TribuTrace

Insights

What 2026 surveys say about AI ROI in private equity

Private equity is not short of AI activity. What it is short of, in the public surveys of 2026, is a financial count that a finance owner will sign. This note is a reading of those surveys, not a new one. The figures below were checked against the live sources on November 10, 2026. Each number is third-party evidence, not a TribuTrace result, and not a promise about what a given portfolio company will earn. Several of the same figures appear as sourced context in the problem statement on the TribuTrace homepage.

How to read these surveys

The sources do not measure the same population, and they should not be averaged.

Accordion’s PE AI adoption benchmark is a Q2 2026 vendor survey of technology and AI operating partners. The sample size was not shown on the source page (Accordion, 2026). CloudZero surveyed 260 senior finance leaders, 135 of them CFOs, not a PE-only sample (CloudZero, 2026). Coupa’s Strategic CFO Report 2026 is a survey of 600 global finance leaders (Coupa, 2026). Bain and StepStone surveyed 103 investment and investor relations professionals, primarily in North America and Europe, from December 2025 to January 2026 (Bain and StepStone, 2026). Grant Thornton’s 2026 PE AI Impact Survey is reported on the firm’s site as a survey of private equity leaders; the sample size and fieldwork dates were not shown on the page we opened (Grant Thornton, 2026). FTI Consulting’s 2026 Private Equity AI Radar surveyed 200 PE decision-makers worldwide in December 2025, at firms with at least $1 billion in assets under management, and the results are self-assessed (FTI Consulting, 2026). McKinsey’s April 24, 2026 article draws on its latest Global Survey on AI, a cross-industry sample whose size was not stated in that article (McKinsey, April 24, 2026).

Vendor surveys should be read as vendor surveys. Self-assessed “targets met” is not the same as a line in the P&L. Where a source does not publish its sample size, that limit is part of the finding.

Measurement often stops before the P&L

Accordion’s 2026 benchmark is the clearest PE-specific picture of what gets measured today. Most of the operating partners in that survey are measuring something. They are often measuring the floor, not the financial result.

FTE impact or time savings was the most common measure, selected by 67%. Only 38% were measuring revenue or margin impact. Only 29% included AI impact in board and LP reporting (Accordion, 2026).

Those three figures sit together. Time saved is not meaningless, but it becomes a P&L result only if finance can show where the capacity went: a cost that came out, output that went up without adding staff, or work that stopped being done. The gap between 67% and 38% is a gap in the count that reaches EBITDA, and a further gap in what boards and limited partners actually read.

The same survey found that 47% of operating partners described a tool-by-tool, use-case-by-use-case approach with no shared measurement (Accordion, 2026). If each tool has its own dashboard, and none is reconciled to the same ledger, board reporting will stay thin.

Finance is being asked for a number it often cannot give

The finance surveys, which are broader than private equity, describe the same pressure from the CFO’s side of the table.

CloudZero’s 2026 survey found that only 22% of finance leaders could tie AI spend to business outcomes, while 87% said they needed to do so within the year (CloudZero, 2026). Two-thirds of boards, 66%, were conditioning further AI funding on proof of return, and 43% of finance leaders said they were already being asked for a number they could not give (CloudZero, 2026). Sixty percent said they were already spending more on AI than they could justify (CloudZero, 2026). The same report notes a split in who spends and who accounts: AI was driven outside finance in 74% of companies, while finance owned the bill in 60% (CloudZero, 2026).

Coupa’s 2026 Strategic CFO Report, from a survey of 600 global finance leaders, found that 76% said difficulty quantifying AI ROI was hindering further AI investment; the firm’s FAQ on the same page describes that difficulty as “moderately to significantly” hindering further investment (Coupa, 2026).

Read together, the finance surveys say something simple. The blank-check period is ending. The people who have to answer the board often cannot connect the spend to an outcome, and a large majority say that this gap is now in the way of further investment. They do not say what the return usually is. They say that it is frequently unmeasured.

Many GPs do not expect a material financial impact this year

On the sponsor side, the surveys describe activity that has not yet settled into a financial result.

Bain and StepStone’s GP outlook, from the 103-person survey described above, found that 39% of GPs did not expect AI to have any material financial impact on portfolio companies in 2026 (Bain and StepStone, 2026). Among expected primary outcomes in the portfolio for 2026, 46% pointed to cost savings or efficiency, and 10% pointed to revenue growth (Bain and StepStone, 2026). A further 3% expected a net negative impact, or costs of implementation that outweigh short-term gains (Bain and StepStone, 2026). These are expectations for the year, not measured results after the fact.

Grant Thornton’s 2026 PE AI Impact Survey is summarized on the firm’s site under the headline “PE is seeing AI activity, but not returns.” On that page, 24% of private equity leaders report revenue growth from AI, and 45% say they are still piloting, a rate the firm describes as 11 percentage points above the cross-industry figure (Grant Thornton, 2026). The sample size was not shown on the page we opened, so these figures should be used with that limit attached.

Bain, in a September 16, 2026 article on what it calls the “AI Value Paradox,” put the pattern in plain language: AI activity is everywhere in private equity, and meaningful value creation is not (Bain, September 16, 2026). The same article reports Bain’s CEO Survey 2026 (n=100): 82% of CEOs said their AI transformation was at best partially realizing its intended results (Bain, September 16, 2026). That CEO sample is cross-industry, not PE-only. A cross-industry figure from McKinsey points in the same direction: in its April 24, 2026 article, McKinsey reported that 60% of respondents to its latest Global Survey on AI had still not seen enterprise-wide EBIT impact from their AI programs (McKinsey, April 24, 2026).

Day-to-day use is still uneven

FTI Consulting’s 2026 Private Equity AI Radar is useful here if it is read carefully. The survey is of 200 decision-makers at larger firms, with at least $1 billion in AUM, and the answers are self-assessed (FTI Consulting, 2026). It is not a middle-market sample, and it should not be treated as one.

On that survey, only 36% of respondents said their portfolio companies were using AI in day-to-day operations; 7% reported comprehensive integration across the portfolio (FTI Consulting, 2026). Most of the cost-saving and revenue-generating programs in the study were aimed at a 5% to 10% improvement, not a transformation of the P&L (FTI Consulting, 2026).

Those figures are a brake on treating “AI in the portfolio” as already equivalent to “AI in the operating routine.” They are also a reason not to lean on self-assessed claims that programs have “met or exceeded targets” without asking who set the target, and against what baseline. We have not used those self-assessed target figures here, because the published summaries do not make the base of the percentage unambiguous.

What the surveys do not say

They do not tell you the return on the initiatives already running in a given portfolio company. A survey average cannot be copied into a board memo as if it were a baseline.

They do not tell you that buyers will pay more for “AI-enabled” operations. Accordion reported that 44% of operating partners said buyers were already asking about AI-enabled finance capability in diligence, and that only 9% had seen a demonstrable premium in a completed transaction (Accordion, 2026). That is evidence of questions, and of how rarely a price difference has been observed in that sample. It is not evidence of a premium you can underwrite.

They do not settle the method. McKinsey’s April 2026 article argues that AI impact is measurable if it is treated like any other capital investment, with a baseline, a comparison such as a staggered rollout, and a recurring review (McKinsey, April 24, 2026). The surveys above show how often that discipline is still missing. They do not substitute for it.

They do not license a contested claim that almost all AI pilots fail.

What an operating partner or a CFO can do with this

If you are an operating partner, the useful move is to stop asking whether “the portfolio is doing AI” and start asking, company by company, which initiatives can be counted. One shared method is worth more than another dashboard. The 29% of operating partners in Accordion’s 2026 survey who already include AI impact in board and LP reporting are the group with a reason to keep a ledger that rolls up (Accordion, 2026). The rest have a reporting gap that a fund-level slide will not close on its own.

If you are a portfolio-company CFO, the useful move is to own the number before the next board meeting asks for it. CloudZero’s 2026 survey is not a PE sample, but the situation it describes will be familiar to operating partners and portfolio-company CFOs: the spend is often incurred outside finance, and finance still has to speak to it. A short inventory, a reconstructed baseline where the data will support one, and a label of measured, estimated, projected, or not yet attributable on each line is a better answer than a vendor’s saving.

The surveys are consistent on the shape of the problem. Measurement often stops at time saved. Boards are asking for proof. A large share of GPs do not expect a material financial impact this year. None of that is a reason to invent a return. It is a reason to count the spend already on the books, and to say when data cannot support a claim.

If you want that count done once, for one portfolio company, the AI Value Baseline is the fixed-scope version of the work.

Sources

Figures were checked against the live pages or, for Bain and StepStone, the published PDF, on November 10, 2026.