AI projects stalling because organisations aren't data-ready: Report
AI is outpacing data readiness as organisations struggle with harnessing usable information, according to Dominic Del Giudice at Iron Mountain.
AI projects are stalling because organisations aren't data-ready, according to Iron Mountain’s new report.
In the last 12 months, more than 86 percent of surveyed companies experienced delays or cancelations due to data concerns, with 38 percent of them significant delays or complete cancelation, according to the Information Supercycle: Aligning Data, Infrastructure and Capital for AI Scale report.
Iron Mountain managing director ANZ, Dominic Del Giudice, told CRN Australia that AI technology and use cases are not the problem.
“AI is becoming increasingly mature at a rapid pace, and it’s being adopted at enterprise scale. It's the underlying information and data that's the issue,” he said.
Del Giudice said organisations are also struggling to identify useable data for AI, which may be redundant, obsolete or trivial, or what the industry terms ‘ROT’, or dark data.
“Dark data is unknown and the value unclear. So it leaves a very small subset of data available to leverage AI and extract maximum value in terms of insights and specific business use cases,” he said.
The scale of work required to make information usable for AI is significant. Iron Mountain estimates 4 in 10 organisations spend at least 41 percent of their AI project timelines on manual data preparation, while only 39 percent fall into the report's 'digitised and structured' category, where dark data is centrally managed, indexed and readily usable for analytics and AI.
Fragmented and legacy information is another major barrier, with 56 percent of organisations citing information spread across disconnected systems or departmental silos, while 53 percent point to legacy formats or physical archives.
“It leaves a very small subset of data available to leverage AI and extract maximum value in terms of insights and specific business use cases,” Del Giudice said.
Those gaps become more consequential as businesses move beyond generative AI towards agentic systems capable of acting across workflows. According to Iron Mountain, 70 percent of organisations say they cannot effectively track the lifecycle (origin or journey) of the data used in their AI models.
“AI and AI use cases are only as good as the underlying information and data that sit underneath them,” he said.
The risk is that fragmented information can create security exposure, compliance and audit risk, higher discovery costs and loss of institutional knowledge.
“It's important that organisations understand what information they have, where it sits, and what the quality of that information is,” he said.
“There needs to be clear ownership and custodianship of that data,” he added.