There’s a question I’ve been asking in boardrooms a lot lately: If you knew your company was carrying a liability that was quietly inflating operating costs, slowing your most strategic initiative, and compounding interest with every quarter you waited — would that be an IT conversation, or a CFO conversation?
For most organizations right now, technical debt is exactly that liability. And yet, too many finance leaders are still treating it like a maintenance backlog buried somewhere on a ticket queue. That framing is costing them in delayed AI ROI, missed competitive windows, and real dollars.
Recent data tells the same story. MIT’s GenAI Divide: State of AI in Business 2025 report found that roughly 95% of generative AI pilots fail to deliver measurable financial impact, with only 5% translating into meaningful revenue gains. RGP’s AI Foundational Divide: From Ambition to Readiness survey of 200 U.S. CFOs, 86% said legacy systems and technical debt limit their AI readiness, even as most expect meaningful AI ROI within the next two years. Not talent gaps. Not budget. Not governance. The underlying infrastructure and data architecture is what’s blocking the investment from landing.
If you’re expecting AI to show up in your growth story, you can’t afford to treat technical debt as an IT-only conversation.
Why AI ROI Stalls Before It Starts
I spend a lot of time with enterprise technology and finance leaders. The pattern I see most often goes something like this: the organization approves an AI initiative, a team identifies compelling use cases, pilots launch, and early results look promising. Then things slow down. The value curve flattens. And by the time someone asks, “what’s the ROI on this AI investment?”, the honest answer is, “we can’t quite measure it yet.”
This is what we call the AI Value Paradox. More AI, without the right foundation beneath it, doesn’t produce more value. It produces more complexity. And when it comes to the measuring AI ROI, complexity is the enemy.
Two root causes drive this:
1. The Trust Ceiling
Without governance, security, and explainability embedded in the foundation, AI’s value hits a low threshold fast. Leadership won’t fund what they can’t trust. Employees won’t adopt what they don’t understand. Regulators won’t approve what they can’t audit.
The AI investment stalls because the infrastructure beneath it can’t be trusted at scale. The use cases were never the issue.
2. The AI Graveyard
Most organizations don’t fail at AI because they stop trying. They fail because they never stop starting. A new pilot here, a new tool there, three teams running three different LLMs with no shared governance, shadow AI spreading with no data controls. Every experiment looks reasonable in isolation, but none of them connect. Disconnected AI investments don’t compound, and the ROI never aggregates into true enterprise-wide impact.
Both of these root causes trace back to the same fragmented, aging technology foundation that wasn’t built to support the orchestration, data integrity, and security that enterprise AI requires.
Most organizations don’t fail at AI because they stop trying. They fail because they never stop starting.
Technical Debt: What the P&L Doesn’t Show
I want CFOs to understand that the cost of technical debt in an AI strategy isn’t maintenance spend alone. That’s the visible line item. The invisible cost is what you can’t do (and can’t measure) because the foundation isn’t there.
Consider what technical debt actually blocks:
- Data readiness: AI is only as good as the data it works with. Legacy systems fragment data across silos, make lineage impossible to trace, and create quality issues that produce hallucinations (due to inputs, not the model). When a CFO says, “my data isn’t in good enough shape,” that’s a revenue risk disclosure, not a technology one.
- System connectivity: Agentic AI has to reach the systems and data your business already runs on in a consistent way. When the environment is fragmented, every connection turns into a one‑off project, and the time and cost of getting from idea to production increases with each new use case.
- Governance at scale: As AI agents proliferate across departments, the questions get real, fast. Can your marketing team’s AI access finance data? Should customer service AI have read-write or read-only permissions on customer records? These are operational risks that compound with every new use case added to a legacy foundation.
At some point, the cost of not modernizing legacy technology shows up on the wrong side of the ledger. For most organizations, AI is accelerating that timeline.
Related Read: ENTERPRISE AI GOVERNANCE: HOW TO PLAY DEFENSE WHEN YOU CAN’T STOP EVERY YARD
The AI ROI Framework CFOs Are Missing: Clarify, Realize, Scale
One of the most common pitfalls in calculating enterprise AI ROI is that organizations try to measure value before they’ve established the baseline conditions for value creation. They’re measuring output before they’ve fixed the inputs.
The framework I recommend is built into Presidio’s AI Blueprint and organized around three stages — Clarify, Realize, and Scale.
Stage 1: Clarify — Build the AI ROI Framework Before You Build Anything Else
Clarify is the stage where you define the vision, prioritize use cases against actual business outcomes, assess your data and infrastructure readiness, and build the governance muscle to sustain what comes next. Skip this stage and you’re guessing at ROI. Do it right and you have a sequenced, deployment-ready plan with defined value metrics before a single dollar of AI infrastructure spend is committed.
As part of this stage, Clarify also includes an honest audit of your Foundation layer — the data architecture, network, security posture, and integration fabric that AI will depend on. If that foundation has structural debt, it shows up here and can be quantified in business terms. How many use cases are blocked? What’s the cost of manual workarounds? What’s the timeline impact on your highest-priority AI initiatives?
Questions and answers that belong in a CFO conversation, not on an IT ticket.
Stage 2: Realize — Turn One Department into Proof of ENTERPRISE-WIDE ROI
Realize is when the AI strategy stops being theoretical and starts running in production. During this stage, organizations should start small on purpose. Pick one department, prove the value there, and let that success set the pattern for the rest of the enterprise. The explicit goal of Realize is to produce a replicable model before committing to scale.
Presidio’ P.A.T.H. innovation lab was built for exactly this stage. It gives enterprise customers a hands-on environment that’s backed by NVIDIA, Cisco, and Vertiv infrastructure to test real workloads against enterprise-grade compute before committing production dollars. The AI ROI calculation gets grounded in actual performance data, not projections.
Step 3: Scale — Where AI ROI Compounds, If the Foundation Holds
Scale is the stage where AI shifts from a series of wins to part of how the company operates. You take what works in one part of the business and extend it across teams and use cases without losing control of risk, cost, or trust.
Organizations that rush to Scale before taking the steps to Clarify and Realize will hit the AI Value Paradox. They scale the complexity, not the value. But organizations that have done the foundational work first (i.e., cleaned the data, established governance, modernized the integration layer) find that Scale produces compounding returns. The same infrastructure that supports 10 use cases can support 100. The governance model that worked for one department can extend across the enterprise.
That’s where AI ROI strategies move from interesting to transformational.
Related Read: FROM AGENT SPRAWL TO COMPOUNDING VALUE: WHY AI ORCHESTRATION IS THE ONLY STRATEGY THAT SCALES
The Proof of AI ROI is in Production
As a CTO myself, I would never offer a framework to support AI ROI without data to back it up. Here’s what we’ve seen in practice.
A leading investment firm managing more than $20 billion in assets engaged Presidio for an 18-month, $6 million agentic AI transformation. We started with design-thinking workshops to identify 25-plus use cases. Then we built the governed data foundation and deployed agentic automation for back-office operations. The projected outcome is $10 billion in AUM growth enabled without adding headcount. That’s a measurable ROI calculation. And it was only possible because we did the foundational work first.
On the manufacturing side, the pattern holds. Agentic AI for invoice processing is now running three-times faster with cost reductions up to 70%. Predictive maintenance AI is catching failure events before they happen. These are production workflows with defined ROI metrics tied to operational cost reduction, not pilots waiting for a budget review.
Internally at Presidio, we’ve already enabled over 700 employees with agentic workflows, reduced AI session costs from $3.25 to $2.30, and processed over 15 billion tokens — all while iterating on what governance needs to look like at scale. We didn’t outsource this. We built it, measured it, and iterated on it. Because the only way to credibly help customers navigate this is to have navigated it ourselves.
What CFOs Should Be Asking Their Technology Teams
If you’re a CFO building an AI investment case for your enterprise, here are the questions I’d want on the table before the next
budget cycle:
- What percentage of our planned AI use cases are blocked by data quality or infrastructure gaps? If your CTO can’t answer this quantitatively, the Clarify work hasn’t been done.
- What is the total cost of our current technical debt? Not in maintenance spend, but in AI value we can’t capture? This is where finance and technology need to sit on the same side of the table. Your technology teams can surface the constraints and scenarios while finance can turn that into a shared view of risk, trade-offs, and where to prioritize AI investment next.
- Do we have a department-first proof of concept that connects AI investment to measurable business outcomes? If the answer is still, “we have several pilots running,” that’s not proof. That’s the AI graveyard in early formation.
- Is our AI governance model built into our foundation, or bolted on after the fact? Governance by design enables scale. Governance as an afterthought becomes a brake on every use case you add.
- Who is accountable for AI ROI across the organization? And do they have the span of control to actually move the levers? Fragmented accountability is one of the most common reasons AI initiatives stall between pilot and production.
Related Read: FROM BOLT-ON TO BUILT-IN: THE CASE FOR SECURE-BY-DESIGN PLATFORMS
Turn AI ROI Into a Shared CFO-CTO Mandate
Technical debt used to sit in the IT budget as a line item to manage down over time. In an AI-first enterprise, it shows up as stalled initiatives, unclear returns, and growth bets that never make it out of pilot. If AI is part of the story you tell your board and investors, then the state of your foundation is already a finance conversation, whether it’s framed that way or not.
The organizations that will see real AI ROI over the next few years will be the ones where CFOs and CTOs share a single view into where technical debt is blocking value, use that to Clarify where AI should be implemented first, Realize a working model in one part of the business, and Scale only what proves out in production. That’s where AI shifts from a set of experiments to a lever your enterprise can depend on.
If you’re trying to decide where to start, or how to course-correct a strategy that’s stalled, our AI Blueprint goes deeper into Presidio’s Clarify-Realize-Scale framework and the foundational work that supports it.
Download Presidio’s AI Blueprint eBook to use as a starting point for the next conversation between your finance and technology teams.
Rob Kim is Chief Technology Officer at Presidio, where he helps organizations modernize with purpose — turning AI, cloud, and digital technologies into real business outcomes. With over 20 years of experience in enterprise technology strategy, Rob serves as a technology orchestrator for clients navigating complex transformations with a strategy-first, value-led mindset. Connect with Rob on LinkedIn.
