AI in private equity: how funds turn AI into value in their portfolio companies

Why AI has become a value creation topic for private equity

For years, much of the return in buyouts came from leverage and rising multiples. That tailwind has weakened. In Bain’s illustrative deal model, a typical buyout today needs 10–12% EBITDA growth a year to return 2.5 times the equity over five years, with less leverage and flat multiples. In 2015 about 5% was enough (Bain, Global Private Equity Report 2026).

That gap has to be closed in the operations of each portfolio company: more margin per order, less manual work per invoice, faster answers to customers. This is where AI fits, provided it reaches the profit and loss statement and does not stay a pilot.

Three places where AI shows up in private equity

Fund-level tools save time for the deal team. The large value pool sits in the portfolio, because every point of margin there is multiplied by the exit multiple. This article focuses on that part. How to assess a single company’s starting point is covered in AI readiness of a portfolio company.

The AI value gap: why most pilots stall

Most portfolio companies are already testing AI. Far fewer can show it in their numbers. In a BCG survey of 1,250 senior executives, 5% of companies achieve AI value at scale, 35% are scaling, and 60% report minimal gains so far (BCG, September 2025).

Two findings explain much of the gap. First, AI helps where a task fits it and hurts where it does not. In a field experiment with 758 consultants, AI users completed 12.2% more subtasks, 25.1% faster and with 32% higher graded quality on tasks inside the AI’s capability frontier. On a task outside it, they were 19 percentage points less often correct (Dell’Acqua et al., Organization Science 2026). Choosing the right process matters more than choosing the right model.

Second, building alone is harder. In preliminary, self-reported findings from 52 organisations, AI tools sourced through external partnerships reached deployment about twice as often as tools built in-house, roughly 67% against 33% (MIT NANDA, July 2025). The sample is small, but the direction matches what we see: pilots stall on data, permissions and ownership, not on the model.

Where AI moves value in a portfolio company

We look at four levers in every company. Each has documented results at named or clearly described companies; AI use cases in portfolio companies goes through them with ranges and caveats.

• Commercial: price and terms discipline in proposals, faster answers to requests. A $15bn distributor gained more than 250 basis points of margin, more than 200 of them from AI price and discount tools over about 18 months (McKinsey 2026, anonymised).
• Operations: order intake, scheduling, routine checks. Danfoss cut order processing time by 50% with AI agents that read B2B order e-mails (Google Cloud case).
• Customer service: Klarna’s customer service and operations cost fell 15% in 2024, the year it launched its AI assistant (F-1 filing).
• Finance and reporting: Eurofins says automated invoice processing freed up 30% of its accountants’ time (Rossum case).

People and organisation is the fifth area, and the one that decides whether the other four last: knowledge search, onboarding and leadership teams that know how to run AI-enabled processes.

What PE owners have reported so far

Public figures from fund owners are still rare, and all of them are self-reported:

• Partners Group puts two AI use cases at Foundation Risk Partners at $10m of EBITDA, 120 basis points of margin; close rates doubled, and the first use case went from concept to production in 14 weeks (Partners Group 2026).
• Cengage, in which Apollo holds a minority stake, cut customer-care cost by 15% in early results (Bain 2025).
• Assent, a Vista portfolio company, reports more than 10% full-time employee savings in support at more than 33% deflection (Vista 2026).

Note the size: one portfolio company, a few use cases, effects in the low hundreds of basis points. At an exit multiple of ten, that is meaningful. Promises of AI multiplying a company’s value are not backed by any of these sources.

From time saved to EBITDA: the arithmetic that matters

Many AI business cases count hours saved as money saved. They rarely are. Uber cut invoice handling time by 70% but process cost by 25–30% (Uber 2025). Maximus cut invoice processing time by 85% and, through natural attrition, its accounts payable team by 20% (Rossum). By our conversion, 24–43% of the saved time left the cost base.

For an investment team that means three rules. Count cost cases with a realisation share, not at full time saved. Count extra volume at contribution margin, never at revenue. And subtract the running costs of the tools: licences, cloud and upkeep. A price case is the strongest lever because it flows almost fully to profit: at a 12% EBITDA margin, one point of price on all revenue adds 8.3% to EBITDA, as long as volume holds.

How to get from pilot to portfolio value

1. One lens across the portfolio. Look at every company through the same questions: processes, data, systems and people. Talk to the teams who do the work, not only to management: the cases they name are the cases they will use. The result is a value map per company and a heatmap for the investment team.
2. Calculate each case before building it. Every number derived and sourced, every assumption visible, with an owner in the company. The model and its cockpit stay with the company: every number with its source and quality grade, every scenario at the push of a button. Cases that do not add up are dropped early.
3. Build controlled MVPs on existing systems. The AI does the groundwork, a person approves, and the ERP or CRM stays the system of record. A controlled MVP typically takes two to four weeks of build time.
4. Track impact against the value creation plan. Each live tool gets a KPI that the management team already reports, not a separate AI dashboard. AI in the value creation plan shows how.
5. Reuse what works. A tool that works in one company becomes a pattern for the next. Patterns travel faster than platforms.
6. Train the leaders who run it. Tools change processes; leaders change companies. Management teams need to know where AI fits, where it does not, and how to steer a process in which AI does part of the work. How to take teams along is described in AI change management in portfolio companies.

Who should drive AI in the portfolio?

Large owners such as Partners Group and Vista report AI programmes of their own; Partners Group’s example above came from collaboration between its portfolio companies. In our experience, many mid-market funds have an operating partner or a small value creation team, but no engineers. They need a partner who can do three things at once: calculate value like a strategy adviser, build tools like an engineering team and enable leaders like a trainer.

That is how we work with private equity funds: as an impact partner, led personally by two former partners at McKinsey & Company and Roland Berger. We find where AI moves value in each company, build the tools in weeks on the systems the company already runs and stay until it works. Code, data and know-how can go fully to the portfolio company. Details are on pe.leadbeyond.de.