As AI responsibility shifts to the C-suite, a new question is emerging: Do senior leaders have an inflated view of how advanced their companies really are? That tension is especially relevant in 2026 as executive leaders are under more pressure to show results from their AI investments.
A new survey of 1,000 global senior technology and data leaders shows that despite plenty of investment, most organizations are confronting a roadblock that keeps AI implementations in the realm of personal productivity gains rather than unlocking enterprise-wise ROI. The report, commissioned by autonomous knowledge platform Teradata, reveals key reasons why agentic AI stalls at the enterprise level, and the specific issues organizations must solve to achieve measurable returns.
The difference between AI for organizations and AI for individuals
A key challenge to enterprise AI adoption is conflating the productivity gains individual workers using AI can achieve with enterprise-wide efficiency gains and returns. It’s not especially difficult for an individual worker or a small pilot team to build an AI agent that improves a contained workflow, but translating those agents to work at enterprise scale is significantly more difficult.
Not everyone considers this difference when they’re evaluating AI adoption in an organization. This discrepancy in understanding may help explain one of the report’s findings—that 69% of C-suite executives say their organization is already operating with agentic AI, while only 57% of VPs say the same.
Personal AI gains come from tools like custom chatbots that help people (or small teams) do their work more efficiently. But any benefits organizations will realize from enterprise-wide AI implementations come from connecting agents to core business systems. That requires shared knowledge, a balance of appropriate access levels, and well-designed governance.
Why do enterprise AI initiatives stall?
Forty percent of respondents say more than 40% of AI pilot projects pause before production because infrastructure is not ready for autonomy. And 78% of leaders report difficulty creating a connected data foundation for agentic AI to act on. Enterprise AI is stalling because the foundation underneath it was built to inform humans, not to ground autonomous systems operating at scale.
A core challenge in connecting agents to enterprise data is context fragmentation. Organizations have tremendous amounts of data, but typically the data has been built and organized for humans who can interpret the broader context and meaning around that data. Now, it has to be contextualized for AI agents. As the report summarizes: “Getting context right means knowing which data, assembled in what order, on behalf of which process, needs to reach an agent at the moment of a decision.”
The difference between getting value from personal AI and enterprise AI is the same issue keeping organizational pilots from advancing: Agentic AI pilots are run on clean, well-organized data. But when organizations try to deploy a successful pilot into production, they stall out, because it’s normal within enterprises to have data that means different things across systems. Billing counts accounts, finance counts entities, marketing counts emails. While each is locally correct, an agent moving across these systems has no way to know they disagree. There’s a lack of shared meaning. Even superbly organized data usually hasn’t been formatted and contextualized for agentic AI to act on across an entire enterprise.
The maturity shift that unlocks enterprise ROI
The hard part isn't spending the money. The report found that 90% of leaders expect to increase their agentic AI investments in the next year. The bigger task is converting that spend into something a CFO recognizes as a return.
The report contextualizes how close organizations are to realizing ROI from investments using an Agentic AI Maturity Index. There are four stages: Experimenting, Developing, Building and Operationalizing.
Companies in the Experimenting stage are running pilots, and they’re trying to determine how AI is working for their teams and how to expand these successes more broadly. Those in the Developing stage have taken things a step further by establishing some successful models, but they haven’t yet connected those to autonomous knowledge. At these levels, companies have yet to make the leap from personal AI gains to organizational-wide AI gains.
In the Building stage, organizations have achieved structured local governance and have automated basic workflows. They still lack the data foundations and a context layer that can support agents making autonomous decisions across enterprise functions, but they can realize initial returns from automation within certain systems. In the Operationalizing stage, companies have effectively harmonized enterprise systems with the context and governance for AI agents to make autonomous decisions reliably. The Operationalizing stage is what unlocks robust enterprise ROI from agentic AI investments.
Operationalizing agentic AI unlocks enterprise ROI
The report shows that 68% of organizations are stuck in the Experimenting and Developing phases of the Agentic AI Maturity Index. Only 25% have reached the Building phase, where some enterprise-level returns start to emerge, and just 7% of organizations have reached the Operationalizing stage, where significant ROI from reliable autonomous decision-making gets unlocked.
That maturity gap shows up as a measurement gap, too: The report shows 62% of leaders prioritize enterprise-wide ROI over individual productivity gains when evaluating AI, but only 30% actually use margin improvement as a success metric. Meanwhile, 63% admit seeing no more than an emerging, minimal positive return on their AI investments to date. Taken together, these findings indicate that organizations claiming to optimize for enterprise returns but measuring something else will keep funding work that feels successful without ever producing a number a CFO recognizes.
Pilots are useful for identifying potential areas for ROI. Actually achieving enterprise-level returns from agentic AI requires that companies connect data across the organization and build the context layer that lets AI agents act on their own across teams, with a consistency and accuracy that organizations can rely on.