Private equity has declared AI its top strategic priority. GPs and operating partners rank it above revenue growth initiatives, cost reduction programs, and M&A activity. It is the first item in every value creation framework and the centerpiece of nearly every LP presentation.
It is also the ninth most frequently used value creation tool out of ten.
That is the finding of FTI Consulting's 2025 PE Value Creation Index, which surveyed more than 500 PE decision-makers. Despite unanimous agreement on AI's importance, only 23–28% of firms report using it frequently as a value creation lever. The gap between what firms say about AI and what they actually do with it is the defining execution failure in private equity right now.
Understanding why that gap exists - and where it doesn't - is more useful than optimism about AI's potential.
The Industry Is Investing in the Wrong Kind of AI
Most PE AI investment is concentrated at the deal level. Sourcing platforms, diligence acceleration tools, portfolio monitoring dashboards - these have become standard infrastructure at larger firms. EQT's Motherbrain scans 50 million companies to surface investment targets. Hebbia compresses due diligence from 90 days to 21. AlphaSense is used by roughly 80% of top PE firms for market research.
These tools are genuinely valuable. They make deal teams faster and more productive.
What they don't do is improve the performance of portfolio companies. That requires a different kind of AI entirely - predictive tools deployed inside businesses to improve operational decisions: demand forecasting, inventory planning, workforce scheduling, capital allocation. This is the AI that lives inside the income statement rather than the deal team's workflow.
The industry has heavily invested in the former. It has largely ignored the latter.
The Numbers Are Worse Than You'd Expect
The data on portfolio company AI adoption is sobering regardless of which source you use.
McKinsey's 2026 Global Private Markets Report found that while roughly 60% of portfolio companies are experimenting with AI, only about 5% have scaled it to production with measurable results. Bain's 2025 Global PE Report - drawing on investors representing $3.2 trillion in AUM - found that only 20% of portfolio companies have operationalized any AI use case with concrete outcomes. Accenture's analysis of approximately 40 portfolio companies found that 90% of AI use cases never move beyond the pilot stage.
A 2025 MIT study put the broadest frame on it: despite $30–40 billion in global enterprise AI investment, 95% of companies have seen little to no P&L impact.
These numbers describe a consistent pattern: pilots that never reach production, dashboards that describe the past without improving decisions, and technology investments that exist in strategy decks but not in operating results.
Why Portfolio Companies Get Stuck
The failure is rarely algorithmic. It is almost always foundational.
The most common barrier is data infrastructure. Only 13% of business leaders believe their data architecture is well-suited for AI, according to an AWS/Harvard Business Review study of 623 executives. Most middle-market portfolio companies have operational data scattered across disconnected systems - legacy ERPs, departmental spreadsheets, point-of-sale platforms that don't communicate. Before any predictive model can function, that data has to be cleaned, integrated, and structured. That work is unglamorous and chronically under-resourced.
The second barrier is structural. FTI found that 40% of PE firms manage AI at the portfolio company level in a fully decentralized model - meaning each company is largely on its own to evaluate, procure, and implement AI tools. For a portfolio company CFO running a 10-person finance team, that is an unrealistic ask. The firms with the strongest operational AI track records - Blackstone, Vista Equity, Apollo - succeed partly because they provide centralized infrastructure, shared playbooks, and dedicated technical support across the portfolio.
The third barrier is misdiagnosis. Many portfolio companies deploy analytics tools in the form of dashboards and call it AI. Dashboards have value for reporting and accountability. What they don't do is improve decisions about the future. BI tool adoption has been stuck at 25–29% of employees for seven straight years. The reason isn't poor implementation - it's that dashboards answer the wrong question. Most operational decisions don't require a better view of the past. They require a credible view of what comes next.
Where the Real Opportunity Is
The highest-impact operational AI applications in PE-backed companies are specific, not generic.
McKinsey research shows demand forecasting reduces forecast errors by 20–50% and can decrease stockout losses by up to 65%. Workforce scheduling in seasonal service businesses - HVAC, landscaping, hospitality, field services - can reduce labor cost inefficiency by 10–20% when staffing decisions are driven by predicted demand rather than historical averages. Seasonal modeling for businesses with 200–600% demand swings between peak and off-peak periods can materially improve revenue capture, margin, and working capital - but only if the forecast incorporates external signals like weather data and regional demand trends, not just last year's numbers.
These are the decisions that move EBITDA. They are also the decisions most portfolio companies are still making with spreadsheets.
The ambition-execution gap in PE AI is real, but it is not inevitable. Closing it starts with investing in the right kind of AI - not the kind that improves how you find deals, but the kind that improves what the deals are worth.
ARC Consulting helps private equity firms and portfolio companies build predictive analytics systems focused on operational decisions. Get in touch.