Denodo study finds 290% AI data layer return on investment

Denodo has published research by Veqtor8 on the financial impact of its AI data layer for enterprise agentic AI. The study found a 290% return on investment over three years.

The report is based on interviews with enterprise practitioners, hyperscalers and technology partners in sectors including banking, telecommunications, chemicals and logistics. Veqtor8 used those interviews to build a composite enterprise model, scaling from 100 agents in the first year to 900 by the third year across 100 distributed data sources.

The findings show that organisations using a governed AI data layer generated 2.9 times more business value from AI initiatives over the three-year period. Agent deployment was also three times faster, as teams could use pre-built data products instead of creating new integrations for each deployment.

The research also focused on operating costs. Veqtor8 found that AI operating costs fell by 68% when queries were routed through the data layer and data was filtered before model ingestion, reducing the context payload per task from about 18,000 tokens to about 5,760.

The report also pointed to savings in integration and governance work. Enterprises could avoid more than 75% of integration and governance costs, while maintenance costs fell by 79% and policy administration effort by 76%.

Integration pressure

The research examined what it described as MCP server sprawl, where enterprise agents need custom Model Context Protocol servers, retrieval-augmented generation pipelines and separate security controls for each underlying data source. In regulated sectors, a single production agent may need to interact with 30 to 50 cloud, software, legacy and operational systems.

That complexity has become more visible as companies move from generative AI pilot projects to broader production deployments involving autonomous and semi-autonomous agents. The study identified access to live, governed and semantically consistent enterprise data as an emerging constraint in those roll-outs.

Veqtor8 framed the issue as one of scale as much as cost. As the number of agents rises, enterprises face a growing volume of connectors, security rules and maintenance work if they rely on custom-built architecture for each use case.

"Agents are only as effective as the data they can reach," said Andrew Milroy, Chief Analyst, Veqtor8.

"Connecting an individual system is straightforward, but orchestrating dozens of systems into a coherent, reliable workflow under consistent security policies is extraordinarily complex. Denodo's AI Data Layer abstracts this complexity, allowing enterprises to profile data once and reuse governed data products across every agent entitled to access them," Milroy said.

Enterprise use

The findings are relevant to enterprises seeking to expand the use of AI agents while maintaining governance across distributed data estates. Denodo positions its platform as a layer that lets businesses unify data access, semantics and policy enforcement without moving large volumes of data into a single repository.

The study's model reflects a large enterprise environment rather than a small-scale deployment. By the third year, the composite organisation had expanded its use of agents ninefold, allowing Veqtor8 to assess the cumulative effect of faster deployment, lower operating costs and reduced integration overheads.

For companies in heavily regulated industries, those factors can determine whether AI projects remain limited pilots or become production systems used more widely across the business. The report suggests the economics of data access and governance may be as important as model performance in determining how far agentic AI can be rolled out.

"Enterprises cannot afford to choose between speed, governance, and cost efficiency when scaling AI," said Alberto Pan, Chief Technology Officer, Denodo.

"By unifying access, semantics, and centralized policy enforcement across distributed data estates without requiring massive data relocation, the Denodo Platform provides the essential foundation for trusted, enterprise-grade AI agents," Pan said.