Insights

AI Adoption Statistics for Established Businesses - everything you need in 2026

The latest AI adoption statistics for established businesses, including the gap between AI usage, mature rollout, employee adoption, and measurable returns.

Riaz Kanani ai ai adoption business strategy digital transformation

78% of organisations now use AI in at least one business function. Only 1% describe their rollout as mature. (McKinsey, 2025)

That gap is not a technology problem. It is a deployment problem.

Most of the conversation about AI adoption focuses on whether organisations are starting. The more important question is what happens after they start — and the data on that is considerably less encouraging than the headline adoption numbers suggest.

Here is what the research tells us about where AI adoption actually stands, why the numbers look better than the reality, and what the organisations seeing genuine returns are doing differently.

We update this page as new research becomes available.

Key statistics at a glance

  • 78% of organisations use AI in at least one business function (McKinsey, 2025)
  • Only 1% describe their AI rollout as “mature” with embedded AI across multiple functions (McKinsey, 2025)
  • Only 6% qualify as AI high performers seeing meaningful financial returns (McKinsey, 2025)
  • 85% of employees have the capability to use AI tools. Only 25% use them regularly (IBM, 2026)
  • 48% of executives describe their organisation’s AI adoption as a “massive disappointment” (Writer, 2026)
  • $301 billion in global AI spending in 2026 — with only a fraction generating returns commensurate with investment (IDC, 2026)

How far AI adoption has actually spread

The headline numbers on AI adoption look impressive. The underlying picture is more complicated.

  1. 78% of organisations report using AI in at least one business function — up from around 50% in 2022 (McKinsey Global AI Survey, 2025)
  2. Only 1% of those organisations describe their AI rollout as mature, with AI embedded across multiple business functions and generating consistent returns (McKinsey, 2025)
  3. Only 6% of organisations qualify as AI high performers — meaning they see meaningful, sustained financial results from their AI programmes (McKinsey, 2025)
  4. Larger enterprises are approximately twice as likely to have adopted AI as smaller companies — though adoption rate and return on that adoption are different things (McKinsey, 2025)
  5. SMB AI adoption reached 35% in 2025, up from 18% in 2023 — significant acceleration, but from a low base (Verizon Business, 2025)
  6. 40% of enterprises report they lack adequate AI expertise internally to execute the AI programmes they have committed to (McKinsey, 2025)
  7. Global AI investment reached $301 billion in 2026, up from $223 billion in 2025 — investment is growing faster than returns (IDC, 2026)

The 78% figure is technically accurate. But using AI in one business function — even successfully — is a long way from building the kind of embedded capability that shows up in financial results.

The gap between AI capability and AI use

The most underreported problem in AI adoption is not whether employees have access to AI tools. It is whether they actually use them. These two things have almost no correlation.

  1. 85% of employees have the technical capability to use AI tools their organisation has deployed. Only 25% use them on a regular basis — a 61-point gap (IBM Global CEO Study, 2026)
  2. 83% of CEOs say AI success depends more on people’s adoption of the technology than on the technology itself (IBM, 2026)
  3. Team resistance to AI tools ranks among the top three barriers to AI success across all company sizes (IBM, 2026)
  4. 79% of organisations report facing significant AI adoption challenges — a double-digit increase from the prior year (Writer, 2026)
  5. 39% of organisations have no formal plan to drive business value from the AI tools they have already deployed (Writer, 2026)
  6. 54% of C-suite executives say adopting AI is creating significant internal friction or conflict within their organisation (Writer, 2026)

Most organisations have solved the wrong problem. Getting the tool in front of employees is not the challenge. Getting employees to change how they work is the challenge — and it requires a different kind of investment than a software licence.

What organisations are actually seeing from their AI investments

The gap between AI spending and AI returns is one of the defining stories of 2025 and 2026.

  1. Only 39% of organisations that use AI report any measurable impact on their bottom line (McKinsey, 2025)
  2. 48% of executives describe their organisation’s AI adoption as a “massive disappointment” (Writer Enterprise AI Adoption Report, 2026)
  3. 75% of executives admit their organisation’s AI strategy is “more for show” than for genuine business transformation (Writer, 2026)
  4. 80% of companies using generative AI report no significant effect on their bottom line (McKinsey, 2025)
  5. Despite $301 billion in global AI spending in 2026, only a small fraction of organisations report financial returns commensurate with their investment (IDC, 2026)
  6. 42% of companies abandoned most of their AI initiatives in 2025, up from 17% in 2024 — a 147% increase in project abandonment (S&P Global, 2025)

The data does not suggest AI does not work. It suggests most organisations are not set up to make it work. Those are different problems with different solutions.

The skills and workforce dimension

The capability gap inside organisations is not just about today’s AI rollouts. It is about whether companies have the workforce to sustain the AI programmes they are building.

  1. Between 2026 and 2028, organisations expect 29% of their workforce to require reskilling for a different role entirely, and 53% to need upskilling to perform their current role more effectively with AI (IBM, 2026)
  2. 40% of enterprises acknowledge they do not have adequate AI expertise internally to execute their stated AI strategy (McKinsey, 2025)
  3. Only 6% of organisations have the combination of leadership alignment, data infrastructure, talent, and change management that characterises AI high performers (McKinsey, 2025)
  4. Companies that invested in workforce AI education and change management at the same level as technology deployment were significantly more likely to see positive returns within 12 months (IBM, 2026)
  5. The organisations seeing the best AI returns treat reskilling as an active, ongoing programme — not a one-time training event at the point of tool deployment (McKinsey, 2025)

What separates the 6% seeing real returns

McKinsey’s research consistently identifies a small set of behaviours that distinguish the organisations seeing real results from the majority generating activity without outcomes.

  1. High performers define what AI success looks like before deployment, with a specific measurable signal — not a general productivity goal (McKinsey, 2025)
  2. High performers redesign the workflow the AI will operate in before they select or deploy the AI — not after (McKinsey, 2025)
  3. High performers treat adoption as the primary implementation challenge and budget for change management at the same level as technology (IBM, 2026)
  4. High performers run fewer AI programmes simultaneously and build from demonstrated results rather than maximising the number of pilots underway (McKinsey, 2025)
  5. High performers maintain active, visible executive sponsorship throughout a programme — not just at the kickoff (McKinsey, 2025)
  6. High performers are 2.5x more likely to have a documented, consistent process for deciding which AI investments to pursue (McKinsey, 2025)
  7. High performers pilot inside real operational processes rather than alongside them — so what they test is what they will scale (McKinsey, 2025)

Why the adoption curve looks slow from the inside

The headline adoption rates suggest most organisations are moving quickly on AI. The lived experience inside most businesses feels considerably slower. Both things are true.

Adopting a tool is fast. Changing how a team works is not. The 85% capability / 25% usage gap is a direct reflection of this. Most organisations have moved fast on procurement and slow on the harder work of making AI part of how people actually do their jobs.

The organisations that close this gap tend to do one thing differently: they define what changed behaviour looks like before they buy anything. Not “our team will use AI” but “our account managers will prepare client briefings using AI, and the time saved will show up here.” That specificity is what separates adoption that compounds from adoption that stalls.

Final thought

The 78% adoption figure is real. So is the 1% maturity figure. The gap between them is where most of the wasted investment sits.

Getting to the other side of that gap is not primarily a technology question. It is a question of what you defined as success before you started, how you set up the people who need to change, and whether your leadership stayed engaged beyond the announcement.

For a practical framework on structuring AI investments that actually deliver, see Most AI Projects Don’t Fail Because of the Technology. And if you want a structured view of where your business stands before investing further, the AI Readiness Assessment is a useful starting point.

Sources include: McKinsey Global AI Survey (2025), IBM Global CEO Study (2026), Writer Enterprise AI Adoption Report (2026), IDC Worldwide AI Spending Guide (2026), S&P Global Market Intelligence (2025), Verizon Business AI Report (2025). This page is updated continuously as new research becomes available.