By Bain's Global Private Equity Report, roughly 20% of portfolio companies have operationalized generative AI with concrete, measurable results, while about 70% of general partners expect high impact within three to five years and only 6% see it today. The interesting number is not the optimism. It is the 20%, and what separates it from everyone else.
That separation is the operating partner's problem, not the deal partner's. Buying AI tools is easy and nearly universal; McKinsey's work on value creation in private equity and BCG's 2026 study of AI-first firms converge on the same finding — many portfolio companies have deployed AI, few have reworked the operating model or value proposition enough to capture it. This is a piece about what the operating partners in the 20% are actually doing.
Fund AI is not portfolio AI
The first distinction operating partners draw is between AI the fund uses on itself and AI that changes a portfolio company's economics. They are different programs with different owners. EQT's Motherbrain, the firm's long-running data platform for sourcing and screening deals, is fund-level AI: it improves origination, not a portfolio company's margin. Useful, but it does not move EBITDA at an asset.
Portfolio value creation is the other program, and it is the one that shows up at exit. It lives inside operating companies — their pricing, their cost to serve, their sales motion, their working capital — and it is owned by the operating partner and the portfolio CFO, not the deal team. Conflating the two is how a firm convinces itself it has an AI value-creation story when what it actually has is a better deal-sourcing funnel.
Where the EBITDA actually is
Operating partners map AI to the value-creation plan the same way they map any lever — by where it lands in the P&L and the balance sheet.
- 01Revenue — dynamic and predictive pricing, sales-rep effectiveness, lead scoring and conversion, churn prediction and retention, expansion targeting. The highest-conviction lever, and the one investors themselves rank first.
- 02Gross margin — service and support automation, procurement and spend analytics, supply-chain and inventory optimization, quality and yield.
- 03SG&A — finance close and reporting, HR and recruiting, legal review, customer support deflection, marketing production.
- 04Working capital — demand forecasting, receivables and collections prioritization, inventory right-sizing — often the fastest cash impact and the easiest to attribute.
There is a reason revenue leads. In Open Future Forum's investor research, when investors describe where AI is already producing value in their portfolios, “better products” leads at 51%, ahead of cutting costs at 35% and helping customers at 34%. That cuts against the reflexive PE instinct to treat AI as a cost-out program. The operators seeing value are describing a better offering, not just a leaner one.
The 20% changed the operating model, not the tool stack
The firms in the measurable-results minority did the harder thing: they changed how the business runs, not just what software it licenses. A pricing model is worth nothing until the commercial team is allowed to reprice; a support agent saves nothing until the staffing plan and the service levels are rebuilt around it. This is the same conversion problem finance leaders describe elsewhere — capacity created by AI only becomes value when the operating model is redesigned to bank it.
That is why buy-and-deploy underperforms buy-and-transform. The tool is the cheap part. The operating-partner work — reworking the process, the targets, the incentives and the headcount plan so the tool's output actually changes an outcome — is the expensive part, and it is what the 20% paid for.
The question operating partners skip: what could AI invalidate?
Value-creation content almost always asks where AI can add value. The sharper operating partners ask the opposite question about the thesis they underwrote: which parts of this investment could AI erode?
- 01Product substitution — could a general-purpose model do most of what this company charges for?
- 02Software commoditization — is the moat a workflow that is now cheap to rebuild?
- 03Labor-based revenue — does the model bill for hours that AI compresses?
- 04Pricing pressure — will customers expect AI-driven cost to be passed through?
- 05Cyber and technical debt — does faster AI adoption widen the attack surface faster than controls?
For a fund with a defined hold period, the downside question is not pessimism; it is underwriting. A thesis that assumes AI is only upside is an incomplete thesis. The operating partners who benchmark AI across a portfolio are, in effect, running an early-warning system for which assets the technology helps and which it quietly threatens.
The real prize: a repeatable portfolio capability
The individual use case is not the durable advantage. Any single company can hire a consultant and buy a pricing tool. The advantage a sponsor can build that a standalone company cannot is a repeatable capability across the portfolio: centralized AI expertise, benchmarking of what worked at one company against another, common vendor standards and portfolio-wide contracts, and playbooks that turn a one-off win into a default first move at the next acquisition.
This is where firms with real operating DNA — the Vista Equity, Thoma Bravo and Bain Capital lineage of operating teams — start ahead, and why platforms that benchmark AI maturity across a portfolio are spreading. Alvarez & Marsal's 2026 value-creation work frames the near-term prize the same way: move from pilots to embedded operating routines, starting where outcomes are measurable and building the data quality, talent and process maturity to scale. The operating partner's edge in 2026 is not tool selection. It is being the seat that can deploy the same win five times.
Last updated: August 19, 2026
Frequently Asked Questions
Where private equity operators compare AI playbooks
Open Future Forum's Private Equity Executive Forum convenes sponsors, operating partners and portfolio executives on the deal-to-value work, including where AI actually moves EBITDA. Global, founded in Silicon Valley.