Enterprise AI is spreading, but clear financial returns remain elusive, according to McKinsey’s latest global survey. The findings also point to a shift in software buying: nearly a third of respondents say their organisation has passed on a product or feature because it could build something similar with AI coding tools.
Watch Desk analysis
What happened
A report by Diginomica on the survey says 40% of respondents at organisations with more than $1 billion in revenue report scaling AI agents, up from 27% last year. Among smaller organisations, the figure is flat at 22%. The survey also finds that 20% of companies are scaling software coding agents, rising to 31% among larger enterprises.
The Diginomica report says 37% of respondents attribute at least some positive EBIT impact to AI, roughly unchanged from last year; about 6% qualify as high performers, attributing at least 5% of EBIT to AI and describing its impact as significant. Meanwhile, 60% expect their organisation to increase AI investment over the next year, even as about 20% say AI-related costs are constraining its use.
Key findings
- Agent adoption is skewed towards larger organisations
40% report scaling AI agents at firms above $1 billion in revenue, against 22% at smaller organisations. - Coding agents are changing some software-buying decisions
Nearly a third of respondents say their organisation has opted not to buy at least one product or feature it could build with agentic coding tools. - Reported financial impact remains limited
37% attribute some positive EBIT impact to AI, while the high-performer share remains about 6%.
Why it matters
The survey describes two trends moving at different speeds: organisations are scaling AI and reconsidering what they buy, while reported financial gains remain patchy. That makes “we’re investing more” a poor stand-in for “this is paying off”. It also gives software suppliers a reason to watch whether coding agents become a credible alternative to buying packaged tools, rather than just a promising internal experiment.
Our read
The most useful signal is the gap between adoption and returns. Building a feature in-house may save a purchase, but it still has to be maintained and deliver value. Treat these results as a snapshot of what respondents say, not a universal scorecard for enterprise AI. The survey’s headline numbers are interesting; the business case still has homework.
What to watch
- Whether more organisations report measurable financial returns as deployments scale.
- Whether in-house AI-built software becomes a sustained alternative to buying products.
- Whether AI costs keep constraining use despite plans to increase investment.
Discussion spark: If AI coding tools let a company build a feature instead of buying it, what evidence should it require before calling that a genuine saving?
Sources and evidence
- Want to hear more about AI adoption? (4 November 2026, 11:05 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.