The most important finding across every credible 2026 survey is a gap rather than a number. Adoption is close to saturated and measured financial impact is not, and the distance between those two figures is the single most useful statistic for anyone planning AI investment. This page collects the adoption data from primary research sources, with each figure attributed and dated, and explains what the gap between adoption and impact actually indicates.
Headline Adoption Figures
Adoption has effectively stopped being a differentiator, which changes what these numbers are useful for. They establish that you are not early rather than telling you anything about advantage.
Organisations Using AI
(cite index=”3-1″>88% of organisations now use AI in at least one business function, up from 78% a year earlier, per McKinsey’s State of AI Global Survey published November 2025.</cite>
The Growth Rate Behind It
(cite index=”7-1″>Enterprise AI adoption grew 10 percentage points year over year in 2025 per McKinsey</cite>, which is the increase that took the figure from a majority to near-universal.
Adoption Versus Production Deployment
(cite index=”7-1″>Only about a third of organisations have scaled AI beyond pilots, which makes the adoption rate and the production deployment rate significantly different numbers.</cite>
What This Means for Planning
Using AI somewhere is now the baseline. Our AI consulting services engagements start from what is deployed rather than from whether anything is.
The Agent Adoption Gap
Agentic AI is where the distance between experimentation and deployment is widest, and the numbers are worth reading carefully because coverage frequently conflates the two.
Experimentation Versus Scaling
(cite index=”4-1″>McKinsey’s 2026 survey finds 62% of organisations at least experimenting with AI agents but only 23% scaling them in even one business function.</cite>
Actual Deployment Rates
(cite index=”3-1″>Fewer than 10% of organisations have deployed AI agents in any given business function, even as 88% now use AI somewhere, per Stanford HAI’s 2026 AI Index Report published April 2026.</cite>
The Forecast Against It
(cite index=”4-1″>Gartner projects 40% of enterprise applications will feature task-specific AI agents by the end of 2026, up from less than 5% in 2025.</cite>
The Cancellation Projection
(cite index=”3-1″>Gartner projects more than 40% of agentic AI projects will be cancelled by the end of 2027 due to escalating costs, unclear business value, or inadequate risk controls.</cite> Our AI agent development work scopes against exactly those three risks.
Financial Impact Data
This is where the reporting diverges most sharply from the adoption narrative, and it is the section worth reading if you read only one.
Enterprise-Level EBIT Impact
(cite index=”4-1″>Only 39% report enterprise-level EBIT impact from AI, and about 6% attribute more than 5% of EBIT to it.</cite>
Executive-Level Results
(cite index=”5-1″>PwC’s 2026 CEO Survey of 4,454 executives found only 12% of CEOs have achieved both revenue gain and cost reduction from AI.</cite>
Concentration of Returns
(cite index=”5-1″>PwC’s 2026 AI Performance Study found roughly 20% of companies capture nearly 74% of AI’s economic gains.</cite>
Spending Ahead of Deployment
(cite index=”4-1″>Menlo Ventures estimates enterprise generative AI spend reached about $37 billion in 2025, roughly triple the prior year</cite>, while deployment figures remain far lower.
What Separates the Organisations Getting Returns
The most actionable finding in the 2026 data is not an adoption figure. It is what distinguishes the minority capturing value, and it is consistent across sources.
Process Redesign Over Tool Adoption
(cite index=”4-1″>High performers are 2.8 times more likely to report having fundamentally redesigned the process rather than layering AI onto it, and the gap between adoption and financial impact tracks workflow redesign more closely than it tracks spending.</cite>
Data Quality as the Stated Barrier
(cite index=”7-1″>52% of businesses cite data quality and availability as the biggest barriers to AI adoption.</cite> Our data analytics work addresses this before AI delivery for that reason.
Governance Maturity Is Low
(cite index=”5-1″>McKinsey’s 2026 AI Trust Maturity Survey reports only about 30% of organisations reach maturity level three or higher in strategy, governance, and agentic AI controls.</cite>
The Implication for Investment
Spending without process redesign produces adoption without impact. Our business process automation work treats the redesign as the deliverable.
Sector and Function Data
Adoption varies meaningfully by function, and the engineering figures are the most reliable because the behaviour is directly measurable rather than self-reported.
Developer Tool Adoption
(cite index=”7-1″>84% of developers use AI coding tools per Stack Overflow’s 2025 Developer Survey, and GitHub Copilot is deployed at 90% of Fortune 100 technology companies.</cite>
Healthcare Adoption
(cite index=”7-1″>63% of physicians use AI tools and 80% of hospitals deploy AI in at least one function.</cite>
Marketing Adoption
(cite index=”7-1″>87% of marketers use generative AI in at least one workflow per Salesforce State of Marketing 2026.</cite>
Why Regulated Sectors Lag
(cite index=”5-1″>In healthcare, banking, insurance, legal, and government the blocker is audit, explainability, and accountability rather than capability.</cite> Our AI and automation engagements in those sectors begin with governance.
FAQs
What percentage of companies use AI?
88% of organisations use AI in at least one business function, up from 78% a year earlier, per McKinsey’s State of AI Global Survey from November 2025. However, only about a third have scaled beyond pilots into production deployment.
How many companies have actually deployed AI agents?
Fewer than 10% in any given business function, per Stanford HAI’s 2026 AI Index. McKinsey finds 62% at least experimenting with agents but only 23% scaling them in even one function, so experimentation and deployment are very different numbers.
Are companies making money from AI?
A minority. Only 39% report enterprise-level EBIT impact and about 6% attribute more than 5% of EBIT to AI. PwC’s 2026 CEO Survey found just 12% of CEOs achieved both revenue gain and cost reduction.
Why do so many AI projects fail?
Gartner projects more than 40% of agentic AI projects will be cancelled by end of 2027 due to escalating costs, unclear business value, or inadequate risk controls. Separately, 52% of businesses cite data quality and availability as the biggest barrier.
What separates organisations getting returns from AI?
Process redesign. High performers are 2.8 times more likely to have fundamentally redesigned the workflow rather than layering AI onto it, and the adoption-to-impact gap tracks redesign more closely than it tracks spending levels.
Which functions have the highest AI adoption?
Software engineering, where 84% of developers use AI coding tools, and marketing at 87% using generative AI in at least one workflow. Healthcare shows 63% of physicians and 80% of hospitals using AI in at least one function.
