AI will fail without unified information estate

Organizations widely understand the challenge of data debt, yet progress remains blocked, most often by fragmented authority, ownership, and budget gaps.

"We still retain a significant portion of our historical data in physical “boxed” form. Additionally, digital data sits across multiple legacy claims and policy systems obtained through years of corporate acquisitions. A primary roadblock to implementing AI for us is the cost and effort required to extract, translate, and pull this fragmented data forward into modern platforms."

— Finance leader, Insurance1

The cost of dark data.

Fragmentation often delays progress, compounding the already significant data debt inside an organization. They’re collecting and storing, often unnecessarily, unanalyzed records that are redundant, obsolete, and trivial (ROT).

What’s needed is a mindset shift to understand that tackling ROT can actually free up the valuable budget needed to invest in AI innovation, while also improving the ability to deliver effective, responsible AI.

As much as 1/3 of enterprise data can be considered ROT (Redundant, Obsolete, or Trivial).

Another 52% is dark data with unknown value, leaving only 15% to feed AI systems.2

The cost of ROT for enterprises is substantial, and likely growing at a rate equal to the CAGR of enterprise data itself, estimated at +23% annually.

This puts average costs of ROT for companies of varied size and industries in a range as follows, which left unchecked will continue to grow annually.3

Cost of ROT by company size

digital governance ownership is highly diluted across the C-suite

Closing the governance gap:
A roadmap for AI maturity

Tackling ROT and growing data debt requires companies to build a modern information architecture that shifts the consolidation of how information is governed inside the enterprise. In most enterprises, disparate frameworks and disconnected teams have existed across IT, data, and legal teams.

Divided ownership and split budgets erode confidence in retention decisions, fueling a culture of data hoarding. Instead of being remediated or activated, data is often stored, where it remains an untapped liability rather than a strategic asset.

Records management is reborn as an engine of enterprise AI

Enterprises are waking up to their governance silos, as seen in the reclassification of the historical “records function” in terms of strategic importance. Accompanying the shift in perception, Records Management (RM) has undergone a shift in reporting structure and is often positioned under the purview of IT, Legal, or the CDO (Chief Data Officer).

This shift is just the beginning; changing reporting lines for RM is insufficient without creating a shared, active information governance discipline across IT, Data, and Legal teams.

53%

of organizations see records management professionals as “indispensable” or “important” to the task of managing the retention of critical business information.4

The AI Governance Maturity Journey:

37%

of companies are at the Foundational phase (early adopters), establishing comprehensive data and governance frameworks;

45%

are at the Optimized phase (mature adopters), automating governance and compliance and using AI to govern AI.6

Fragmented authority blocks your AI roadmap and governance automation.

Governance in the AI era must be a continuous operational progression—one where information is constantly processed, automated, and activated for intelligence. 

Mature enterprises are establishing AI Governance Councils that unite IT, data, and legal functions. This cross-disciplinary ownership ensures that: Legal/Compliance creates and maintains rules against evolving regulations; IT provides the scalable infrastructure for digitization; data ensure data is “AI-ready”, free of ROT and bias.

Only 1 in 5 (21%) companies report having a mature model for governance of autonomous agents.5

Liquidating ROT to fuel your AI future

By purging ROT and implementing a tiered, agile information architecture (across physical, tape, and hybrid, multi-cloud instances), enterprises can create a lean, high-integrity data environment needed for AI insights.

These lean, data driven organizations are accelerating their AI journey and capturing new competitive advantages from their data.

Data-driven organizations are

23X

more likely to acquire customers and 19X more likely to be profitable.7

Actions you can take today:

If you’re ready to go from ROT to ROI, we’ll take you there.

Iron Mountain will help you aggressively purge data ROT to minimize dark data costs and ensure AI models are grounded in accurate, high-value context. By activating your historical information, we’ll help you transform static records and archives into intelligence that powers better decisions, stronger customer experiences, and AI innovation.

When you unlock the power of your information with Iron Mountain.

With our digital capabilities and RIM services, we help you unify authority across physical records, digital, and governance lifecycles, creating a secure, cost-efficient approach to information management and AI activation. Learn more about our Records and Information Management services.

Learn more

Get all the insights, all in one place.

Download the report >>

If The Information Supercycle is speaking to you, let's talk.

Connect with our experts to explore what the Information Supercycle could mean for your organization, data strategy and AI priorities.

See how Iron Mountain’s offerings can help you
go from AI pilot to enterprise scale.

Sources

1 Iron Mountain analysis, “The information supercycle: Aligning data, infrastructure, and capital for AI scale,” October 2026. Cost of ROT estimates are Iron Mountain analysis and assume:

  • Average combined cloud and paper archive spend by company size and industry reflects industry benchmarks.
  • Storage accounts for 30% of that spend.
  • 30% of stored data is ROT.
  • ROT costs grow 23% a year, in line with data growth.

2 Simon Jelley, “How to keep your data from ROT-ing in the cloud,” Forbes Business Council, Forbes, January 12, 2023.

3 IDC Global DataSphere forecast, as cited in Ocient, “The world is moving beyond big data, according to Ocient survey of 500 data and technology leaders,” Business Wire, August 9, 2022. Forecast of 23% compound annual growth in global data creation and replication, 2020 to 2025.

4AIIM, “Balancing records management and information governance: Automation is the key ingredient,” AIIM Industry Watch, 2020.

5Deloitte AI Institute, “The state of AI in the enterprise,” 2026 edition, January 2026. Survey of 3,235 business and IT leaders across 24 countries.

6Iron Mountain, “Fuel innovation with AI and unstructured data,” May 2025. Survey of 1,400 IT and data decision-makers at organizations with 250 or more employees in the US, UK, France, Germany, India and Australia, conducted by Vanson Bourne.

7McKinsey & Company, “Five facts: How customer analytics boosts corporate performance,” 2014, based on the McKinsey DataMatics 2013 survey. As cited by QuantumBlack, AI by McKinsey, “Unlocking the value of clean, trusted data,” Medium, November 2025.

Sources reviewed October 2026.