Enterprise AI just got a confidence boost. This time the number crunchers have receipts.
KPMG’s latest Global AI Pulse survey, released in late June, found that 22% of organizations now say AI is fully part of everyday work. That’s up from just 13% a quarter earlier — the sharpest single-quarter jump anywhere on the firm’s adoption curve. AI also held its place as a top investment priority for 79% of leaders, up from 74% in Q1. Average planned spending held steady too, at $188 million per organization.
That’s a fast swing for a sector that spent most of 2025 fielding “is this a bubble” questions.
The Maturity Effect
KPMG polled 2,145 senior business leaders across 20 countries. The pattern that emerges isn’t just optimism. It’s a shift in what people are optimistic about. Leaders are moving past whether AI works and into who’s accountable when it doesn’t.
The accountability data backs that up. Organizations where the CEO is directly accountable for AI-driven decisions report far higher confidence in their AI strategy — roughly 60% versus 22% for those without that structure — and are more than three times as likely to report established ROI. Cost discipline tells a similar story. Companies with full visibility into what their AI systems actually cost to run are five times more likely to report established ROI than those flying blind.
Translation: the tech stopped being the bottleneck. Who owns the outcome — and who’s watching the meter — became the bottleneck.
Why the Meter Suddenly Matters
That “watching the meter” problem has a specific, underappreciated cause: the industry’s shift from flat-rate AI contracts to usage-based, token-metered pricing. Vendors are moving enterprise customers onto consumption billing for compute, inference, and agent activity. Costs that used to be predictable line items are turning into variable ones, and a lot of companies weren’t built to forecast them. KPMG found only 26% of organizations have real-time visibility into their AI operating costs. Another 42% have only partial visibility. And 35% say cost management and economic literacy around usage-based pricing is itself a barrier to deployment.
The pain isn’t hypothetical. Uber’s engineering org reportedly burned through its entire 2026 AI budget in four months, largely on coding-agent usage, and had to cap individual employee spending on those tools. It’s a preview of what metered billing looks like once nobody’s watching the dial.
The C-Suite Isn’t Reading From the Same Page
Underneath the aggregate optimism, there’s a persistent executive split. A separate Protiviti survey of 852 global C-suite executives, conducted with the University of Oxford, found CIOs and other tech leaders far more confident that AI is driving revenue growth than CEOs and board members — 61% versus roughly a third. Protiviti’s own read: the gap tracks C-suite alignment on what success looks like more than it tracks AI maturity itself.
That’s not a rounding error. That’s two executive teams looking at the same deployment and reaching opposite conclusions about whether it’s paying off. It also lines up with KPMG’s own finding that only 7% of leaders can currently point to established ROI, even as a quarter of them face growing investor pressure to prove it.
Where the Optimism Might Be Getting Ahead of Itself
Not everyone’s buying the acceleration story. Deutsche Bank’s Jim Reid cautioned this week that real, measurable AI productivity gains at the macro level are likely still years out, even as he expects the technology to eventually reshape the labor market. That read sits uncomfortably next to KPMG’s confidence numbers. If the productivity payoff is genuinely years away, then what’s driving this quarter’s confidence jump isn’t results. It’s belief that results are coming — belief backed by governance structures that make it feel earned.
That distinction matters more than it sounds. Confidence built on visible ROI and clean cost accounting is durable. Confidence built on faith in a future payoff, one still running on usage-based bills nobody fully understands yet, is the kind that evaporates the moment a downturn forces a hard budget review.
The next KPMG pulse check will land in a market already jumpy about AI capex. Whether the CEO-CIO confidence gap finally closes, or whether the usage-based cost shock catches up with the optimism first, is the thing actually worth watching next quarter.
Related: The $1.6 Trillion AI Bet Has a Timing Problem — And Markets Are Starting to Notice
