Part 1 of a multi-part series for middle-market PE Firms
I've held many roles over the years: lender, banker, bankruptcy trustee, CEO, CFO, Chief Restructuring Officer. I've never been a PE firm partner, but I've advised and/or sat on the board of more than a few of their portfolio companies, and I've spent decades inside countless middle market businesses in just about every condition you can imagine: thriving, struggling, scaling, contracting, adapting well, and not adapting at all.
After the last 30 years of doing this work, I have been seeing fundamental changes to the dynamics of that industry. These changes are being confirmed through conversations I have had with PE partners focused on the middle market (which represents something like two-thirds of all PE deal value in the country). Simply stated:
The deals that used to work don't work the same way anymore. In the early days, the approach was buy right, lever it up, tighten the operating model, find Lego-like add-ons, grow organically for five or six years, and sell. Adding leverage and financial engineering was good enough to build platform success, but as the PE industry grew, as did competition for LP capital and platform companies, a more nuanced approach was required. Firms shifted to more specialization, focusing on niche markets and building a better mousetrap of operational efficiencies. However, now these playbooks are no longer as effective and the market returns have stagnated.
Below is a summary of some of the key fundamental shifts that are impacting value:
- Rates are up and likely aren't going back to zero anytime soon. The Fed's own projections say so through 2028.
- Uninvested capital, or "dry powder," is piling up and sitting there longer than it used to, some of it for four years or more.
- Portfolio companies are stacking up in the sell queue, tens of thousands of them, with nowhere to go.
- Holding periods keep stretching, now past six and a half years on average.
- For a company this size, an IPO was never much of a realistic exit, and strategics have gotten picky too, so close to half of all exits now are just one PE firm selling to another.
- Multiples have stopped climbing, sitting flat at or below record highs.
This is a developing multi-layered problem, which is currently culminating in lower returns for Limited Partners, and could cause structural cracks in the foundation of the PE industry.
The one-year IRR (meaning the annualized return the whole asset class generated over the trailing twelve months, the same lens you'd use to judge how the stock market did this year) for buyout funds has fallen from around 16% before the pandemic to about 9% now, which seems especially stark given that the stock market has put up three straight double-digit years. True, the stock market has been an early beneficiary of the AI phenomenon; but the point remains that LP returns from PE investments are dropping with cash going back to LPs running somewhere around 11 to 14% of NAV, against a historical average near 29%, so less than half what it used to be.
Average buyout multiples appear to have hit a ceiling, at 9x through most of the 2010s, and then they jumped a bit post-pandemic to over 11x, and last year hit a record 11.8x. These are the numbers across all sectors. My network, who focus primarily on the middle market, say pandemic driven multiples hit a high in the early 2020s. Since then they have retreated a bit and have not returned to the highs. The middle-market exit multiples are going in the wrong direction.
Some increase in multiples is no surprise, with the proliferation of new PE firms entering the market over the last 15 years. The number of potential buyers for the same deal continues to scale. So some "multiples inflation" is the expected result. But at a certain point the investment thesis can no longer support the price increase. Naturally, multiples will stagnate once further price increases make a deal too much of a stretch to justify the investment. With less clear value deals on the market, one would expect less capital to be deployed as buyers become more selective. This is exactly what is happening with undeployed capital almost doubling in the last 10 years ($0.7 trillion pre-pandemic to $1.3 trillion since 2023).
Bain has a term: "12 is the new 5." For most of the last fifteen years, the industry could hit a healthy 2.5x return on something like 5% annual EBITDA growth, because cheap leverage and multiple expansion did the rest of the heavy lifting. Bain's own research now puts that bar at 10% to 12% a year, more than double, driven by multiple factors, including debt that costs more and multiples that already appear to be at a ceiling.
That's the new environment. The ground just isn't as fertile as it used to be, and putting up a top-quartile return now takes real fundamental improvement, not just financial engineering. However, there is a seismic shift that could reset everything: AI is turning into the biggest valuation lever private equity has seen in years, maybe ever.
Confirmation of the impact of AI on middle-market values is still emerging, with many of these bets currently incubating in PE portfolios. But it is reasonable, given what we do know, to take a rising tide approach to estimated valuation impact: all indications are that there's a tremendous reward for companies that are showing up as "AI-native" (meaning AI isn't just running in the background of the business, it's built into the business model itself). Scarcity of real AI capability and buyer competition for AI-native are significantly driving up multiples for the companies that have it.
In addition, I'm hearing from PE partners directly that they're seeing multiples in the middle market in the neighborhood of 20 times EBITDA for companies that have genuinely arrived at AI-native. This lines up with the data that is emerging on AI's impact on valuations across a number of sectors:
- McKinsey scored 471 PE-backed product companies on how deep their AI use actually runs. Scattered use, versus AI rolled out everywhere internally, moved the multiple almost nowhere, 13x to 14x revenue. Real product-embedded AI moved it to 31x, more than double the floor.
- Real deals back it up. Alphabet paid $32 billion for Wiz on AI-driven threat detection. Palo Alto Networks paid $25 billion for CyberArk on AI identity security. ServiceNow paid roughly $3 billion for Moveworks, built around natural language AI. Workday paid $1.1 billion for Sana, an AI-native HR and finance platform. Recently an AI engineering company with 160 employees sold for $1.5 billion. Let’s do the math: that’s about $10.0 million per employee!
- In the broader private market, AI-native software companies are trading at 8 to 15 times revenue today against 4 to 6 times for otherwise comparable traditional software.
- PwC's 2026 research on AI performance found nearly three quarters of the economic value companies are getting from AI is captured by just one fifth of them. The gains aren't spread evenly. They're concentrated in the companies that actually did the work.
- Buyers have noticed. EY's private equity exit readiness research found the share of GPs who flag AI as a significant factor in an exit more than doubled in a year, from 7 percent to 16 percent.
- Also, there's a real cost to getting this wrong in the other direction. FE International's valuation work found AI claims that don't survive diligence, contain thin data, and no real defensibility, and thus can compress a multiple 15 to 30 percent.
For a company that can genuinely claim both real AI-native capability and the characteristics of an actual growth platform, and especially for one that gets there first in its market, it seems to be safe to expect a doubling of EBITDA multiple off the middle-market baseline. Call it 7x becoming something in the neighborhood of 14x, and I'd argue that's the conservative read given what my own network is reporting at the high end. That being said, whether the real number lands at a 50%, a 100%, or 150% lift, it doesn't change the conclusion. Directionally, the reward for AI-native is considerable.
One of the more amazing elements is that this isn't only for tech-centric companies. It has transcended to old-line industrial businesses that traditionally didn't have much opportunity for true retooling of their processes and market approach. Given the reduced cost to build custom solutions, every “rust-belt” company needs to rethink its vision of what it can become and where it is going. I can assure you that the competition is doing just that. Those that get a jump on this will be rewarded with significant value uplift.
We're still early in the cycle, and AI-native transformation inside a portfolio company doesn't show up and get measured overnight. Given how long a typical hold period runs, it could be years before anyone publishes a conclusive, middle-market-specific study on exactly what AI-native capability is worth at exit for a company this size.
…but the "AI tide" is clearly rising valuations across the board.
We have seen a preview of this tech-inspired valuation pop before, most recently, back in the 2010s, with what everyone called "digital transformation." Digital transformation meant using cloud, mobile, analytics, and connected systems to change how a company served customers, ran its operations, and in the best cases, its actual business model.
MIT research on digital maturity found that companies who made the digital shift saw better than a 25% increase in both profitability and market valuation over their peers. The market backed that up with its own rule of thumb: by around 2015, "the Rule of 40" (EBITDA margin plus revenue growth) became the accepted way to price a digital business, and companies that cleared it traded at about a 75% premium over the ones that didn't.
I think the AI impact ends up bigger. Digital transformation, for all the value it created, was mostly about moving analog processes onto digital rails, faster, more connected, more visible, but still running on the same basic operating model. It made the existing business run better. It rarely changed what the business could actually do.
AI reaches further into the decision-making, judgment, and market analysis work that used to require a person. Now it can be done at a fraction of the cost. Even more impactful, an organization's subject-matter experts, a rare breed who are exceptional in their know-how but also often a bottleneck for the rest of the team given their limited capacity, can now be liberated from much of their routine work, amplifying their impact across the organization.
If moving processes onto digital rails alone leads to a 75% uplift for the top performers, the monumental organizational reengineering that comes with AI-native transformation may be creating the biggest fundamental growth engine and value uplift we have ever seen.
Over the next few posts, I'll build this case:
- First, exactly what's changed in the market since the pre-pandemic era, and what it's already cost in returns and distributions.
- Then, why none of this fixes itself if you just wait it out.
- I’ll show why the market has already started paying a premium for real AI capability and I'll go deeper on the digital transformation precedent.
- After that, I will outline our AI maturity model, giving a roadmap for evaluating any company's "AI story" and what it takes to move it forward along with why so many companies that look like they're transforming are really just stuck.
The PE industry has grown greatly over the last three decades with successful firms always shifting to higher-value approaches as the status quo stagnates. Those firms that have shifted ahead of the competition have been rewarded with superior returns. Today, with the traditional playbooks no longer providing adequate returns, it’s time to shift to the next value wave. PE partners who catch the AI wave first will be the winners of the next cycle.