Covering Scientific & Technical AI

Deloitte’s latest State of the AI report shows that AI adoption continues to accelerate rapidly, however data infrastructure, governance and talent redesign are lagging significantly. While enterprises remain confident strategically, they are operationally not prepared to achieve their AI goals. This widening execution gap is a core theme across this year’s report.

The numbers shared by Deloitte show access to AI tools has expanded dramatically – up 50% year over year. Now 60% of employees have access to such tools, but fewer than 60% of them regularly use them.

Deployment tells a similar story. Just 25% of organizations have converted 40% or more of their pilots into production systems. However, more than half believe they will cross that threshold within months. If that acceleration materializes, it will test the data infrastructure, integration layers, and governance frameworks in ways that isolated pilots never did.

Does scaling alone guarantee transformation? Not according to Deloitte. The report’s findings show that even as AI deployments increase, the depth of change across organizations varies significantly. Around 25% of surveyed leaders now describe AI as having a transformative effect on their organizations – that more than doubles last year’s figure.

While investment levels continue to rise and executive confidence is strengthening, only 34% of companies are reimagining products, services, or business models around AI. Another one third are redesigning key processes without altering the broader structure of the business. The remaining third are layering AI onto existing systems with limited structural change. This shows that efficiency gains are becoming widespread, but reinvention remains selective.

As companies move toward agentic AI, that pace of change becomes even more important. The Deloitte report shows that nearly three quarters of organizations plan to deploy autonomous agents within the next couple of years, but only 21% report having the proper governance in place for those systems.

Unlike earlier AI models that offered recommendations and played more of a supporting role, AI agents are designed to execute decisions directly. We know that capability can help expand operational reach and improve efficiency, but it also increases exposure.

The respondents shared that they are most concerned about security (73%) and data privacy (73%). This is followed by lack of governance oversight and whether models are reliable 50%.

It’s not surprising that data privacy and security remain chief concerns at 73% each. Nearly half of respondents also worry about governance oversight and whether models are reliable and explainable.

(Yossakorn Kaewwannarat/Shutterstock)

The issue of the execution gap is further reinforced by the survey’s preparedness indicators. Just around 40% of the respondents shared that their AI strategy is highly prepared. Governance trails at 30%. Technical infrastructure readiness reaches 43%, data management 40%, and talent readiness falls to only 20%. These numbers have decreased compared to last year’s report indicating that organizations are less prepared for the AI goals in 2026. It could also mean that maybe they are setting too ambitious goals and in that context that are not as prepared.

The implications of that gap extend beyond infrastructure and governance. They reach into the workforce itself.

The report highlights that just 20% of organizations say their talent is highly prepared for AI. At the same time, around a third of the respondents expect meaningful automation within a year. However, most companies have focused on training rather than restructuring how work gets done. This means that employees are being educated about how to use tools without actually reworking how work gets done using these tools.

(WINEXA/Shutterstock)

Another key finding of the report is that many organizations expect revenue growth from AI, but few have actually realized this at scale. This suggests that the industry is still in a value transition phase where efficiency gains are more tangible than top-line expansion.

Overall, this year’s report shows that AI adoption is no longer the main challenge. Most enterprises are investing and planning to scale – but the pressure now is operational. Data systems, governance models, and workforce structures are being asked to support more automation and more autonomy than they were originally designed for. In many cases, those foundations are still catching up.

The next stage of enterprise AI will depend less on how fast companies deploy tools and more on how well they integrate them. Efficiency gains are visible, but sustained revenue growth and structural change take longer. As AI becomes more embedded in everyday operations, the organizations that slow down enough to strengthen their infrastructure and rethink how work gets done may ultimately move further ahead.

This article first appeared on BigDATAwire.

Related