IBM: Data Silos Remain Major Hurdle for Enterprise AI Adoption

According to IBM's research, the main obstacle to enterprise AI adoption isn't the underlying technology, but the persistent challenge of fractured data ecosystems.

Ed Lovely, VP and Chief Data Officer at IBM, identifies data silos as the critical vulnerability in modern data strategy. His remarks follow a new IBM Institute for Business Value study indicating that while AI is primed for scaling, enterprise data readiness lags behind.

The report, surveying 1,700 senior data executives, reveals that departmental data—from finance and HR to marketing and supply chain—remains locked in separate domains without unified standards or a common language.

This fragmentation directly undermines AI initiatives. "When data is trapped in isolated silos, every AI project devolves into a lengthy, six-to-twelve-month data cleanup effort," Lovely explained. "Teams exhaust more time finding and reconciling data than on deriving actionable insights."

This poses a direct risk to competitiveness. For CIOs and CDOs, the mandate has evolved from merely collecting and securing data to effectively deploying it as fuel for advanced AI systems.

From Data Custodian to Business Catalyst

The study's consensus is clear: data leaders must be uncompromisingly focused on business impact, with 92% of CDOs linking their success to this outcome-driven approach.

This highlights a core dilemma: while 92% prioritize business value, only 29% are confident they possess "clear metrics to assess the business value of data-driven results."

This ambition-reality gap is where autonomous AI agents, capable of learning and acting to achieve goals, are poised to assist. Confidence in these tools is growing, with 83% of CDOs in IBM's study believing the potential benefits of deploying AI agents outweigh the associated risks.

At global medical technology firm Medtronic, teams were mired in manually matching invoices, purchase orders, and delivery confirmations. Implementing an AI solution automated this workflow, slashing document processing time from 20 minutes per invoice to just eight seconds with over 99% accuracy. This freed staff from low-value data entry for higher-impact work.

Similarly, renewable energy company Matrix Renewables deployed a centralized data platform to monitor its assets, achieving a 75% reduction in reporting time and a 10% cut in expensive downtime.

IBM Identifies Key AI Hurdles: Architecture, Governance, and Talent

Replicating such successes demands a new architectural approach that avoids silos. The outdated model of expensively and slowly moving data into a central lake is fading. IBM's research finds 81% of CDOs now favor bringing AI to the data, rather than relocating data for AI.

This strategy depends on modern patterns like data mesh and data fabric, which create a virtualized access layer over distributed data. It also promotes "data products"—packaged, reusable data assets built for specific business needs, like a "customer 360" view or a financial forecasting dataset.

However, improving data access intensifies governance demands. A strong CDO-CISO partnership is now vital to balance agility with security. Data sovereignty is a key concern, with 82% of CDOs treating it as a cornerstone of their risk strategy.

The most significant barrier may be human capital. The report highlights a widening talent gap that risks derailing progress. In 2025, 77% of CDOs report struggles in attracting or retaining top data talent, a sharp rise from 62% in 2024.

This scarcity is compounded by rapidly evolving skill requirements. IBM found that 82% of CDOs are "hiring for data roles that didn't exist last year, related to generative AI." Overcoming this cultural and skills challenge is often the most difficult aspect.

Hiroshi Okuyama, Chief Digital Officer at Yanmar Holdings, noted: "Cultural change is difficult, but there's growing awareness that decisions must be grounded in data and facts, requiring evidence collection during the decision-making process."

Unlocking Data Silos to Deploy Enterprise AI

On the technical side, enterprise leaders must champion the shift from siloed data environments. This involves investing in modern, federated data architectures and encouraging teams to build and consume secure, reusable "data products" across the organization.

Secondly, culturally, data literacy must become a company-wide imperative, not solely an IT issue. The 80% of CDOs who affirm that data democratization accelerates their organization are correct. This requires fostering a data-driven culture and investing in intuitive tools that empower non-technical staff to work with data.

The ultimate goal is to advance from running isolated AI pilots to scaling intelligent automation across fundamental business processes. The successful organizations will be those that treat their data not as a byproduct of applications, but as their most strategic asset.

Ed Lovely concluded: "Enterprise AI at scale is achievable, but success hinges on powering it with the right data foundation. For CDOs, this means building a seamlessly integrated enterprise data architecture that drives innovation and unlocks business value.

"Organizations that master this won't just enhance their AI capabilities; they will transform their operations, accelerate decision-making, adapt more swiftly to change, and secure a sustainable competitive advantage."

See also: New data centre projects mark Anthropic’s biggest US expansion yet

Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expo happening in Amsterdam, California, and London. This comprehensive event is part of TechEx and co-located with other leading events including the Cyber Security Expo. Find more information here.

AI News is powered by TechForge Media. Discover other upcoming enterprise technology events and webinars here.