The Five-Year Exit Is Over. Here's How to Get Your Valuation Back.

For twenty years, the private equity playbook for a technology business ran to a familiar clock. Buy it, professionalise it, grow it for four or five years, then sell it to a bigger fund or a strategic buyer at a higher multiple and hand the gains back to investors. Founders building to sell learned to work to the same clock, whether or not they had a fund on the cap table.

That clock has stopped for a lot of businesses. The plan said year five. The buyers said "not at that price". So the owners are waiting.

What's harder to tell is how much of that is AI, how much is the hangover from 2021, and whether the businesses stuck in the queue can do anything about it. They can. But it starts with being honest about what's actually going on.

How big is the problem?

The headline is real. Bain & Company's 2026 Global Private Equity Report puts the average holding period for buyout assets at exit at around seven years, up from five to six years between 2010 and 2021. Almost 40% of PE-owned companies have now been held for more than five years, up from 29% in 2019, and the industry is sitting on 32,000 unsold companies worth US$3.8 trillion. At SuperReturn in June, Apollo's deputy global head of private equity put it more bluntly, saying average hold times have roughly doubled, from about four years to almost eight.

McKinsey counts more than 16,000 companies held for four years or longer, 52% of all buyout-backed inventory and the highest share on record.

The money tells the same story. Bain's midyear report describes a four-year stretch of record-low distributions as a share of net asset value, and an implied capital cycle of roughly seven years. In that same report, a majority of investors said they start losing confidence in a fund manager when an exit comes in more than 5% below the last valuation mark. Read that twice, because it explains a lot of behaviour. If selling below the mark damages your next fundraise, holding and hoping looks rational.

So is 7 to 10 years the new normal? Not quite. The median holding period is closer to six years, which Private Equity Info's data shows is the longest on record. The average is around seven, pulled up by a long tail of older assets. In North American software specifically, more than 27% of PE-owned companies were bought at least five years ago and about 11% at least seven. The honest version is this: the five-year exit is gone as a safe assumption, seven is now typical, and for weaker businesses bought near the 2021 peak, 7 to 10 years is a realistic planning case.

Software has taken the hardest hit

Technology was historically one of the fastest sectors to turn over. Now it's one of the slowest. According to 451 Research data reported by S&P Global, software acquirers paid a median of 6.4 times trailing revenue in 2021, and just 3.2 times in 2026 as of April. Half the multiple, on the same revenue.

February was the shock. Public software valuations dropped nearly 30% in a month, and technology deal value fell 70% from the last quarter of 2025 to the first quarter of 2026. Private equity's own valuations of its software holdings fell only about 8% over the same period. That gap matters. Sellers anchor to the private marks; buyers anchor to public prices. Until those two numbers meet, deals don't clear.

As one 451 Research analyst put it: "The holding periods are longer and they're going to get longer because there effectively isn't an exit market for these companies."

The view from Australia and New Zealand

Local markets have felt it at least as sharply. Over the past year, Xero shares are down about 62% and WiseTech about 60%, even though Xero's revenue grew 31% in FY2026. That's not a business going backwards. That's a market repricing a model.

Latimer Partners tracked 128 ASX-listed technology companies and found 80% had fallen in value this calendar year, with software multiples that peaked at 14 times revenue in 2021 now sitting at 3.5 times. In the same Mergermarket report, a technology M&A banker said private equity sponsors and private owners of software businesses are unlikely to seek an exit right now unless they're under pressure, and lawyers at K&L Gates said earn-outs are being used in most of their deals to bridge the gap between what buyers will pay and what sellers will accept.

Sponsors are also reaching for other tools. EY's Australian PE update noted Pemba Capital Partners' continuation-vehicle transaction for Vets Central as an example of sponsors extending holds while still returning cash to investors. And venture capital has its own version of the problem: the Australian Investment Council notes that VC funds have called up more capital than they've distributed, because successful companies are being held for longer.

What's actually causing it

It's tempting to blame all of this on AI. That would be a mistake, and a costly one, because the fix depends on the cause. Three forces are stacked on top of each other.

1. The maths was stretched before AI arrived

A lot of the businesses stuck in the queue were bought at the top of the market. Analysis reported by Risk.net shows median entry multiples for the software cohort in buyout funds rising from about 13 times EBITDA before 2019 to about 21 times in 2025, often funded with heavy debt. Interest rates then rose, which made that debt more expensive and future cash flows worth less.

Bain spells out what that does to the plan. A deal that needed 5% annual EBITDA growth a decade ago to return 2.5 times the money over five years now needs 12%. Buyouts completed between 2020 and 2022 have returned a median 2.1 times at exit, against 2.9 times for technology buyouts done between 2010 and 2019. When the original plan relied on selling at the same or a higher multiple, any compression at all turns year five into year seven.

2. AI changed what recurring revenue is worth

This is the part that's genuinely new. For fifteen years, software was valued on the assumption that annual recurring revenue was sticky, predictable and cheap to serve. AI has put a question mark over all three.

Seat-based pricing is the most exposed. If an AI agent handles 80% of support tickets, a customer that needed 50 support seats might need 10. HarbourVest's modelling shows how hard that is to price your way out of: a 30% fall in seats with a 10% price rise still means a 23% fall in revenue. Lincoln International's technology co-head summed up the buyer's view simply: "You have this pressure on the seat-based license concept that's been around a long time."

Replicability is the second exposure. As PwC notes, the cost and friction of building software are falling, so products that compete mainly on features or user experience are easier for a customer, a competitor or an AI copilot to reproduce. The third is cost. AI features carry real compute costs every time they're used, which chips away at the 80% gross margins that justified software multiples in the first place.

None of this means a given business is losing customers. It means buyers no longer take ARR at face value. They want to know what it's made of.

3. Buyers can't yet tell winners from losers, so they discount everyone

The third force is uncertainty, and it might be the most damaging. Around 64% of PE firms surveyed by EY said AI disruption risk had made them more selective about software. Latimer Partners reports that one in five strategic dealmakers walked away from a transaction in 2025 because of AI's expected impact on the target. And strategic acquirers whose own share prices have fallen have less currency, and less appetite, to pay up.

When buyers can't price a risk, they price the worst case or they walk. That's why earn-outs are everywhere: they let the buyer pay for proof rather than promises.

It isn't only software. In technology services, EY finds AI is compressing commoditised delivery models while increasing demand for integration, governance and advisory work. Among managed service providers, documented AI and automation capability is attracting 9 to 13 times EBITDA, while undifferentiated generalists sit closer to 5 to 8 times.

The market hasn't closed. It has split.

Here's the part that rarely makes the headlines. Good businesses are still selling, and selling well. Bain cites MSCI analysis showing more than 75% of buyout assets exited above their next-to-last valuation mark, and high-quality assets with strong strategic value continue to attract buyers. In Australia, Pemba's Tom Matthews described the reset as a possible "golden era" for investors in software that leans into AI or has strong AI moats, and Pemba took its whole team off the tools for three weeks to reassess which niches were safe. Employee engagement software moved from green to red. Agtech coupled with hardware moved from red to green.

So the useful question isn't "are technology businesses worth less?" It's "which side of the split are we on, and can we prove it?" Drawing on the frameworks buyers are now using, from HarbourVest, PwC and Latimer Partners, the split looks like this:

Six pathways back to a valuation

None of these are quick fixes. All of them are within management's control, which is more than can be said for interest rates or the public markets.

1. Run the replication test before a buyer does

Ask the uncomfortable question: could a capable customer or competitor rebuild our core workflow with today's AI tools in a quarter? Diligence teams are now asking exactly that, along with whether a product is merely a prototype, a product, or a platform a compliance team will approve.

If the honest answer is "probably", the roadmap needs to move toward the things that are hard to copy: owning the record rather than reading from it, embedding the industry rules and compliance logic your customers rely on, deepening integrations so you're a node in their ecosystem rather than an app beside it, and building data that gets more valuable the longer a customer uses you.

2. Get ahead of seat compression

Model your own downside before someone else does. Take your top 20 customers, assume AI lets each of them do the same work with 30% fewer people, and see what happens to your revenue. If the answer is frightening, that's your pricing roadmap.

The direction of travel is toward hybrid models: a platform fee for access and value, plus pricing linked to usage, transactions or outcomes. Embedded payments are one reason vertical software is holding its value; payment volume doesn't shrink when a customer's headcount does. Pricing changes take time to land with existing customers, so start well before a sale.

3. Rebuild your retention story from the bottom up

A strong net revenue retention number used to close the conversation. Now it opens one. Buyers are asking whether AI add-on upsell is masking seat contraction underneath, so they'll want gross retention as well as net, split by cohort, by product module, and by revenue that's exposed to AI versus revenue that isn't.

Build that view now, from your own data, and know what it says. If there's a weak cohort or a product line quietly shrinking, it's far better to find it and fix it than to have a buyer find it and reprice the whole business.

4. Make AI show up in revenue, not just the deck

Every seller now claims an AI strategy. Few can show what it's worth. The evidence buyers respond to is concrete: AI features that are priced, adoption figures, attach rates, and customers who'd struggle to go back. Xero, for example, reports two million subscribers already using AI features embedded in the platform.

Product speed counts too. Shipping meaningful improvements faster than competitors is becoming one of the few moats AI can't erode, because it's a capability of the team rather than a feature of the code.

5. Plan to earn the multiple, not inherit it

The quiet assumption behind most five-year plans was that the exit multiple would be at least as good as the entry multiple. Remove it. Plan on the basis that value has to come from earnings growth, and that buyers will pay for growth that's both durable and efficient.

AI helps here as much as it threatens. Used inside the business, in support, onboarding, finance and sales operations, it should lift revenue per employee and lower the cost to serve each customer. Those are exactly the numbers that hold up in diligence. Make sure your core metrics are clean, consistent and defined the way a buyer will define them, with one-off fees kept out of recurring revenue and the full cost of sales included in customer acquisition cost.

6. Treat the exit as a process, and widen your options

EY reports that 88% of private equity firms now run targeted exit-readiness work during the hold, and nearly half start formal preparation 12 to 24 months before a sale. Founders should do the same. A buyer's diligence questions are predictable now; answer them in advance.

Widen the definition of an exit, too. Earn-outs can bridge a valuation gap if the business is confident in its numbers. Partial sales and minority investments return some capital without selling at the bottom. Continuation vehicles are an option for sponsor-backed assets, though Bain notes they're getting harder to execute as investors scrutinise them more closely. And consolidators are active in Australia, from Constellation Software's Vela to Banyan, Jonas and Accel-KKR's Humanforce, which bolted on two businesses in April. Decide early whether you're building a platform or a very good bolt-on. Both can work, but they need different plans.

Finally, ask the hard question Bain asks of fund managers: will more time actually create more value? Holding a business for another two years only makes sense if those years change the story. If they won't, the best price may be the one available now.

The clock hasn't stopped for everyone

The five-year exit was partly a product of cheap money and rising multiples. Both have gone, and AI has added a genuine new question about which technology businesses will matter in five years' time. That's uncomfortable for owners who bought at the top.

But the businesses selling well today aren't lucky. They own the work their customers can't do without, they've priced for a world with fewer seats, they can prove their revenue is durable, and they're using AI to get better rather than just talking about it. Every one of those is a choice, and most of them can be started this quarter.

This article is general information only and isn't financial, investment or legal advice. Figures are as reported by the sources listed, current at the date of publication.