AI's Productivity Gains Aren't Reaching the P&L

Big companies are pulling away on AI agents. The gap isn't about budget — it's about data plumbing.

This is an analysis piece built on McKinsey's published 2026 survey data; the framing and conclusions are Edgewisely's own.

Eighty-nine percent of organizations now use AI regularly in at least one business function. Eighty percent of individual workers say it's made them more productive. And yet only 37% of organizations say AI has meaningfully moved their EBIT — essentially flat from a year earlier, according to McKinsey's "The State of AI in 2026: On the Road to ROI", based on a survey of 1,719 respondents across 97 countries fielded in May and June 2026. Just 6% of organizations qualify as AI "high performers" — attributing at least 5% of EBIT to AI and describing the impact as significant, per the same report, covered separately by The Register.

The gap that matters more than the headline number

Buried inside that survey is a split that says more about where AI value actually accrues than the adoption headline does: 40% of organizations with more than $1 billion in annual revenue report scaling AI agents in one or more functions, up sharply from 27% a year earlier. Smaller organizations, by contrast, stayed essentially flat at 22%, according to McKinsey's data as reported by Forbes. Large enterprises are pulling away from smaller ones on agent adoption at exactly the moment agents are supposed to be the technology that levels the playing field between teams with big engineering budgets and teams without.

Why size is winning, not just budget

The obvious explanation is money — larger companies can afford more experimentation, more failed pilots, more specialized hires. But the more durable explanation is data plumbing. Scaling an AI agent past a demo requires clean, accessible, well-governed data connected to real systems of record: CRM, ERP, ticketing, HR, finance. Most large enterprises have spent a decade and tens of millions of dollars building exactly that infrastructure for other reasons — compliance, reporting, prior software rollouts. Smaller organizations often haven't, and an AI agent bolted onto messy, disconnected data doesn't fail because the model is bad; it fails because the agent can't reliably see or act on the information it needs.

For enterprise buyers

If you run AI strategy at a large company, the McKinsey data is permission to keep investing in agent scaling — the gap between your peers who are pulling ahead and those who aren't is real and growing. But it also means the return isn't coming primarily from the model layer, which every competitor can buy from the same handful of vendors. It's coming from whichever team gets its underlying data and workflow infrastructure agent-ready first. That's an argument for prioritizing data governance and system integration work over swapping in a marginally better model.

For smaller companies and startups

The flat 22% figure is a warning less about AI's usefulness at smaller scale and more about what's missing: most smaller organizations don't have a dedicated platform team whose job is making internal data agent-accessible. The startups and small teams that do break out of the 22% plateau tend to be the ones that treat data integration as a first-class engineering problem from day one, rather than retrofitting it after buying an agent product that then can't actually reach anything useful.

Why productivity gains aren't showing up in EBIT

The disconnect between 80% reporting personal productivity gains and only 37% reporting EBIT impact is the report's most important finding, and it has a simple explanation: individual time savings don't automatically become company-level financial results unless the organization redesigns the workflow around the saved time. A support agent who resolves tickets 20% faster doesn't move EBIT unless the company reduces headcount, handles more volume with the same headcount, or redeploys the freed time to higher-value work — and McKinsey's own data shows workforce reduction from AI is happening at less than half the rate companies predicted a year ago (14% report a decline in workforce size from AI versus 32% who expected one). Productivity gains are real and are simply evaporating into unchanged processes instead of showing up on an income statement.

For related reporting on where enterprise AI spending is and isn't paying off, see Edgewisely's coverage of why most enterprise AI pilots are quietly dying and why hospital AI pilots still don't pay off.

The takeaway

The AI ROI story in 2026 isn't a story about whether the technology works — 80% individual productivity gains settle that question. It's a story about which organizations have already built the plumbing to convert that productivity into measurable financial return, and which haven't. That plumbing correlates heavily with company size today, but it isn't actually about size — it's about whether data integration and workflow redesign got funded years before anyone needed an AI agent. Companies waiting for a better model to close that gap are waiting for the wrong thing.

The model was never the bottleneck. The bottleneck was always whether anyone had bothered to make the data reachable.

Frequently Asked Questions

What is the gap between AI productivity gains and financial ROI that McKinsey found?

McKinsey's 2026 survey found that 80% of individual workers report AI has improved their personal productivity, but only 37% of organizations say AI has meaningfully contributed to their EBIT, a figure essentially unchanged from the prior year. Just 6% of organizations qualify as AI "high performers" who attribute at least 5% of EBIT impact to AI.

Why are large companies scaling AI agents faster than small ones?

McKinsey found that 40% of organizations with over $1 billion in annual revenue report scaling AI agents, up from 27% a year earlier, while smaller organizations remained flat at 22%. The main driver isn't just budget — it's that large enterprises typically already have the clean, connected data infrastructure agents need to operate reliably, built over years for other reasons like compliance and reporting.

Why doesn't individual productivity automatically translate into company profit?

Time savings at the individual level only move EBIT if a company redesigns its workflows around those savings — by reducing headcount, handling more volume with the same staff, or redirecting freed time to higher-value work. McKinsey's data shows workforce reductions attributed to AI are happening at less than half the rate companies predicted a year earlier, meaning the saved time is often absorbed rather than converted into measurable financial impact.

What should smaller companies do to close the AI ROI gap?

Smaller companies should prioritize data integration and workflow infrastructure before investing further in AI agent products, since agents fail in practice most often because they can't reliably access the data and systems they need, not because the underlying models are inadequate. Treating data plumbing as a first-class engineering priority is the most direct lever available to smaller teams.

Editor's note — sources: McKinsey "The State of AI in 2026: On the Road to ROI," The Register, Forbes coverage of McKinsey's agent-scaling data.