AI statistics that matter in 2026
AI numbers are often quoted without a source, and then they start to drift. This page lists the statistics I use myself in working sessions and essays, each with its original source. Updated August 2026.
26%
of companies get past the proof-of-concept and generate tangible value from AI
Survey of 1,000 executives in 59 countries: 74% had yet to demonstrate tangible value from AI. 22% were starting to realise value at scale and 4% already generated substantial value. The problem is rarely the model; it is the gap between pilot and production. I therefore do not use the more widely quoted "95% of AI pilots fail" (MIT NANDA, 2025) as a current statistic: that figure counts non-adopters as failures and its method was never released.
14%
average productivity gain with an AI assistant
Measured across roughly 5,000 customer support agents at a Fortune 500 software company. The least experienced workers gained 35%; top performers close to nothing. AI lifts the floor, not the ceiling.
r = −0.68
correlation between frequent AI use and critical thinking
Gerlich found a strong negative association among 666 participants, mediated by cognitive offloading: the more often you outsource the thinking, the less you train the skill. A correlation, not a causal verdict, but the mechanism rings true.
74.4%
of Dutch SMEs have integrated AI into their operations
That puts the Netherlands ahead of the rest of Europe. Fast adoption says nothing about controlled use; the same research warns precisely against uncontrolled deployment.
±70%
of the European cloud market is held by three American hyperscalers
Amazon, Microsoft and Google combined. Relevant for every sovereignty debate: the infrastructure profits of the AI wave largely land outside Europe, however high Dutch adoption gets.
€35M / 7%
maximum fine under the EU AI Act
For prohibited AI practices: up to €35 million or 7% of global annual turnover, whichever is higher. SMEs and start-ups face proportionally lower fines. Deadlines and risk tiers are on my EU AI Act page.
Every number links to its original source. Quote freely, with the source.
What the numbers say together
The productivity gain of AI is real, but it rarely lands in the P&L by itself. The gap sits between pilot and production, and between working faster and thinking better. Take both seriously and you are in the 26%.
Want to know where your AI work stands?
In the Scan I put your pilots next to these numbers: ROI potential, production risk and a verdict per pilot. One week, fixed price.