Deloitte’s latest State of AI in the Enterprise report — drawing on responses from more than 3,200 senior leaders across 24 countries — lands with a stark verdict: most companies have AI tools. Far fewer have any idea what to do with them. Workforce access to AI has grown by 50% in a single year, rising from under 40% to roughly 60% of employees. Yet among those workers who actually have access, fewer than 60% use AI in their daily workflow. Wider access, it turns out, has not translated into deeper impact.
Two tiers, one technology
The report reveals a widening divide. One-third of companies (34%) are using AI to reshape their products, processes, and business models fundamentally. Another third (30%) are redesigning workflows but leaving their underlying models intact. The remaining 37% are barely scratching the surface — using AI cosmetically, with little change to how they actually operate. All three groups are capturing efficiency gains. Only the first is capturing the future.
The pilot purgatory problem
A revealing data point: just 25% of companies have moved 40% or more of their AI experiments into production. Yet 54% expect to reach that level within three to six months. The report calls this the “proof-of-concept trap” — pilots run cleanly in controlled conditions, then collide with the messy realities of integration, compliance, and scale. One healthcare AI leader quoted in the report puts it bluntly. Without a coherent strategy, organizations end up chasing the next shiny object, running a hundred pilots, and creating zero value.
Jobs are next — but companies aren’t ready.
The workforce disruption the industry has theorized about is now imminent. A full 36% of companies expect at least 10% of their roles to be fully automated within a year; 82% expect the same within three years. And yet, 84% of companies have not redesigned a single job to leverage AI capabilities. Most are investing in training employees to use AI tools — far fewer are rethinking roles, career paths, or organizational structures from the ground up.
Three emerging frontiers
Deloitte identifies three forces reshaping the next chapter of enterprise AI. Sovereign AI — designing and deploying AI within national borders — has become a boardroom issue, with 77% of companies now factoring a vendor’s country of origin into their technology decisions. Agentic AI is moving fast: 74% of companies plan to deploy autonomous AI agents within two years, yet only 21% have mature governance frameworks to manage them. And physical AI — robotics, autonomous vehicles, smart sensors — is already in use at 58% of companies, with adoption expected to hit 80% within two years, led by Asia Pacific.
The bottom line
The report’s central argument is uncomfortable but clear: the biggest barrier to AI’s transformational potential is not the technology itself — it is the gap between access and activation. Companies that treat AI as a foundational operating principle, rather than an efficiency layer bolted onto legacy processes, are pulling ahead. Those still running pilots without a roadmap to production risk being left behind — not by a competitor’s model, but by their own indecision.