buyer's own book is being written down while the seller is negotiating.

The practical translation is that revenue quality becomes the whole conversation. Gross margin after the cost of delivering the product, net retention, the share of revenue under contract rather than under renewal risk, customer concentration, and the cash cost of the growth that produced the revenue. Metrics that were leading indicators to a venture investor are now inputs to a debt schedule.

What does the preference stack leave for common?

Whatever is left after the preferred holders are satisfied, which at a price below the aggregate preference is nothing. This is arithmetic rather than negotiation, and it should be run before a process rather than discovered inside one.

The mechanics are worth stating plainly. Each preferred round usually carries a liquidation preference, commonly one times its invested capital, sometimes with participation that lets the holder take the preference and then share in the remainder. Seniority determines the order in which those claims are paid, so a later round frequently sits ahead of an earlier one. At any given price, each holder takes the better of its preference or the value of converting to common, and the conversion point differs by round. Common stock, which is what founders and option holders hold, is paid last.

Two consequences follow. First, there is a range of sale prices across which the company can be sold, the preferred holders are made whole, and the founding team receives materially nothing. Second, inside that range the preferred and the common want different transactions, which is why management retention packages carved out of the proceeds are a standard feature of these deals rather than a generosity. A buyer who wants the team to stay has to solve for the team separately from the cap table.

There is no published dataset measuring what common holders actually receive in venture-to-sponsor transactions, and this article does not estimate one. The model that matters is the company's own, run at several prices, before anyone is approached.

What does a sponsor require before it will engage?

Evidence that the business funds itself, and a debt schedule that a lender will support. The second is currently the binding constraint. Software borrowers pay 75 to 100 basis points above the standard matrix by one advisor's account, or 150 to 300 basis points above comparable non-software credits by another, with interest-only periods cut from three years to two or less and liquidity minimums and cash controls now standard (Houlihan Lokey and SPP Capital Partners, both via SPP, Market At A Glance, July 2026).

The reason lenders are cautious is visible in their existing books. Roughly 40% of private-credit borrowers were running negative free cash flow, and around 30% of outstanding software loans mature by 2028 against 22% for the broad leveraged loan market (Allianz Research, Private equity in transition, 20 February 2026). That publication predates the second-quarter 2026 repricing and is used here with its February vintage labelled. A lender looking at a refinancing wall in its current portfolio is not the marginal buyer of another software credit.

Which means the operational requirements are specific rather than aspirational. Positive or clearly reachable cash generation without the growth spend, revenue that is contracted rather than renewed monthly, retention evidence at the account level, no single customer above the thresholds that gate institutional buyers, a management team that will sign up to stay, and a capitalization table clean enough that the transaction does not require unanimous consent from investors who have written the position down and stopped answering.

What are the alternatives if the numbers do not work?

Three, and each has a different owner of the decision. The first is a strategic acquirer, who can pay for the product, the customer list or the team rather than for the earnings, and who is not constrained by a debt schedule. Where a company has capability a larger vendor would otherwise build, this is usually the highest-value route and it is reached by a built list rather than by waiting for inbound.

The second is time, which is only an option if the business is self-funding. Reaching cash-flow breakeven changes the counterparty from a buyer of last resort to a buyer with a choice to make, and it is worth more than any negotiating position that can be constructed with a shorter runway.

The third is a restructuring of the capitalization table before a sale rather than during one. Preference stacks can be recut with the consent of the holders, and holders who have already marked the position down are sometimes willing, because a clean structure with an incentivized team produces a higher price and therefore a higher recovery for them too. That conversation is slow, it belongs to the board, and it cannot be started once a buyer is in diligence.

What does not work is running a process and hoping the price clears the stack. If the model says a fair price leaves the common with nothing, the process will confirm it in public, and the company will have spent its remaining credibility finding out.