The perpetual route needed 12 to 18 percent compound growth for a decade to win

A ULA and a PULA grant the same deployment right. They differ on one thing: whether you keep the moment where the count freezes and the support base can be reset. This brief prices that difference, so you choose between two instruments rather than accept the one you are sold.

Prepared by Redress Compliance · August 15, 2026 · Oracle advisory. Based on 35 to 45 unlimited agreement decisions advised, 2024 to 2025.

Because the certification clause is an option, and options have a price whether or not anyone writes it down. When you sign a PULA you sell that option to Oracle, and the discount you receive is the premium. Value it the way you would value insurance: estimate the probability that your Oracle footprint is materially smaller in seven years, then multiply it by the support you would stop paying.

• Probability the estate shrinks, taken from the capacity plan and the application road map rather than from the sales narrative.
• Support you could shed, the portion of the annual stream attached to programs you would genuinely retire.
• Years of benefit, everything from the reset date to the end of your planning horizon.
• Compare against the perpetual premium over the ULA, expressed as a single number today.

There is also an asymmetry that drives behavior. Inside the term, deployment is free. The first unit above the certified number is full price, at whatever discount you can negotiate under pressure, with support attached forever. That is why a ULA rewards deploying hard before the freeze and rewards discipline afterwards. A perpetual agreement removes both behaviors, which is genuinely simpler and genuinely more expensive once growth stops.

Model both instruments over ten years, not three, because the comparison Oracle presents almost always stops at the point where the perpetual route still looks cheaper. Four inputs decide the answer.

• Deployment growth on the named programs, compound annual growth in processors or users, taken from funded capacity plans.
• The support base and its escalator, the starting stream and the annual uplift, typically 3 to 4 percent but often higher in drafts.
• The reset you would achieve, the share of support you could stop paying after a disciplined certification.
• The probability of a structural event, a divestiture, a migration off Oracle, or a large acquisition inside the horizon.

Stated as a rule: the perpetual route needs sustained compound growth of roughly 12 to 18 percent a year on the named programs across a full decade before it beats two disciplined ULA cycles. Very few enterprise Oracle estates grow that way for ten years. The ones that do are usually mid program on a database consolidation or carrying a funded acquisition pipeline with Oracle standardization written into it.

• Is the growth funded? Point at approved capital, named projects and dates. Aspiration is not growth.
• Is the growth on these programs? Database growth does not justify unlimited middleware, and the reverse is equally true.
• Could you certify if you had to? If your inventory cannot produce a defensible count today, the ULA exit you are relying on is theoretical.
• Is a divestiture plausible in seven years? Corporate development will not promise, but they will describe the shape of the portfolio.
• Would you buy the perpetual right at twice the price? If not, you are buying comfort rather than economics.

Five clear yes answers is the only pattern that justifies a perpetual agreement. Four yes answers and one doubt points to a ULA with a well drafted certification clause. Two or fewer points somewhere else entirely, usually to a straight purchase of what you actually run, which is the question the decision framework settles before instrument choice arises at all.

Negotiating a perpetual agreement as though it were a ULA is the most common drafting error we correct. On a ULA, the certification clause is the product, and four elements decide whether the exit you paid for actually works: the declaration window and when it starts, the measurement basis and whose method applies, whether authorized cloud deployment counts and on what averaging, and whether Oracle countersigns the certified quantities or merely receives them.

On a perpetual agreement there is no certification to protect, so the negotiable value moves to the support base and the assignment clause. Cap the annual uplift in the same signature that sets the base, and read the assignment language against your own corporate development pipeline. Both instruments leave the named program list, the entity and territory schedule, the virtualization counting rules, and Oracle's audit rights fully intact, which is why scope discipline matters equally on either route. The counting standard is set out in the certification pillar.

The common advice, from Oracle and from most resellers, is that a perpetual agreement is simply a better ULA: it removes certification risk and ends the counting forever, so any large and growing estate should prefer it. The reasoning fails at its first step, because it treats certification as a hazard rather than as an asset. Certification is the single moment in the entire Oracle relationship where the buyer, not the vendor, decides what the number is going to be. Selling that moment away for a discount on the entry fee is the most expensive trade in Oracle licensing, and it is the only one nobody puts a price on.

The framing works because the risk is real and immediate while the value is deferred and abstract. Certification genuinely is difficult: the count has to be defensible, virtualized estates are hard to measure, and a weak inventory turns the exit into an argument. Set against that, an instrument that deletes the whole exercise sounds like relief. But the difficulty is a project management problem with a known solution, whereas the surrendered option cannot be recovered at any price you would want to pay. Buyers who are told they are avoiding risk are in fact paying a premium to avoid a task.

What makes this asymmetric is the direction of travel. A ULA can become a PULA at any renewal, and Oracle will welcome that conversation. Across the perpetual agreements reviewed since 2020, no customer moved back to a timed agreement without paying for the privilege. When two options are close on the numbers, the reversible one is worth more than the model shows, because it preserves the ability to be wrong cheaply. That alone should settle most borderline cases.

Watch the timing of the proposal, too. The perpetual instrument tends to arrive at three predictable moments: after an audit finding, ahead of a known acquisition, and at a support renewal the customer has already called unaffordable. Each is a moment when the buyer feels exposed and the vendor is offering certainty. That is precisely when a ten year model is worth building rather than skipping, and why the instrument that wins on the entry fee usually loses across a decade. The instrument itself is described on the Oracle PULA page, the exit mechanics in the exit playbook, and the four axis comparison in the two models brief.

Watch the briefing · 4:30How to Negotiate an Oracle ULA: No Price List, Just Your Business CaseThere is no price list: the fee is a story built from your estate and your growth, so give conservative answers and keep the product list narrow.

Across roughly 35 to 45 Oracle unlimited agreement decisions advised on in 2024 and 2025, the proposal and the fit diverged sharply:

The patterns: comparisons truncated at the point the perpetual route still leads, growth taken from ambition rather than funded plans, and option value left out of the model entirely.

The buyer side move is to build the ten year model yourself before either instrument is priced. The wider library sits in the Oracle practice.

What actually separates a ULA from a PULA?

One clause. The two agreements are commercially identical until the last day of the ULA term, when certification freezes the count, ends the unlimited fee, and makes the support base reviewable. A perpetual agreement deletes all three of those moments, and every other difference between the instruments follows from that.

How much growth does the perpetual route need to win?

On the models we rebuild, roughly 12 to 18 percent compound annual growth on the named programs, sustained across a full decade, before it beats two disciplined ULA cycles. Very few enterprise Oracle estates grow that way for ten years, and the ones that do are usually mid program on a consolidation or carrying a funded acquisition pipeline.

What premium does a PULA carry over a ULA?

The entry fee is typically 1.4 to 2.2 times a comparable three year ULA. The premium itself is small in the context of the estate. What matters is that what you buy with it is permanent, and what you give up with it is the one day when the buyer sets the number.

Is the decision reversible?

In one direction only. A ULA can convert to a PULA at any renewal and Oracle will welcome the conversation. Moving from a perpetual agreement back to a timed one requires Oracle's agreement, and it prices that agreement. Where the choice is close, take the reversible option.

How should the comparison be modeled?

Over ten years, not three, because the comparison Oracle presents usually stops where the perpetual route still looks cheaper. Four inputs decide it: compound deployment growth on the named programs from funded capacity plans, the support base and its escalator, the reset you would realistically achieve at certification, and the probability of a structural event inside the horizon.

Which clauses carry the risk on each instrument?

On a ULA the certification clause is the product: the declaration window, the measurement basis, cloud counting, and whether Oracle countersigns. On a PULA the support base and the assignment clause carry the value, because there is no certification to protect and the annuity and the corporate perimeter are what remain negotiable.

When does Oracle propose the perpetual instrument?

At three predictable moments: after an audit finding, ahead of a known acquisition, and at a support renewal the customer has already flagged as unaffordable. In the decisions we advised on, the perpetual instrument was proposed first in about two thirds of cases and fitted the estate in fewer than one in three.

What is the single test that settles it?

Whether you would buy the perpetual right at twice the price. If the answer is no, you are buying comfort rather than economics. Five clear yes answers across funded growth, program alignment, certification readiness, divestiture likelihood and that price test is the only pattern that justifies a perpetual agreement.