Are CEOs actually seeing business gains from AI investments?

PwC's 29th Global CEO Survey, based on responses from "more than 4,000 chief executives across 95 countries and territories," finds that "56% saying they have seen neither lower costs nor higher revenue" from AI use in their business. Only a third of CEOs report increased revenue from AI in the past 12 months, and just over a quarter say costs are lower because of it. The survey also finds "only a relatively small proportion of CEOs are applying AI to a large extent" in areas like demand generation and support services — benefit realization is lagging adoption even at the CEO's own strategic level, not only on the front line.

The report frames this as a maturity problem rather than a technology failure: "'Clearly,' says the CEO report, 'we are in the early stages of the AI era.'" PwC ties the finding to its own companion data — its 2025 Global Workforce Survey, which found only 14% of employees were using generative AI on a daily basis — treating low daily usage and low benefit realization as two readings of the same early-stage adoption curve. The same survey records a drop in CEO confidence about near-term revenue growth generally (30% very/extremely confident, down from 38% last year and well off the 56% peak in 2022), though the excerpt attributes that broader confidence decline to tariffs, cyber exposure, and macroeconomic volatility rather than to AI specifically.

This sits in tension with Do AI productivity gains feel larger than they actually measure?, which found executives' perceived AI gains outrunning measured ones. PwC's broader, more general CEO sample shows the opposite bias at scale: a majority report no benefit at all rather than an inflated one — suggesting Baslandze's optimism-gap pattern may be concentrated in the high-skill, finance-heavy sectors it sampled rather than universal across industries. The finding also parallels Why do workers feel confident with AI but get poor results? at a different organizational level: WalkMe shows individual workers overestimating day-to-day reliability, while PwC shows CEOs underestimating or simply not yet seeing aggregate payoff — confidence and benefit realization diverging in different directions depending on where in the organization you ask. Why do ready workers struggle to transform their work? offers a candidate explanation PwC's excerpt doesn't test directly: that organizational and management factors, not individual AI use, are the bottleneck on realized returns.

The excerpt does not establish why benefits are lagging — it does not isolate implementation quality, time-to-value lag, measurement difficulty, or genuine lack of use-case fit as causes, and it does not break the 56% figure down by industry, company size, or region. It also does not show whether the CEOs who do report benefits used AI differently, nor whether "applying AI to a large extent" is the same bar across the surveyed firms. The finding supports, at the strength of a single cross-sectional self-report survey, that AI's business case remains largely unrealized for most large firms as of early 2026 — a caution against treating CEO AI investment announcements as evidence of CEO AI payoff.