Ask a mega-fund what it's doing about AI and you'll get a name. Four in ten firms managing more than $10 billion in private equity have someone whose job is AI, and almost nine in ten say something about it publicly.
Ask a firm managing $250 million and, two times out of three, there's nothing to find.
We looked at 1,256 US private equity sponsorsNote 1, from sub-$100M firms to the largest firms in the market, and recorded what each one says publicly about AI: whether it has a named AI lead, a stated program, a passing mention, or nothing at all.
The gap matters because of what these firms own. Lower mid-market sponsors hold thousands of portfolio companies. Many are HVAC contractors, distributors, clinics, and staffing firms with no technical staff at all. The software companies have engineers, but the skill that matters now is rarer: keeping up with a technology that changes every few months, and rolling it out across an entire business. The length of task AI can complete on its own has been doubling every four to seven monthsNote 2,Note 3. Employers now pay a premium for people who can put AI to work across teams, and those people are hard to findNote 4,Note 5. Across the portfolio, that expertise has to come from outside, and the sponsor is the natural place for it to come from. At most of these firms, there's no one to send.
Below: how the gap breaks down by size, what separates the firms that act on AI from the firms that only talk about it, and why the mega-fund answer, a dedicated hire, stops making sense at a $250 million firm.
AI leadership tracks firm size
Stacked bar chart of 1,256 private equity firms in five size bands. The larger the firm, the more likely it shows a public AI signal: 22% of small firms (under $100M), 34% of lower mid-market firms ($100–500M), 46% of mid-market firms ($0.5–2B), 62% of large firms ($2–10B) and 87% of mega firms ($10B+). The share with a named AI lead climbs more steeply, from 4% of small firms to 42% of mega firms.
- Named AI lead
- Other public AI signal
- No public AI signal found
The chart sorts every firm by its strongest public AI signal: a named AI lead, a described AI program, a passing mention, or nothing we could find. Moving down the size bands, two things happen at once.
The share of firms saying anything about AI falls steadily, from 87% of mega-funds to 22% of firms under $100 million.
The share with someone accountable for AI falls much faster. From the mega-funds to the mid-market, the chance a firm has an AI lead halves or more with each step down: 42%, then 21%, then 9%. In the lower mid-market it's one firm in twenty. Below $100 million, it's one in twenty-five.
So the two measures pull apart. At the mega-funds, about half the firms that talk about AI have named someone to lead it. In the lower mid-market, it's about one in seven. Interest in AI reaches much further down the market than the people to act on it.
Firms managing more than $2 billion are a quarter of the sponsors we studied, and they employ more than half of the AI leads we found.
The lower mid-market owns more than 4,000 companies
The lower mid-market is large: the 356 firms without an AI lead hold more than 4,100 active portfolio companies. That works out to about a dozen companies per firm, on average run by a team of around ten people.
Services companies, healthcare administration, distribution, and field operations run on the kind of work generative AI has already been shown to speed up: customer support, clinical documentation, bookkeeping, and routine writingNote 6. Lenders who finance this end of the market say it has "less software correlation" than the larger market, where software takes a much bigger shareNote 7.
The expertise gap inside the portfolio
A software company already employs the people who build with AI. At US software publishers, 46% of the workforce is in computer and math roles. At durable-goods distributors, the share is 4.4%. At physician practices and outpatient clinics, it's 0.6%. At heating and air-conditioning contractors, it's 0.25%Note 8: in a 100-person company, a quarter of one person.
Four grids of 1,000 dots, one per kind of business, with a dot filled for each employee in that function per 1,000. Software publishers are far ahead at 456 per 1,000, about ten times durable-goods distributors at 44. Outpatient clinics (6) and HVAC contractors (3) barely register.
- Software publisher456 per 1,000
- Durable-goods distributor44 per 1,000
- Outpatient clinic6 per 1,000
- HVAC contractor3 per 1,000
A SaaS portfolio company can grow AI capability from its own engineering team. An HVAC contractor, a distributor, or a billing operation has almost no one to grow it from. Across the middle market, 70% of companies using generative AI say they need outside help to get full value from itNote 9.
For most of the lower mid-market's portfolio, the AI expertise has to come from outside the company.
Plenty of interest, few AI leads
About a third of lower mid-market firms say something about AI publicly: in their investment thesis, on their value creation pages, in portfolio updates. Only 18 of 374 (5%) have named someone to lead it.
"We talk about AI in every board meeting, but no one has the slightest clue about ROI." — Deal lead, lower mid-market firmNote 10
AI leads show up where someone's job is value creation. Among lower mid-market firms:
- With a dedicated operations team: 16% have an AI lead (11 of 68).
- Relying on an operating partner network: 4% (7 of 178).
- Where the deal team runs value creation: none. Not one of 128.
Firms with an ops team are more likely to have an AI lead at every size. At the mega-funds, half of them do. In the lower mid-market, 16% do. The team matters, and so does what the firm can afford to hire, which comes down to arithmetic.
Why the gap exists: it's arithmetic
The median lower mid-market firm in our data has 10 employees and $250 million in private equity assets. At a 2% management fee, that's around $5 million a year to run the whole firm.
A senior AI hire (for example, a former consultant brought in as a director or operating principal) costs roughly $350,000–$450,000 a year fully loadedNote 11.
For that firm, it's a 10% increase in headcount and roughly 8% of its fee income, for one person who then has to cover every company in the portfolio.
The median mega-fund has 146 employees and $21 billion in private equity assets. The same hire is about 0.1% of its feesNote 12.
Two squares sized by annual fees: a lower mid-market firm earning about $5M a year, and a mega-fund earning about $420M, 84 times as much. One senior AI hire, $400K fully loaded, is drawn as the same small square inside each. It is 8% of the smaller firm's fees but only about 0.1% of the mega-fund's.
Lower mid-market firm
~$5M a year in fees
Mega-fund
~$420M a year in fees
0.1%
the same hire
One senior AI hire, $400K fully loaded
So the large firms hire, and the small firms mention. The lower mid-market sees the opportunity. The mega-fund way of capturing it, a dedicated in-house hire, doesn't scale down.
What this means
If you run a lower mid-market firm, you may already be the one pushing AI. A third of your peers write about it in their thesis and value creation pages. The constraint is capacity. Your portfolio is full of the companies least able to build AI capability from inside: contractors, distributors, clinics, service businesses. Whoever champions it at the firm has a full deal load. As we heard from one head of value creation:
"We thought we were getting our arms around an AI strategy, but the world changes every two months." — Head of value creation
The model the mega-funds use, a dedicated hire, would cost about 8% of your fee income. Picture what that person would walk into: a dozen companies across several industries. An HVAC contractor on one dispatch system, a distributor on another ERP, a physician practice on its own billing stack. Each needs its own diagnosis, its own vendor choices, and someone on the ground to make the change stick. Few have a technical team to hand the work to.
So the work ahead is getting the capability without the headcount. That's what the rest of this series is about.
Notes
How we did this: We started from every adviser filing SEC Form ADV and kept the 1,256 that sponsor private equity funds, sized by the gross asset value of their private equity funds. For each firm, research agents read the firm's website, team pages, LinkedIn and press coverage, and recorded the strongest AI signal they found: a named AI lead, a described AI program, a mention of AI or nothing public.
As a caveat, this measures what firms say publicly. Some firms doing real AI work won't talk about it, so these figures are a floor, not a census.
METR, "Time Horizon 1.1," January 2026.
AI Digest, "A new Moore's Law for AI agents," March 2026.
Lightcast, Beyond the Buzz: Developing the AI Skills Employers Actually Need. July 2025.
Lightcast, "Emerging skills in AI jobs," July 2026.
Erik Brynjolfsson, Danielle Li, and Lindsey Raymond, Generative AI at Work, The Quarterly Journal of Economics, May 2025.
PitchBook, "Private credit lenders finding opportunities in lower middle market," July 2026. The lenders quoted are describing borrowers with $5–30M in EBITDA, a close match for the companies lower mid-market sponsors own.
BLS, Occupational Employment and Wage Statistics, May 2025, national industry-specific estimates. Share of employment in computer and mathematical occupations (SOC 15-0000) for software publishers (NAICS 513210), durable goods merchant wholesalers (423000), offices of physicians (621100), and plumbing, heating, and air-conditioning contractors (238220).
RSM US, Middle Market AI Survey 2025.
Quotes are from conversations with PE professionals during this research, shared anonymously
Heidrick & Struggles, 2024 North American Private Equity Operating Professional Compensation Survey (June 26, 2024).
Simplified: fees are charged on committed or invested capital, and large funds charge less. However, even at 1%, the mega-fund earns about 40 times the lower mid-market firm's fees at 2%.