Enterprise AI Adoption — US, EU, and Global Business Use (2023–2026)

The most important fact about enterprise AI adoption statistics is that three widely-cited figures — 23.8% of US businesses, 20.0% of EU enterprises, and 88% of organizations in McKinsey's global survey — are not measuring the same thing. Presenting them as a single trend would be analytically misleading. Each figure comes from a different survey instrument, a different question framing, and a different sampling frame, and understanding why they diverge tells a decision-maker more about the state of AI deployment than any one headline figure on its own.

The US Census Bureau's Business Trends and Outlook Survey (BTOS) provides the most statistically rigorous baseline. Its sample is drawn from the Census Bureau's Business Register and its estimates are firm-weighted, so the denominator includes the small retailers, contractors, and service firms that make up most of the US business count. Under the original question — whether a business used AI "in producing goods or services" — 3.7% of US businesses reported AI use in September 2023, 6.0% in September 2024, and 10.0% in September 2025. That near-tripling in two years is strong evidence of genuine deployment momentum at the firm level. Under the broader question introduced in November 2025, the rate has since climbed from 17.3% to 23.8% in the survey collected in September 2026.

In November 2025, the Census Bureau revised the BTOS question to cover AI use "in any of its business functions", and the reported US adoption rate moved from 10.0% to 17.3%. That step is not comparable to earlier readings. Analysts extending this series should treat November 2025 as a methodological break, not a month of exceptional real-world adoption.

The EU Eurostat figure — rising from 8.1% in 2023 to 13.5% in 2024 and 20.0% in 2025 — comes from the annual Survey on ICT Usage and E-Commerce in Enterprises, which covers all EU-27 member states and firms with 10 or more employees outside the financial sector. The Eurostat definition of "AI technologies" is broad, spanning text mining, speech recognition, image recognition, machine learning, and AI-based process automation. Under the legacy BTOS question the EU rate ran well above the US one; under the revised question the two now sit in a similar range, though differences in firm-size coverage and sector scope mean the comparison is indicative only. The 2025 figure of 20.0% represents an 11.9-percentage-point increase in two years, a meaningful shift given the structural inertia of an enterprise-count-weighted survey across 27 member states.

The McKinsey State of AI survey operates in a different register entirely. The share of respondents saying their organizations have adopted AI in at least one business function was 55% in the survey fielded in April 2023, 78% in July 2024, and 88% in the survey fielded from June to July 2025. The question is broad enough to capture a single department's use of an AI tool. The respondent pool leans toward large organizations — 38% of 2025 respondents work at organizations with more than $1 billion in revenue — and results are weighted by each country's share of global GDP, which likely pushes the reported adoption rate well above what a firm-count-weighted survey of the full business population would show. The McKinsey series is best understood as a leading indicator for large enterprise behaviour — not as a representative cross-section of global business activity.

Sectoral concentration within the US data sharpens the picture further. In the September 2026 BTOS collection, the Information sector reports 49.0% AI use, about three times the rate for Retail Trade (16.5%) and more than double the rate for Manufacturing (19.5%). Professional, Scientific, and Technical Services (43.9%) and Finance and Insurance (41.0%) also sit well above the all-business rate of 23.8%. This stratification suggests that aggregate adoption figures understate how far AI is already embedded in the information and professional services economy, while overstating its reach into goods-producing and consumer-facing businesses.

In the EU, the more economically significant stratification runs by firm size rather than sector. Eurostat's 2025 data shows large enterprises (250+ employees) at 55.0% adoption — more than three times the rate for small enterprises with 10 to 49 employees (17.0%). Medium enterprises (50–249 employees) sit at 30.4%. The practical implication for corporate strategy and M&A analysis is that EU firms with under 50 employees are mostly at early-adoption or pre-adoption stages, while large-cap European corporates have already cleared the initial deployment threshold and are differentiating on depth of integration rather than presence or absence of AI.

Key series characteristics for professional use:

• BTOS (US): Biweekly Business Register-based survey, firm-weighted; legacy "producing goods or services" question through September 2025, broader "any of its business functions" question from November 2025; the chart shows September readings for the legacy series and the latest readings (December 2025, September 2026) for the revised one.
• Eurostat isoc_eb_ai: Annual enterprise-count-weighted survey, EU-27, enterprises with 10+ employees excluding the financial sector; broad "AI technologies" definition; most recent data 2025.
• McKinsey State of AI: Annual executive survey weighted by national GDP; organization-level question on AI use in at least one function; respondent pool skews large; figures likely overstate representative global business adoption by a significant margin.

For investment analysis, the BTOS series is the most defensible for modelling US AI penetration, particularly for mid-market and SMB-oriented business models where large-firm surveys are uninformative. The Eurostat series provides the most comparable cross-country framework for EU exposure assessments, with the firm-size breakdown offering the clearest segmentation for target company analysis. The McKinsey series, despite its sampling limitations, tracks directional momentum within the large-enterprise cohort most likely to drive near-term enterprise software revenues and AI platform market share.