Why Most Private Equity Firms Aren't Seeing Returns From Their AI Yet

Why Most Private Equity Firms Aren't Seeing Returns From Their AI Yet

Private equity has not been shy about spending on AI. Deloitte's inaugural 2025 GenAI in M&A Survey, which asked 1,000 senior corporate and PE leaders in the U.S. about their GenAI use, found that 88% of PE respondents have already invested $1 million or more in the technology for their deal teams, and most plan to increase that spending again over the next year. Boston Consulting Group's research on AI-first private equity firms, published in January 2026, delivers the less comfortable half of that story: despite the spending, "few firms can show meaningful returns from AI in many of their portfolio companies." The gap between the two findings is the actual story.

The Investment Is Real. The Returns Mostly Aren't.

Deloitte's numbers describe genuine enthusiasm. Eighty-six percent of PE and corporate respondents say they've already integrated GenAI into their M&A workflows, and 65% of them did that integration within the past year alone, meaning this is a very recent wave, not a slow multi-year rollout. Eighty-one percent of PE respondents expect measurable ROI within one to three years. That's a firm bet on a near-term payoff.

BCG's take, drawn from its work across hundreds of AI transformations, is that the bet mostly hasn't paid off yet, and it names a specific reason rather than a vague "it's early." According to BCG, the outlier firms showcased at PE conferences for their AI wins "remain the exception," and the reason most others aren't seeing results is that "although many PE-backed companies have deployed AI tools across the portfolio, few have done the hard work of transforming their operating model or value proposition to capture real value."

Deploy, Reshape, Invent: Why Handing Out Licenses Isn't Enough

BCG frames the gap using a three-stage framework it applies across its AI transformation work: deploy, reshape, invent.

Deploy is what most PE portfolio companies are actually doing: giving employees a license to a large language model and assuming productivity gains will follow on their own. BCG is direct about the ceiling here: "Deploying AI is worth doing. It will make some employees more productive. But it rarely creates measurable and meaningful value." That lines up with where Deloitte says GenAI adoption is concentrated in dealmaking today: M&A strategy and market assessment (40% of adopters), target identification and screening (35%), and due diligence (35%), all pre-sign activities where a tool assists an existing task rather than changing how the task gets done.

Reshape is the harder step firms are mostly skipping: rethinking roles, org structure, and the operating model itself so productivity gains are "real and scalable and flow through to the P&L." BCG's specific advice for PE is to write a single AI playbook that can be piloted at one portfolio company and then translated across others, even across different industries, rather than letting each portfolio company experiment in isolation. R&D, sales, marketing, and customer service are the functions BCG flags as the most promising early targets, both because of their size and because the AI tooling for those functions is already relatively mature.

Invent, the third stage, means building AI directly into the customer-facing product or business model. BCG calls this "high investment, high risk, and high reward" and explicitly says most companies should skip it: it's reserved for industries, like software, digital services, and media, where AI carries outsized upside or downside, and there's no standard playbook for it. For most PE portfolios, reshape is the stage that actually matters.

The Money Is Already Committed to Do the Harder Work

The Deloitte numbers make BCG's point sharper rather than softer: this isn't a funding problem. PE firms are already outspending corporates on GenAI for dealmaking (88% past the $1 million mark versus 77% of corporates), and more than half plan to increase that spending again in the next 12 months. The capital to reshape operating models is already committed, largely to deploy-stage tooling. The bottleneck BCG identifies isn't budget, it's the harder organizational work of redesigning how a portfolio company's teams actually operate, function by function, so the tools translate into P&L impact instead of just individual productivity that never shows up in the numbers investors are underwriting.

What This Means for PE Firms Going Forward

The firms getting real value from AI, per BCG, aren't the ones with the most licenses distributed across the portfolio. They're the ones treating reshape as a deliberate, sequenced program: pilot a new playbook at one portfolio company, confirm it moves the P&L, then carry that same playbook to other portfolio companies. Running that kind of test across several portfolio companies in parallel, as BCG recommends, is also how a firm tells the difference between AI that's genuinely changing how work gets done and AI that's just quietly running in the background without ever showing up in EBITDA. Given how much has already been spent on deploy-stage tools, the reshape work is the part of the AI bet PE firms haven't cashed in yet.

Frequently Asked Questions

Have private equity firms actually invested in AI, or is this mostly talk? The investment is real and well ahead of corporates: Deloitte's 2025 survey found 88% of PE respondents have put $1 million or more into GenAI specifically for their deal teams, versus 77% of corporate respondents, with most planning to increase that spending further over the next year.

Why aren't PE portfolio companies seeing returns despite the spending? BCG's research attributes this to most firms staying at the "deploy" stage, handing out AI tool licenses without changing how work actually gets done. Real returns require the harder step of reshaping roles and operating models so productivity gains flow through to the P&L, which few firms have done.

What is BCG's "deploy, reshape, invent" framework? It's a three-stage way of categorizing AI transformation. Deploy means distributing AI tools without organizational change. Reshape means redesigning roles and workflows so gains are measurable and scalable. Invent means building AI into the core customer product, which BCG reserves for a narrow set of industries with outsized AI upside.

Where in the deal process is GenAI adoption currently concentrated? Per Deloitte, adoption clusters in pre-sign activities: M&A strategy and market assessment (40% of adopters), target identification and screening (35%), and due diligence (35%). These are largely deploy-stage use cases rather than deeper operating-model changes.

What should a PE firm do differently to get real returns from AI? BCG recommends building a single AI playbook piloted at one portfolio company, confirming it produces measurable P&L impact, and then translating that playbook to other portfolio companies, even across different industries, rather than letting AI adoption stay tool-by-tool and company-by-company.