PE CxO AI Report · August 2026

There’s No Right Answer: AI

If you read the AI research that came out this summer, you get told opposite things. One set of reports says the risk now is spending too little. Another says most of what companies already spend cannot be measured, cannot be traced, or is going to waste. Both are backed by real data. Sometimes the same firm says both, three weeks apart.

There is not a clean answer in here. Nobody has one. The number that turns out to be right — how much to spend, how fast to move, when to hold — will not be knowable until later, when you can look back and see which companies guessed well.

For a PE-backed CxO the pressure is sharper, because the question your sponsor and board are really asking is whether AI is creating enterprise value or whether you are reporting activity. The companies pulling ahead are the ones that can see their own costs and trace their own results. That visibility, not the size of the budget, is what will separate the winners once the numbers are finally in.

Inside this issue

Confidence has come loose from results — Accenture & EY
Spend more, or the caution costs you — BCG
Prove it first, because most of the spend is invisible — McKinsey & KPMG
Almost everyone is adopting; almost nobody is in production — Forrester
The gains are real, but not in the P&L — S&P Global & IBM
In private equity, conviction is ahead of evidence — Bain

Confidence has come loose from results

Accenture’s Pulse of Change found 86% of leaders planning to increase AI investment and 78% calling it a driver of revenue growth, but only 32% reporting sustained, enterprise-wide impact. Fewer than one in five have rebuilt a process around AI.

EY’s AI Pulse survey found 98% of leaders whose companies invest in AI saying they have seen positive ROI, yet actual spending came in well below what the same population projected a year earlier. A year ago, 35% expected to be spending $10 million or more by now; only 23% actually are. Nearly everyone reports a return, and at the same time they are quietly pulling money back.

Source: Accenture, “Pulse of Change,” June 24, 2026; EY, “AI Pulse Survey, Fifth Wave,” July 28, 2026.

Spend more, or the caution costs you

BCG argues the bigger risk in AI now is underspending. Companies with deeper AI integration are seeing three times the cost reduction of their peers and higher return on invested capital, and the leaders are putting more behind it — about 1.7% of revenue against 0.8% for the rest.

The report’s position is that running small pilots and waiting to prove each dollar before scaling is not caution, it is a way to fall behind while competitors move budget toward what works.

Source: BCG, “Three Ways Leaders Can Invest Smarter in AI,” July 31, 2026.

Prove it first, because most of the spend is invisible

McKinsey’s QuantumBlack team looked at what AI spending actually buys and found a lot of it going to work nobody planned for. In agentic systems, the expensive part is not the first answer. It is the checking, correcting and re-running that follows, which they put at around 60% of the cost of an agentic coding task.

KPMG’s Global AI Pulse points the same way: 79% of leaders call AI a top investment priority, but only 7% say they have established a measurable return. The advice to spend faster runs straight into the problem that most companies cannot see where the money they are already spending goes.

Source: McKinsey & Company, “Cost versus value: Managing agentic AI system performance,” July 8, 2026; KPMG, “Global AI Pulse: Q2 2026,” July 2026.

Almost everyone is adopting; almost nobody is in production

Forrester’s State of Agentic AI describes a wide gap between saying you use AI and actually running it. About three-quarters of enterprise leaders report they are adopting agentic AI. Only a small share have it running in real production beyond chatbots, and true multi-agent systems are rarer still. Forrester’s read is that the technology has arrived and enterprise readiness has not caught up.

Source: Forrester, “The State of Agentic AI, 2026,” June 9, 2026.

The gains are real, but they are not showing up in the P&L

S&P Global’s AI and labor report found companies still increasing AI investment across every region it tracks, while the net employment effect turned slightly negative over the past year. IBM reports the same disconnect from the finance side: 79% of leaders see productivity gains, but only about 29% say they can measure the return with confidence.

Productivity improves, headcount pressure shows up, and the financial return stays hard to pin down.

Source: S&P Global, “The AI and Labor Landscape 2026,” June 30, 2026; IBM, “How to Maximize AI ROI in 2026,” June 12, 2026.

In private equity, conviction is ahead of evidence

Bain’s 2026 Midyear Report found GPs reporting the clearest AI returns in deal sourcing and diligence — the front end of the deal — while inside portfolio companies the benefits skew to cost savings and nearly 40% of GPs do not expect material financial impact from AI in 2026. Firms are naming AI a top priority and rotating capital toward assets they see as less exposed to AI disruption at the same time.

Source: Bain & Company, “2026 Private Equity Midyear Report,” Hugh MacArthur et al., June 8, 2026.

Where this leaves you

Spend more and prove it first are both defensible. Which one was right for your company will be clear later, not now. The goal is not to guess it correctly.

The goal is to be able to change your mind fast when you learn you guessed wrong. That comes down to one thing you can settle this quarter, without waiting for the research to agree: whether you can see your own AI costs and trace your own results. Build that first. Then spend as aggressively as it lets you defend, and size each bet so a wrong guess is cheap to unwind.

Don’t try to predict the right AI investment. Build the visibility that lets you move the investment the moment you are proven wrong.

Schedule a Consult → Read the PE CxO Report