Only 37% of Firms See EBIT Gains From AI as Returns Plateau

Only about a third of organizations — 37% — can attribute any EBIT impact to AI, and that figure has been flat since 2025, according to McKinsey & Company's "State of AI in 2026: On the road to ROI" survey. For HR and people-operations leaders budgeting AI spend for 2027, the survey synthesis signals a workforce technology market where adoption is nearly universal but financial returns remain rare.

The aggregate picture comes from seven 2025–2026 research releases: the U.S. Census Bureau's Business Trends and Outlook Survey (BTOS), McKinsey's survey of 1,719 global participants, Deloitte's poll of 3,235 business leaders, PwC's survey of 300 senior executives, IBM's CEO study, a Gartner survey of 1,303 enterprise respondents, and Google Cloud's survey of 2,403 executives conducted with the National Research Group.

How widespread is AI use right now?

Adoption is broad but uneven by firm size. BTOS data collected between December 2025 and May 2026 shows large firms with at least 20 employees are the heaviest AI users, and another 20% to 23% of firms plan to start using AI by the end of 2026. Information, finance and insurance lead sector-level adoption.

McKinsey found that 89% of its 1,719 global respondents use AI in at least one business function. Just over 50% of large enterprises — those with at least $1 billion in revenue — reported scaling AI across their organization.

What is the scaling bottleneck?

Deployment, not experimentation, is the choke point. Only 25% of the 3,235 business leaders Deloitte surveyed have moved 40% or more of their AI experiments into production. Gartner's 2026 survey of enterprises with at least $50 million in annual revenue found an even narrower result: just 22% have successfully scaled AI across multiple business units.

Put differently, roughly three out of four organizations are still stuck between pilot and production — a gap that shapes hiring, upskilling and change-management demands on people teams.

Why does agentic AI matter for governance?

AI agents drew C-suite attention in 2026. Deloitte found 74% of companies plan to deploy agentic AI within two years, yet only 21% have a mature governance model for autonomous agents — a 53-point gap between intent and oversight.

IBM's "Rewiring the C-suite" report found 72% of CEOs predict agentic AI will be a primary growth driver by 2030. PwC's survey of 300 senior executives identified the top business use cases for agentic AI, which closely mirror Gartner's top three overall AI use cases: cybersecurity threat detection and response, IT service desk automation and automated code generation.

Gartner's Q3 2026 Quarterly Emerging Risk Report, based on 308 enterprise risk management leaders, flagged AI discovery of cyber vulnerabilities, AI strategy miscalculation and AI-driven competitive displacement as emerging concerns. McKinsey also found that enterprises it identifies as high performers are more likely to mitigate AI-driven exploitation of vulnerabilities.

Both Deloitte and IBM found that 83% of business leaders consider AI sovereignty at least moderately important, if not essential, to their AI strategies.

What does the ROI data actually show?

Expectations are outrunning results. Deloitte found 74% of organizations hoped to grow revenue through AI, but only 20% achieved it — fewer than one in five. Google Cloud's research puts a similar number on the problem: only 26% of enterprises reported accelerating year-over-year returns, the group Google dubbed "AI ROI leaders."

IBM's 2026 CEO Study found nearly 80% of executives expect AI to drive significant revenue by 2030, but only 24% know where that revenue will come from. Even so, commitment keeps rising: Deloitte reported 84% of organizations are increasing AI investments, and 78% of leaders expressed greater confidence in AI.

What should employers watch for 2027?

The data shows no dramatic gap between large-enterprise adoption and everyone else's. Most organizations attribute some return to AI, EBIT figures are plateauing, and nearly 20% of organizations report inconsistent, plateauing or declining AI ROI.

The synthesis offers one central prediction: AI has found its foothold in the enterprise, but without specialized tooling its progress will continue to plateau, and deployments will grow more expensive for work largely comparable to what people already do. As 2027 planning begins, most organizations should not feel behind the curve — but leaders will need to scrutinize whether their AI investments deliver more benefit than burden.