Most business cases for replacing a legacy data warehouse start in the wrong place: the licence renewal. It is the number on the invoice, so it becomes the number in the spreadsheet. But the licence line is the tip of the iceberg. The real legacy data warehouse cost sits below the waterline - in escalating support, audit exposure, maintenance drag, security debt, scarce skills, and the opportunities you simply cannot pursue while your data stays where it is.

Three pressures are bringing those hidden costs to the surface at the same time. Support fees are climbing on autopilot. The AI conversation is exposing what legacy data cannot do. And statutory reporting cycles are getting tighter and more frequent. Put together, they reframe a comfortable assumption: doing nothing is not free. It is an active, compounding spend decision - one that gets more expensive every year you defer it.

Here is what an honest total cost of ownership actually includes.

The cost you see versus the cost you carry

The visible cost - annual licensing and support - is genuinely large, but it is the part finance already tracks. The dangerous costs are the ones that never appear as a single line: the audit settlement, the half a team kept busy "keeping the lights on", the cyber premium on an ageing estate, the analyst you cannot hire, the AI use case you cannot deliver. A credible modernisation case prices all of them, not just the renewal.

1. Support and licensing that only ever goes up

Start with the line you do see, because even that is usually underestimated. For Oracle estates, annual support typically runs at around 22% of net licence fees and rarely comes down, even as your actual usage shrinks. On top of that, support uplifts have been running at roughly 8% a year - which compounds to about a 50% increase within five years of signing. That is not a one-off; it is a baked-in trajectory that your five-year model has to reflect.

UK higher education has just lived through a sharp version of this. After Oracle moved Java to a per-employee subscription model - which analysts at Gartner assessed as several times more expensive than the older approach - sector body Jisc negotiated a national framework aiming to save UK institutions up to £45 million against standard list pricing. One sector insider, speaking to The Register, described the exposure institutions faced beforehand in stark, near-existential terms. If your renewal trajectory is not modelled with annual uplifts built in, it is already wrong.

2. The audit exposure you are not pricing in

There is a second, less predictable cost attached to legacy estates: the licence audit. Audits are not random, and they are not neutral events. Licensing specialists note that scrutiny tends to intensify around the very moment an organisation begins discussing a move of its workloads to the cloud - precisely when you are most exposed and least prepared.

The findings concentrate in predictable places: unused database options and management packs that were switched on and forgotten. For a business case, that means an audit contingency is not pessimism - it is prudence. Treat it as a real line item with a real number.

3. "Keeping the lights on" is eating your budget

Then there is the maintenance drag - the cost of simply keeping an ageing platform running. The public-sector picture is sobering: analysis cited across UK government commentary suggests roughly half of IT spend goes on maintaining legacy systems, and around a third of central government systems are now classed as legacy.

Reframe that as opportunity cost. Every pound spent on life-support is a pound not spent on new capability. The platform is not just expensive to own; it is expensive because of what its upkeep crowds out.

4. Security and compliance debt

Ageing systems are also a growing attack surface, and that carries both a risk and a hard cost. Reporting on UK public-sector security has highlighted that a substantial share of cyber incidents - in the region of 40% - exploit vulnerabilities in ageing technology. Unsupported or near-unsupported components attract higher insurance premiums, more patching effort, and greater UK GDPR exposure if sensitive data sits on a platform that can no longer be properly secured. Security debt does not show up on the licence line, but it shows up.

5. The skills you cannot hire - and the people you cannot lose

Legacy platforms run on legacy skills. Deep ODI and PL/SQL expertise is increasingly scarce and ageing, and much of the operational knowledge often lives in one or two people's heads. That key-person dependency is a genuine balance-sheet risk: lose the wrong person and the platform becomes a black box.

The regulatory clock is now forcing the issue. The shift to in-year statutory reporting under HESA Data Futures pushed institutions to break free of manual, key-person-dependent processes that simply could not sustain a faster, more frequent reporting cadence. When the reporting rhythm changes, fragile, undocumented processes stop being an inconvenience and start being a compliance risk.

6. The biggest cost of all: the things you cannot do

The largest cost rarely appears in any model because it is an absence - the capability you cannot deliver while your data stays on legacy infrastructure. This is the conversation dominating 2026. More than 28,000 organisations are now running Microsoft Fabric in production, as fragmented legacy stacks become difficult to justify in an AI-first world.

The public-sector verdict is blunt: out-of-date technology and poor data quality and sharing are actively putting AI adoption at risk. You cannot bolt natural-language querying, real-time operational dashboards, or reliable forecasting onto an estate that refreshes overnight and cannot conform its data. The opportunity cost of a legacy warehouse is not theoretical - it is every AI and analytics initiative that stalls before it starts.

7. The carbon line nobody costs yet

Finally, a cost that is small today but rising and increasingly board-visible: the energy and carbon footprint of an on-premises server estate. As organisations commit to Net Zero targets, the emissions associated with running and cooling physical infrastructure sit awkwardly against those commitments. It belongs in a modern total-cost conversation, even if it is not yet the deciding factor.

How to actually quantify the cost of staying

You do not need a forensic model to make the case - you need an honest one. A simple "total cost of staying" worksheet captures the costs above on five lines:

• Support trajectory - current support, escalated at a realistic annual uplift over five years.

• Audit contingency - a provision for likely findings, not zero.

• Maintenance effort - the FTE cost of keeping the platform running.

• Security and compliance - patching, insurance and remediation on an ageing estate.

• Opportunity cost - the value of the analytics and AI initiatives currently blocked.

Add them up and compare against a modern platform's run cost. The point is rarely that cloud is automatically cheaper on day one - it is that the legacy number is escalating and capability-blocking, while the modern number turns spend into governed, AI-ready capability.

From cost centre to capability

A legacy data warehouse looks like a fixed, manageable cost. Examined properly, it is a rising one - and the steepest part of the curve is the capability it denies you. Modernising is not about chasing a cheaper invoice; it is about converting money you are already spending on life-support into a platform that can actually do what the organisation now needs.

The first step is simply to price the truth. Once "doing nothing" carries an honest number, the business case tends to make itself.

Ready to price your own cost of staying? Book a short, no-obligation cost-of-staying review.