Introduction to Enterprise AI
According to IBM, the primary barrier holding back enterprise AI isn’t the technology itself but the persistent issue of data silos. Ed Lovely, VP and Chief Data Officer at IBM, describes data silos as the “Achilles’ heel” of modern data strategy. Lovely made the comments following the release of a new study from the IBM Institute for Business Value that found AI is ready to scale, but enterprise data is not.
The Problem of Data Silos
The report, which surveyed 1,700 senior data leaders, found that functional data remains stubbornly isolated. Finance, HR, marketing, and supply chain data all operate in isolation, with no common taxonomy or shared standards. This fragmentation is having a direct, negative impact on AI projects. “When data lives in disconnected silos, every AI initiative becomes a drawn-out, six-to-twelve-month data cleansing project,” said Ed Lovely, VP and Chief Data Officer at IBM. “Teams spend more time hunting for and aligning data than generating meaningful insights”. This is a direct threat to competitive advantage. For CIOs and CDOs, the mission is no longer just to collect and protect data, but to deploy it effectively to power these new AI systems.
From Data Janitor to Value Driver
The consensus from the study is that data leaders must be relentlessly focused on business outcomes, with 92 percent of CDOs agreeing their success depends on this focus. Herein lies the central tension: while 92 percent are aiming for business value, only 29 percent are confident they have “clear measures to determine the business value of data-driven outcomes.” This gap between ambition and reality is where AI agents that can learn and act autonomously to achieve goals are expected to help. Leaders are showing a growing confidence in these tools, with 83 percent of CDOs in IBM’s research stating the potential benefits of deploying AI agents outweigh the risks.
Real-World Examples
At global medical technology company Medtronic, teams were bogged down matching invoices, purchase orders, and proofs of delivery. By deploying an AI solution, the company automated this workflow. The result was a drop in document matching time from 20 minutes per invoice to just eight seconds, with an accuracy rate exceeding 99 percent. This allowed staff to be redeployed from low-value data entry to higher-value work. Similarly, renewable energy company Matrix Renewables implemented a centralised data platform to monitor its assets. This led to a 75 percent reduction in reporting time and a 10 percent reduction in costly downtime.
Overcoming the Hurdles
Achieving these results requires a new approach to data architecture while avoiding silos. The old model of costly, slow data relocation into a central lake is being replaced. IBM’s study finds 81 percent of CDOs now practice bringing AI to the data, rather than moving data to AI. This approach relies on modern architectural patterns like data mesh and data fabric, which provide a virtualised layer to access data where it lives. It also champions the concept of “data products” (packaged, reusable data assets designed for a specific business purpose, such as a “customer 360” view or a financial forecast dataset.)
Architecture, Governance, and Talent Gap
However, making data more accessible introduces governance challenges. The CDO-CISO alliance is now essential to balance speed with security. Data sovereignty is a particular concern, with 82 percent of CDOs viewing it as a core part of their risk management strategy. The biggest hurdle, however, may be people. The report reveals a widening talent gap that threatens to stall progress. In 2025, 77 percent of CDOs report difficulty attracting or retaining top data talent, a sharp increase from 62 percent in 2024. This scarcity is exacerbated by the fact that the required skills are a moving target. IBM found that 82 percent of CDOs are “hiring for data roles that didn’t exist last year related to generative AI”. This cultural and skills challenge is often the hardest part.
Opening the Data Silos
On the technical front, enterprise leaders must champion the move away from siloed data estates. This means investing in modern, federated data architectures and pushing teams to develop and use “data products” that can be securely shared and reused across the organisation. Second, on the cultural front, data literacy must become a business-wide priority, not just an IT concern. The 80 percent of CDOs who say data democratisation helps their organisation move faster are correct. This means fostering a data-driven culture and investing in intuitive tools that make it simpler for non-technical employees to interact with data.
Conclusion
The goal is to elevate the organisation from running isolated AI experiments to scaling intelligent automation across core business processes. The companies that succeed will be those that treat their data not as an application byproduct, but as their most valuable asset. Ed Lovely, VP and Chief Data Officer at IBM, said: “Enterprise AI at scale is within reach, but success depends on organisations powering it with the right data. For CDOs, this means establishing a seamlessly integrated enterprise data architecture that fuels innovation and unlocks business value. Organisations that get this right won’t just improve their AI, they’ll transform how they operate, make faster decisions, adapt to change more quickly, and gain a competitive edge.”
FAQs
Q: What is the primary barrier holding back enterprise AI?
A: The primary barrier holding back enterprise AI is the persistent issue of data silos.
Q: What is the impact of data silos on AI projects?
A: Data silos have a direct, negative impact on AI projects, causing teams to spend more time hunting for and aligning data than generating meaningful insights.
Q: What is the solution to overcoming data silos?
A: The solution is to invest in modern, federated data architectures and push teams to develop and use “data products” that can be securely shared and reused across the organisation.
Q: What is the biggest hurdle in implementing AI solutions?
A: The biggest hurdle is the widening talent gap that threatens to stall progress, with 77 percent of CDOs reporting difficulty attracting or retaining top data talent.
Q: What is the key to successful AI implementation?
A: The key is to establish a seamlessly integrated enterprise data architecture that fuels innovation and unlocks business value, and to foster a data-driven culture across the organisation.
