Oracle Infrastructure Modernisation Report 2026: The Bill Is the Architecture Decision

More organisations are hurt by Oracle licensing than by Oracle technology. The 2026 figures make that unusually easy to demonstrate.

Oracle support runs at 22% of net licence fees per year and rises 8% annually by default, so an obligation signed three years ago now costs about 26% more than the number that was approved. Nobody re-approved it. It compounded.

A note on that figure, because it is widely misquoted, including by sources we used here. The increase after three years is commonly given as 24%, which is 8% added three times. The clause compounds, so the arithmetic is 1.08 cubed, or 25.97%. It is a small difference in year three and it stops being small: after five years simple addition says 40% and compounding says 47%.

Meanwhile the Java SE Universal Subscription prices per employee, not per user, across seven published bands from $5.25 to $15.00 per employee per month. Organisations that previously licensed a handful of processors report effective cost increases of 300% to 500%.

And roughly one in five Java users is estimated to face an Oracle audit within three years.

What the numbers look like#

The per-employee metric is the structural change#

The 2023 shift to per-employee Java pricing is the single most consequential licensing change in this estate, and it is still catching organisations out three years later.

Under the old metrics, your Java cost tracked your Java usage. Under the Universal Subscription it tracks your headcount. A company of 6,000 people running Java on four servers pays for 6,000 employees. Growth in an unrelated part of the business raises a technology bill.

Two consequences follow, and both are commercial rather than technical:

1. Your Java cost is now coupled to hiring. Any headcount forecast is now also a licensing forecast, and almost no organisation models it that way.
2. Partial migration saves nothing. Removing Oracle Java from 80% of your estate does not cut the bill by 80%, or at all. Under a per-employee metric it is all or nothing, which changes the business case for a migration from incremental to binary.

That second point is the one that most often turns a "gradual modernisation" plan into a stalled one: two years of genuine engineering effort with no line-item saving to show for it.

Audit is a licensing event, not a security event#

An audit rate near one in five within three years makes this a planning assumption rather than a tail risk. The reported triggers are worth knowing because none of them requires you to have done anything wrong:

• Download telemetry. Oracle can see downloads through its own Java Management Service.
• Renewal gaps. Contract renewal is when discrepancies surface.
• Third-party tooling. License monitoring reports can prompt a review.

The defensible position is not "we believe we are compliant". It is an evidenced deployment inventory you could hand over tomorrow: what is installed, on which hosts, which distribution, which version, and who installed it. Organisations that can produce that settle audits quickly. Organisations that cannot spend the audit period building it under time pressure and with an adversary setting the deadline.

Migration off is a rewrite in disguise more often than it looks#

79% have migrated, are migrating, or are planning to migrate to open-source Java. Completed large-enterprise OpenJDK migrations run 9 to 14 months.

Note what that duration implies. Swapping a JDK is a week of work. Nine to fourteen months is what happens when you discover the rest of it: applications certified against a specific vendor JDK, build pipelines pinned to a vendor distribution, monitoring agents with version dependencies, third-party software whose support contract names Oracle Java specifically, and the internal tooling nobody documented.

The same shape applies to the database side. The data moves; the stored procedures, scheduler dependencies, reporting built on vendor-specific behaviour and undocumented integrations do not. A programme scoped as a data copy and delivered as an application rewrite is the standard way these overrun.

What we would do#

1. Establish the licence position before the architecture. Which agreements, which metrics, which uplift clauses, which renewal dates. This determines which options were ever real, and it is routinely done last.
2. Build the deployment inventory now, while nobody is asking for it. It is the audit defence and it is also the migration scope. One artifact, two jobs.
3. Price the exit at the point of entry. The moment to know what leaving costs is while you still have negotiating leverage, not three years later under audit.
4. Model the per-employee metric against your headcount plan. If you are hiring, your Java cost is rising whether or not your Java usage is.
5. Decide all-or-nothing deliberately. Under a per-employee metric, a partial migration is a cost with no saving. Either commit to full removal with a date, or stay and negotiate properly. The middle option is the expensive one.

Our read#

Oracle modernisation is almost never blocked on whether the technology can be replaced. It is blocked on a contract position nobody has fully mapped, an inventory nobody has built, and a support line that has been compounding quietly since it was signed.

Do the commercial work first. It is unglamorous, it takes weeks rather than months, and it is the only part that changes which options are available to you.

Method and limitations#

This report synthesises published 2026 licensing analyses. It is not primary research: we did not run a survey, and we say so because a report that overstates its method is not worth citing.

The limitation here is specific and important. Oracle does not publish audit rates, migration durations or realised cost increases, so those figures come from licensing advisory firms and specialist consultancies. Those firms make their living advising on Oracle cost reduction, which gives them both genuine visibility into real contracts and a commercial interest in the numbers sounding alarming. We have used their figures because no better source exists, and we are telling you that rather than presenting them as neutral.

List prices for the Java SE Universal Subscription are published by Oracle and are reliable. The 300% to 500% increase, the one-in-five audit rate and the 9 to 14 month migration duration are advisory estimates and should be treated as indicative of direction and order of magnitude, not as measured population statistics.

Nothing here is legal or contractual advice. Your agreement governs, and Oracle agreements vary more than any other enterprise vendor's.

Published 2026-08-10. Figures current to that date.

Sources#

• Oracle Java Licensing Cost 2026, Redress Compliance
• Oracle Price List 2026: Key Changes and Analysis, Redress Compliance
• Licensing Oracle Java: The Complete 2026 Guide, LicenseQ
• Oracle Database and Java Licensing Cost 2026, Software Pricing Guide
• Migrate from Oracle Java to OpenJDK, LegacyLeap

Download the data (CSV) Every figure charted above, machine-readable, with its source on each row.

The findings, in one place

Quote these directly. They are the sentences we stand behind, which is not always true of a sentence assembled out of a paragraph.

• Oracle support runs at 22% of net licence fees and rises 8% annually by default, so a three-year-old obligation now costs about 26% more than the figure that was approved. The 24% widely quoted is 8% added three times; the clause compounds.
• Java SE is priced per EMPLOYEE, not per user, at $5.25 to $15.00 per employee per month. Reported effective increases against processor licensing run 300% to 500%.
• Under a per-employee metric a partial migration saves nothing, which turns modernisation from an incremental programme into a binary one.
• About one in five Java users is estimated to face an audit within three years, and 79% have migrated, are migrating or plan to migrate to open-source Java.

How to cite this

Every report here may be quoted, charted and reproduced, including commercially, with attribution to BvLogic and a link to the report page.

Oracle Infrastructure Modernisation Report, 2026 edition. BvLogic Research, 2026-08-10. https://bvlogic.com/research/oracle-modernization/

Every report here may be quoted, charted and reproduced, including commercially, with attribution to BvLogic and a link to the report page. No permission needed and no form to fill in. We would rather be cited widely than control the copy, and a citation policy that requires an email is a citation policy designed to fail.

Use this

Everything below is generated from this report, so no figure appears in them that is not published above. Free, no form, no attribution required beyond a link.

Press and analyst enquiries

Figures, the underlying data, or a named comment for a piece you are writing: hello@bvlogic.com. We answer these properly and we will tell you when a number is weaker than it looks.

What else is coming for Oracle Modernisation