For PE sponsors and mid-market bankers, the comfortable assumption that legacy portfolio companies can simply outlast a soft exit market is breaking down. The average global PE hold period now stands at 6.6 years, a historic high (1), and McKinsey estimates that 16,000 buyout-backed companies, roughly 52% of total inventory, have been on sponsor books for more than four years, the highest concentration of stranded assets on record (2). Waiting it out is not really a strategy at this point; for a lot of these portfolios it is just an expensive way to keep compounding the same problem.
The pressure is showing up where it always does first: in LP statements. Distributions as a share of NAV fell to roughly 17% in 2025 against a ten-year average of 26% (3), and five-year DPI is at its weakest level in more than a decade (4). Sponsors raised about $414 billion in 2025, the lowest total since 2018, with funds closing at a 19% average discount to target and timelines extending to 20 months (5). At this point LPs have stopped politely listening to narratives about timing the cycle. They want cash back, and that pressure is now reshaping how GPs talk about value creation.
Reframe AI as a valuation driver, not an IT line item.
That reframe matters. For the last three years, AI has been discussed inside most PE firms as a technology agenda: a CIO line item, a digital transformation budget, a series of pilots in two or three portfolio companies. That framing is now actively destroying value. In a market where exits are stalled, leverage is expensive, and financial engineering has run out of room, AI is effectively the only EBITDA a sponsor can still mint inside a hold period. It is the most leverageable EBITDA and multiple-expansion tool still available to businesses that were not built AI-native, and treating it as anything less is a portfolio-level mispricing.
The numbers support the reframe. Accenture estimates that every dollar invested in AI transformation delivers an annualized EBITDA uplift of two to four times by exit (6), and EY's 2026 survey of PE firms found that 63% now use AI in value creation, up from 41% a year earlier (7). But only about 20% of portfolio companies have operationalized AI use cases that deliver measurable returns; the other 80% sit in what the industry has started calling "pilot purgatory," experimenting in silos, duplicating costs, and producing decks rather than EBITDA (8). The firms that win the next exit cycle will be the ones that replace scattered pilots with a portfolio-wide AI operating model: shared infrastructure, a common stack, reusable agents, and operating partners who track AI EBITDA the way they track working capital.
Converting legacy assets toward AI-native is a sequence, not a slogan.
The work of moving a non-AI-native business toward something that reads as AI-enabled is concrete and sequenced, and operating teams should treat it less like a digital transformation and more like a synthetic acquisition: you are buying AI characteristics into an asset that did not come with them. Four moves, in order. First, consolidate fragmented customer, transaction, and operational data into a single governed layer, because no use case compounds value until the data foundation exists. Second, pick two or three EBITDA-anchored use cases the P&L will actually feel, such as dynamic pricing, demand forecasting, procurement optimization, or churn prediction, and resource them like cost-out programs with named owners and weekly tracking. Third, embed AI into the product or service the customer touches, not just internal operations, because that is what shifts the multiple rather than just the margin. Fourth, install AI fluency in the org chart, with a Head of AI or AI operating partner at the portco level and a workforce plan that quantifies the productivity gain. Done in sequence, this is what turns a legacy asset into a credibly AI-enabled one inside a single hold period.
The harder truth is that most firms cannot execute this sequence themselves. They either freeze, unsure where to start, or scatter, chasing AI experiments without a cohesive strategy. Neither produces EBITDA. Very few sponsors, and even fewer of the mid-market portfolio companies they own, have the in-house expertise to map current state across the two axes that actually matter (people and processes), score AI opportunities on an impact-versus-effort matrix, model ROI for the top candidates, and translate the result into a phased roadmap a CEO can act on Monday morning. A focused four-week AI audit that produces exactly that, with a quick-win build attached so leadership can see what is possible rather than just read about it, is increasingly the first move for sponsors who are serious about the conversion. It is also the cheapest possible insurance against the six to twelve months of scattered pilot spend that "figure it out internally" tends to produce.
The mid-market view: a K-shaped deal market.
Mid-market deal teams are seeing the same dynamic from the sell side, but more acutely. The market has gone K-shaped. Roughly 600 transactions above $1 billion drove a 36% jump in global M&A value in 2025, while the remaining ~47,000 deals were essentially flat year over year (9). Plenty of strong midmarket companies are stalling in process despite perfectly defensible fundamentals, because buyers will no longer underwrite a growth story without a credible AI thesis sitting underneath it.
The bar has moved fast. Skadden's 2026 M&A outlook notes that buyers now expect AI narratives backed by evidence: metrics, dashboards, customer outcomes, and a credible workforce story (10). When the CIM says "AI-enabled" and the data room shows three disconnected pilots with no measured impact, AI flips from premium to discount factor. The public market gap shows where this is heading: companies with demonstrable AI capability trade at 30 to 50% premiums to peers, and AI-native software commands 8 to 12x revenue against 3 to 5x for everyone else (11).
AI belongs in sell-side diligence, not just the management deck.
What the sharper bankers have started doing is folding AI into the standard sell-side diligence scope alongside QoE, legal, tax, and IT. This is a different conversation from the now-common one about AI tools accelerating diligence workflows. Here, AI capability itself is the diligence subject. A proper AI sellside diligence answers the questions a sophisticated buyer is now going to ask before they price the asset: where does AI already create measurable economic value, what is the maturity of the data foundation, what does the IP and model ownership picture look like, what realistic AI value-creation initiatives could a buyer underwrite in years one through three, and what gaps would otherwise become discount factors in the bid. The point is to surface those issues twelve months before the bake-off, not in the diligence Q&A. The bankers and sponsors winning premium outcomes today are using that preprocess window to remediate: stand up two or three use cases with measurable EBITDA impact such as pricing, demand forecasting, procurement, or sales productivity; instrument the KPIs so the data room shows the lift in dashboards rather than claims; bring in interim AI or CTO talent who can credibly speak to the architecture in management meetings; and target the buyer pool that values AI evidence rather than the generalist consolidators that do not. The CIM then leads with the AI thesis, supported by quantified proof points, and when those underlying numbers are real, the buyer-side story largely takes care of itself.
There is also a window argument that does not get made enough. CB Insights logged 266 AI-related M&A deals in Q1 2026 alone, a 90% increase year over year (12). Every quarter that legacy businesses delay operationalizing AI, the AI-native comp set deepens, buyers get more sophisticated about distinguishing real capability from feature bolt-ons, and the implicit AI discount on non-AI-native companies keeps widening, which is to say the penalty for waiting is not flat; it accrues.
The diligence question is changing.
The right way to think about all of this: AI is no longer a portfolio company initiative reported up to the GP. It is a fund-level value creation lever that will determine whether the 2026 to 2028 exit window is a recovery or a quiet markdown. For LPs evaluating GPs, this is becoming a diligence question in itself. The question is not "do you use AI internally?" but "what is the AI EBITDA contribution across your top ten holdings, and how do you track it?"
The board is now part of what gets diligenced.
All of this is changing what gets asked at the board level, and the change is sharper than the usual "AI is the new cybersecurity for boards" refrain. Buyers in this market are starting to underwrite governance the way they underwrite financials. Board composition, the seniority and AI fluency of the director who actually owns the AI agenda, and the cadence and substance of AI items in board minutes are being read in diligence as a signal of how seriously the asset has been governed for value. Directors who treat AI as a management-team update rather than a fiduciary question are quietly creating exit risk for the company they sit on, and increasingly for themselves. Search firms are already responding, with AIexperienced director mandates up sharply, and the boards that move first will sit on the assets that price first.
The legacy investments that get marked up over the next 18 months will not be the ones that wait for multiples to recover. They will be the ones whose sponsors, bankers, and boards stopped treating AI as a technology choice, started treating it as the valuation event it has already become, and ran the conversion to AI-enabled as the synthetic acquisition it actually is: diligenced, sequenced, resourced, governed, and priced into the next exit.
Sources:
(1) McKinsey & Company: Beating the odds: How private equity firms can improve exit prospects (Global Private Markets Report 2026). https://www.mckinsey.com/industries/private-capital/our-insights/beating-the-oddshow-private-equity-firms-can-improve-exit-prospects
(2) McKinsey Global Private Markets Report 2026: 16,000 company backlog / 52% of buyout inventory. https://www.mckinsey.com/industries/private-capital/our-insights/beating-the-odds-how-private-equityfirms-can-improve-exit-prospects
(3) Allianz Trade: Private equity in transition: from distribution drought to selective recovery (distributions ~17% of NAV vs. 26% 10-yr avg). https://www.allianz-trade.com/en_global/news-insights/economic-insights/privateequity-transition.html
(4) thrv: The Private Equity Liquidity Crisis: Why DPI Is Stalling Across Portfolios. https://www.thrv.com/blog/theprivate-equity-liquidity-crisis-why-dpi-is-stalling-across-portfolios-and-leaving-saas-companies-unsellable
(5) With Intelligence: Private Equity Fundraising Report 2025 ($414B raised; 19% discount to target; 20-month timelines). https://www.withintelligence.com/insights/private-equity-fundraising-report-2025/
(6) Brightwave: How AI Is Reshaping the Private Equity Operating Model (Accenture: 2 to 4x EBITDA uplift at exit per $1 of AI investment). https://www.brightwave.io/blog/how-ai-is-reshaping-the-private-equity-operatingmodel
(7) EY: How AI is sustainably transforming value creation in private equity (63% of PE firms using AI in value creation, up from 41%). https://www.ey.com/en_ch/insights/strategy-transactions/ai-in-private-equity
(8) Glean: How private equity can turn AI pilots into portfolio value (only 20% operationalized; 80% in "pilot purgatory"). https://www.glean.com/blog/ai-value-gap-private-equity
(9) Bain & Company: M&A Report 2026 (K-shaped market: ~600 deals >$1B drove 36% value growth; ~47,000 other deals flat YoY). https://www.bain.com/insights/topics/m-and-a-report/
(10) Skadden: M&A in the AI Era: What Buyers Can Do to Confirm and Protect Value (2026 Insights). https://www.skadden.com/insights/publications/2026/2026-insights/sector-spotlights/ma-in-the-ai-era
(11) FE International: AI M&A Trends 2026: Why Acquirers Pay Premium Multiples (30 to 50% premium; 8 to 12x revenue for AI-native vs. 3 to 5x). https://www.feinternational.com/blog/ai-ma-trend
(12) FE International citing CB Insights: 266 AI M&A deals in Q1 2026, +90% YoY. https://www.feinternational.com/blog/ai-ma-trend
