Deloitte State of AI in the Enterprise 2026: The Untapped Edge

Originally published by Deloitte. Source: The State of AI in the Enterprise: The Untapped Edge – Deloitte, January 2026

---

Released in Davos in January 2026, Deloitte’s annual State of AI in the Enterprise report captures where organizations actually stand—not where they say they stand. Based on 3,235 director-to-C-suite business and IT leaders across 24 countries, surveyed in August–September 2025, the findings reveal a sector still operating well below its potential.

One-Third Deep. One-Third Surface Level.

When it comes to AI integration, organizations split almost evenly into thirds:

• 34% are using AI to deeply transform—creating new products and services, reinventing core processes, or reshaping business models
• 30% are redesigning key processes around AI
• 37% are using AI at a surface level, with little or no change to existing processes

That means nearly two in five organizations are experimenting with AI without fundamentally changing how they work. The title of the report—The Untapped Edge—reflects exactly this: the gap between what organizations have access to and what they’re actually doing with it.

Worker Access Jumped 50%

One of the most significant movements in 2025 was access. Worker access to sanctioned AI tools grew from under 40% to approximately 60% of the workforce within a single year. That’s a meaningful expansion in reach.

But access alone doesn’t create value. Only 25% of respondents have moved 40% or more of their AI pilots into production—though 54% expect to reach that threshold within three to six months. The bottleneck isn’t availability; it’s activation.

Agentic AI: High Ambition, Low Governance Maturity

Autonomous AI agents are the next frontier nearly every organization is racing toward:

• Nearly 75% of companies plan to deploy autonomous AI agents within two years
• 85% expect to customize those agents for specific business needs
• Yet only 21% of companies planning agentic AI deployment report mature agent governance models

This combination—rapid deployment ambition paired with immature governance—is one of the report’s most urgent warnings. Organizations are moving faster than their risk frameworks can support.

Physical AI, Sovereign AI, and Geopolitical Considerations

The 2026 survey also captures emerging dimensions that weren’t prominent in prior years:

• 58% of companies currently use physical AI (AI in robotics, manufacturing, and physical systems)
• 83% consider sovereign AI strategically important
• 77% now factor country of origin into vendor selection decisions

The geopolitical dimension of AI strategy—who makes the models your enterprise depends on—has become a board-level concern in a way it wasn’t twelve months ago.

What Separates Leaders from the Rest

Deloitte’s Global AI Leader Nitin Mittal put it plainly: “Organizations are pivoting from experimentation toward integrating AI into core business operations, focusing on scale and measurable impact.”

US Head of AI Jim Rowan added: “Successful AI organizations invest in both automation tools and talent development, empowering teams to adopt reimagined business models.”

The pattern among high performers is consistent: they’re not just deploying more AI—they’re rebuilding how work is organized around it, investing in talent alongside technology, and measuring outcomes from the start.

The organizations still at the surface level face a compounding disadvantage. Every quarter they wait, the gap between them and the transformation-oriented third widens.

---

What This Research Misses

The one-third / one-third / one-third breakdown doesn’t measure transformation quality. The 34% “deep transformation” tier is self-reported — organizations classify themselves based on their own perception of depth. Research on enterprise technology adoption consistently finds that organizations overestimate how fundamentally they’ve changed vs. how much they’ve added a capability layer on top of existing processes. Capgemini’s 2026 AI research specifically warns that organizations need to be more deliberate about governance, skills, and accountability to realize “full transformative value” — implying most “transformation” claims are premature.

75% planning agentic AI with only 21% mature governance is a governance crisis in the making. Deloitte flags this tension but doesn’t fully explore the liability implications. The EU AI Act (effective in stages from 2024–2027) classifies many agentic AI applications as “high risk” requiring conformity assessments, technical documentation, and human oversight — obligations that 79% of organizations deploying agents would fail under current governance maturity levels. This regulatory gap is absent from the report’s framing.

The production deployment statistic (25% with 40%+ of pilots in production) tells a story the report underplays. If 75% of companies have not reached production scale with AI — despite years of investment — the bottleneck is structural, not incremental. Slalom’s 2026 research is more specific about why: only 42% of organizations have robust data foundations, and 87% still rely on annual investment models that can’t support the continuous adaptation AI deployment requires.

Physical AI and sovereign AI data may be inflated by definitional flexibility. “58% currently use physical AI” is a striking figure. But the definition of physical AI ranges from industrial robotics (a decades-old category) to newer AI-native autonomous systems. Similarly, “83% consider sovereign AI strategically important” may reflect awareness rather than action — KPMG’s Global Tech Report 2026 finds the gap between stated AI priorities and actual implementation is consistently large across firms.

---

Research Methodology: 3,235 director-to-C-suite business and IT leaders across 24 countries. Industries: Consumer; Energy, Resources & Industrials; Financial Services; Life Sciences & Healthcare; Technology, Media & Telecom; Government & Public Services. Fielded August–September 2025.

Source: Deloitte State of AI in the Enterprise 2026