Deloitte, State of AI in the Enterprise 2026: “The untapped edge”
TL;DR
The Deloitte AI Institute’s annual survey, January 2026. 3,235 director-to-C-suite leaders in 24 countries, surveyed August to September 2025. The Netherlands is in the sample (n=50).
- A quarter have scaled pilots. “Only 25% of respondents said their organization has moved 40% or more of their AI experiments into production to date”; 54% expect to reach that level within three to six months (p. 4, p. 8).
- A fifth govern agents maturely. “1 in 5 (21%) report having a mature model for governance of autonomous agents” (p. 6, p. 20), while 74% expect to use agentic AI at least moderately within two years, against 23% today (p. 17).
- A quarter report transformative effect. 25% of leaders say AI is having a “transformative effect” on their company, up from 12% a year earlier (p. 10). Only 20% are growing revenue through AI today; 74% hope to (p. 10).
- Three thirds of depth. 34% are “starting to use AI to deeply transform” the business, 30% are redesigning key processes, and 37% use AI “at a surface level” with little or no process change (p. 4).
- Jobs are not redesigned. 84% of companies have not redesigned jobs around AI, and insufficient worker skills are the biggest barrier to integrating AI (p. 5, p. 13).
Key data
| Measure | Figure | Unit | Page |
|---|---|---|---|
| Moved 40% or more of AI experiments into production | 25% | respondents (for their organisation) | 4, 8 |
| Expect to reach that within 3-6 months | 54% | respondents | 8 |
| Mature model for governance of autonomous agents | 21% | companies surveyed | 6, 20 |
| Use agentic AI at least moderately: today / in two years | 23% / 74% | companies | 17 |
| AI having a transformative effect | 25% (last year 12%) | leaders | 10 |
| Increasing revenue through AI: achieving / hoping to | 20% / 74% | organisations | 10 |
| Deep transformation / process redesign / surface use | 34% / 30% / 37% | companies | 4 |
| Have not redesigned jobs around AI | 84% | companies | 5, 13 |
| Workers with sanctioned AI access, a year ago to now | under 40% to “around 60%” (p. 4) or “under 60%” (p. 8) | workers | 4, 8 |
| Strategy highly prepared for AI / risk and governance | 42% / 30% | companies | 7 |
Method and definitions
- Sample: 3,235 leaders, “director-level to C-suite-level,” including board members, presidents and vice presidents. Half IT leaders, half line-of-business leaders. Six industries, 24 countries. Fieldwork August to September 2025. Fifteen executive interviews added (pp. 7, 40).
- Countries: the United States is the largest group (n=1,200, 37%). Netherlands n=50. Others include the UK (220), India (200), Canada (175), Germany (170), France (150) (p. 40).
- Selection: “All participating organizations have one or more working implementations of AI being used daily” (p. 40). The sample is AI users, not all firms. Respondents had to influence or make AI decisions, or oversee implementations.
- The 25% question: “In your estimation, what percentage of your AI experiments (e.g., pilots, test cases, etc.) have been deployed to date into your organization (moved into production)?” (p. 8). The threshold of 40% is Deloitte’s cut-off on that estimate.
- “Mature model for governance” of agents is not defined beyond the label.
- Unit: each figure is a share of respondents answering for their organisation. With one respondent per organisation this is close to a share of organisations; the report does not say whether that holds.
Scope and reliability
Self-report by AI decision-makers in organisations that already use AI daily. That selection pushes adoption figures up, and makes the low shares (25%, 21%) more striking, not less. Deloitte sells AI services, and the report closes with actions for leaders. The sample is large, the country breakdown is published, the key question is printed, and the series runs year on year. The Dutch press release (21 January 2026) gives the 21% as “only 21% of those companies”, meaning those planning agentic AI; the report gives it as 21% of companies surveyed (p. 6, p. 20).
Dynamic-capabilities reading
digital-seizing/balancing-digital-portfolios: the pilot-to-production rate is a portfolio measure; the Dutch press release calls its absence “pilot fatigue.”digital-transforming/improving-digital-maturity: the split into deep transformation (34%), process redesign (30%) and surface use (37%) is a three-level maturity reading.contextual/internal-barriers: insufficient worker skills are named the biggest barrier, and 84% have not redesigned jobs around AI.
How it connects
- enterprise-ai-adoption and the thread ai-maturity-measurement-comparison: a mid-strictness instrument. It counts respondents in AI-using organisations, with a pilot-to-production cut-off at 40%.
- IBM CEO Study 2025: the same pilot-to-scale topic, counted as initiatives (16% scaled enterprise-wide).
- Rendtorff-Smith (OECD) on BCG: the OECD names skills, data and managerial processes as the bottleneck. Deloitte’s leaders name worker skills first.
- agent-oversight-and-delegation: 21% report mature agent governance; the report asks organisations to set “clear boundaries for agent autonomy” and human approval points (p. 20).
- ai-sovereignty: 77% say the location of AI development is a key factor when choosing technology (p. 5).
- ai-agents: agentic AI use expected to rise from 23% to 74% in two years.
What was actually ingested
The full global PDF from Deloitte’s own asset URL, converted with pdftotext; overview, key findings, agentic-AI, preparedness and methodology pages read in full, physical-AI and sovereign-AI sections skimmed. The Deloitte Netherlands press release of 21 January 2026 was read in full. The five industry cut-outs on the same page were not acquired. A different Deloitte instrument, the 2026 Global Health Care Outlook, is a separate source in this batch.
Linked entities and concepts
- Deloitte — publisher (Deloitte AI Institute).
- Deloitte, Deloitte AI Institute: publisher, no entity page yet (see the batch’s entity proposals).
- enterprise-ai-adoption, agent-oversight-and-delegation, ai-sovereignty, ai-agents: concepts.
Dangling (single-source mention, deferred): Nitin Mittal (Deloitte global AI lead, quoted), Jorg Schalekamp (AI lead, Deloitte Netherlands, quoted).