The AI Mandate–Adoption Gap: Why Adoption Is Slow and ROI Is Invisible
Almost every sponsor conversation I've had this year has come around to AI.
Almost every sponsor conversation I've had this year has come around to AI. Boards are telling portfolio companies to "deploy AI," and more and more sponsors are writing it straight into the value creation plan. Then you walk the floor at the portfolio company, and very little has changed. Pilots have stalled. The licenses are paid for and mostly unused. And when someone asks the CFO where the EBITDA from AI shows up, the room gets quiet.
The data backs up what I'm hearing. Grant Thornton's 2026 AI Impact Survey found that 45% of private equity respondents are still piloting AI, 11 points above the cross-industry rate. Only 5% have fully integrated AI into operations, compared with 14% across all industries, and only 24% report revenue growth from it. FTI Consulting's 2026 Private Equity AI Radar, which surveyed 200 senior decision-makers, found that only 36% of portfolio companies are deploying AI in production, just 7% have reached enterprise scale, and 43% have not implemented AI in any meaningful way.
The part that should change the conversation comes next. Among the portfolio companies that actually moved AI into production, FTI found that 95% met or exceeded their business case.
Read that again. When AI makes it into the work, it pays. The technology isn't the problem. The problem is everything between the mandate and the workflow.
A mandate is not an operating plan
A mandate tells people what to do. It does not tell them the three things they actually need to know: who is accountable, where the AI fits in the work, and how anyone will know it's working. Leave those three open and you get exactly what the surveys describe: lots of activity, very little result.
Nobody owns it
The typical pattern goes like this. The mandate lands with the CEO. The CEO hands it to the CIO. IT runs a vendor bake-off, picks a tool, and rolls it out. Somewhere along the way, the person who owns the P&L line the tool was supposed to move drops out of the conversation.
Attention at the top isn't the issue. Grant Thornton notes that board-level involvement in AI is above average in private equity. But attention at the top without ownership in the middle produces pilots, not results. The VP of procurement has a quarterly savings target, a supplier base to manage, and a team that's already stretched. If the AI tool is "IT's project," it will lose to the day job every time.
It's bolted on, not built in
McKinsey's State of AI 2026 survey of 1,719 participants found that 37% of respondents attribute at least some EBIT impact to AI, about the same share as last year. AI high performers, defined as those attributing 5% or more of EBIT to AI and reporting significant value, have stayed flat at about 6% of respondents.
What separates that 6%? According to McKinsey, nearly three-quarters of high performers report fundamentally redesigning workflows because of their AI use. Just one-quarter of other respondents report doing so.
That's the whole story in one comparison. A tool layered on top of a broken process gives you a faster broken process. Take spend analytics in procurement. An AI tool that flags off-contract spend in real time is useless if no one has the authority to redirect that spend, and if the buyer's day is still consumed by expediting late orders. The insight shows up in a dashboard. The behavior never changes. Neither does the P&L.
Nobody defined "working"
If you can't name the financial line before launch, you won't find it after. Too many pilots are measured on adoption, usage, or hours saved. Hours saved that never leave the cost base aren't EBITDA. They're a nicer afternoon.
Governance is weak too. Only 9% of PE respondents in the Grant Thornton survey said they were very confident they could pass an AI governance audit within 90 days, the lowest rate of any sector surveyed. If you can't show how a use case is governed, you almost certainly can't show what it returned.
Closing the gap
None of this requires a bigger AI budget. It requires running AI like any other value creation initiative. In practice, that comes down to four disciplines.
Put a P&L owner's name on every use case. Not IT's name. The procurement leader owns the sourcing agent. The plant manager owns predictive maintenance. The head of logistics owns the freight-audit model. IT enables; operations owns.
Start with the workflow, not the tool. Map how the work gets done today. Decide what changes: who makes the decision, which handoffs disappear, what the person stops doing because the AI now does it. If nothing about the job changes, nothing about the result will either.
Write the number down first. Every use case gets a baseline, a target, a specific financial line (cost of goods, freight, inventory days, overtime, working capital), and a date. Finance signs off on the baseline before launch, not after.
Scale it or stop it at 90 days. A pilot that can't show movement on its line in a quarter gets redesigned or shut down. A pilot that can gets funded, staffed, and rolled out. Letting pilots drift is how a portfolio company ends up with fifteen experiments and zero EBITDA.
Supply chain and operations are a good place to start. The data already exists in ERP, procurement, and warehouse systems. The financial lines are clear. And the results show up quickly in cost and cash, which is where a sponsor needs to see them.
Why this matters now
Hold periods are long and the return math is harder than it used to be. Buyers are also learning to tell the difference between a stack of scattered pilots and AI that is embedded, governed, and evidenced in the numbers. The first is a talking point in a CIM. The second is something a buyer can underwrite.
The sponsors who close the gap won't be the ones with the most pilots. They'll be the ones who can point to a specific use case, a named owner, and a line on the P&L that moved.
If your board asked tomorrow which AI use case moved EBITDA last quarter, and by how much, would anyone in the company be able to answer?
David Jadwin is Managing Director and Head of Private Capital at Maine Pointe, where he works with private equity firms, family offices, lenders, and their portfolio companies on supply chain and operations value creation.
Sources
- Grant Thornton, "Private equity insights: 2026 AI Impact Survey Report" (published April 21, 2026; survey of 950 leaders fielded Feb. 23–March 18, 2026; PE subgroup of 100 respondents, which Grant Thornton notes is directional)
- FTI Consulting, "AI is Already Creating Winners in Private Equity" (2026 Private Equity AI Radar) (published May 11, 2026; 200 senior decision-makers)
- McKinsey & Company, "The state of AI in 2026: On the road to ROI" (published August 25, 2026; 1,719 participants in 97 nations, fielded May 4–June 8, 2026)
Facing this in your business?
Tell us what you're trying to fix. A partner who has done it before will come back to you.
Talk to a partner



