Dropping Oracle support, the arithmetic of leaving and returning
Dropping Oracle support is easy. Coming back is priced so that it is not: three published clauses, matching service levels, repricing after license reduction, and reinstatement at 150 percent, decide the arithmetic, and the buyers who model all three before the letter consistently beat the ones who discover them after it.
Prepared by Redress Compliance · August 6, 2026 · Oracle advisory. Based on 25 to 35 support exit and reinstatement reviews worked 2024 to 2026.
Executive summary
Reinstatement is 150 percent, prorated. Returning to support costs 150 percent of the last annual fee you paid, prorated across the lapsed period, plus the fee for the new support period. The consequence is exact: a round trip hands back half of what you avoided.
Skip two years on a $2M line, $4M avoided, and the return fee is $6M, an incremental round trip cost of precisely 50 percent of the support you skipped.
Matching service levels binds the license set, not the CSI. A license set is all licenses of a program plus its options and management packs, or all licenses sharing source code, and support decisions apply to the whole set.
Across our reviews, roughly two thirds of estates had drawn the boundary around the CSI number, a billing construct, and had scoped the entire decision wrong before the modeling started.
Terminating support is reversible at a price; terminating licenses is not reversible at any price. The two letters look similar and are not: partial moves that terminated licenses to satisfy matching service levels destroyed perpetual positions that no reinstatement fee can resurrect.
The repricing clause completes the trap: the licenses you keep reprice at list less your smaller volume's discount, with a floor stopping the fee falling below what the survivors already carried.
The clock ignores who you pay meanwhile. Sustaining Support removes rights, not fees, a rights decision rather than a savings lever, and third party support does not pause the reinstatement meter, which runs from the day Oracle support ended regardless of who supports the estate in between.
The exit model has to price the return option honestly, because Oracle already has.
The three clauses, and what each one prices
| Clause | What it says | What it does to the model |
|---|---|---|
| Matching service levels | Support decisions apply to the whole license set: the program plus its options and packs, or programs sharing source code | Kills the cherry pick: the unit of decision is the set, and drawing it wrong voids the modeling |
| Repricing after reduction | Surviving licenses reprice at list less the smaller volume's discount, floored at their prior fee | Kills the partial saving: dropping half the estate rarely halves the bill |
| Reinstatement at 150 percent | Return costs 150 percent of the last fee, prorated across the lapse, plus the new period | Prices the round trip at half the avoided spend, and the clock never pauses |
The clauses are published, and they are a system. Each one exists to close the exit the previous one left open: matching levels blocks the selective drop, repricing blocks the partial one, and reinstatement taxes the temporary one.
Modeling any clause alone flatters the move; the honest model runs all three against the actual license sets.
The license set, where two thirds of models go wrong
The CSI is how Oracle bills; the license set is how Oracle decides, and they are different maps of the same estate.
A set spans every license of a program with its options and management packs, and reaches across programs sharing source code, so a drop scoped to one CSI can trip matching service levels obligations on licenses three invoices away.
Two thirds of the estates we reviewed had modeled at the CSI boundary, which meant their savings cases were arithmetic on the wrong universe.
The correction is documentary: the entitlement register mapped to license sets under the definitions, before any scenario is priced, the same consolidation the license position guide builds for every other Oracle decision.
The set map also reveals the genuine opportunities, sets that are cleanly separable, fully deployed on stable versions, with no reinstatement scenario worth pricing, which is where the real money is.
The third party support analysis
The exit economics end to end: the clause system, the license set mapping, the third party route with its legal position, and the reinstatement modeling for the estates that might return.
Get the white paper →The round trip math, worked
The worked pair from the engagement file: a $2M annual support line, dropped for two years, avoids $4M. Reinstatement charges 150 percent of the last annual fee prorated across the lapse, $6M on the return, plus the new period's support.
The incremental cost of having left and returned is exactly half of what was avoided, which converts the drop from a saving into a loan at punitive interest for any estate that might come back.
The strategic reading: the drop only pays where return is genuinely off the table, stable estates on frozen versions, platforms with a dated decommission plan, or moves to third party support intended as permanent.
The reinstatement clock running through the third party period is the clause that decides that last case: the route is a destination, not a parking space, and the provider landscape should be evaluated on that basis.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
The irreversible letter, licenses versus support
Terminating support suspends a relationship; terminating licenses destroys an asset.
The confusion arises inside matching service levels, where estates terminate licenses to shrink the set that must stay supported, and every such move in our reviews produced a repricing of the survivors plus the permanent loss of the terminated positions.
The perpetual license, once surrendered, is repurchased at then current list and discount or never, and no support decision is worth that trade unmodeled.
Sustaining Support completes the map of non exits: it removes rights, new versions, new patches, and certifications, while the fee continues, which makes it a rights decision for frozen estates rather than a savings lever for active ones.
The genuine levers sit a level up, in the support cost analysis and, for estates inside unlimited agreements, the PULA analysis where the support stream is the agreement's defining economics.
What we saw across exit reviews, 2024 to 2026
Across the 25 to 35 Oracle support exit and reinstatement reviews Fredrik Filipsson worked between 2024 and 2026, the cash saving was never the hard part; the boundaries were:
Estates that had drawn the license set at the CSI number and scoped the whole decision on a billing construct.
Partial moves terminating licenses to satisfy matching levels, each producing survivor repricing plus irreversible loss.
The successful exits shared a shape: cleanly separable license sets, genuinely frozen estates, return scenarios priced and consciously rejected, and the savings banked annually rather than projected optimistically.
The failed ones shared a letter, signed before the set map existed, whose consequences the three clauses then priced exactly as published.
Your first five moves
- Map the license sets before anything else: programs, options, packs, and shared source lineages, from the entitlement register, not the CSI list.
- Run all three clauses against every scenario: matching levels on the set, survivor repricing on the partials, reinstatement on the round trips.
- Price the return honestly, 150 percent prorated with the clock running through any third party period, and drop only where the answer is never.
- Never terminate licenses to shrink a support set without pricing the irreversibility; the letter that saves support can destroy the perpetual estate.
- Treat third party support as a destination, evaluated on the permanent case. The legal position, the provider field, and the Oracle practice carry the decision with you.
Frequently asked questions
What does it cost to reinstate Oracle support after dropping it?
One hundred fifty percent of the last annual support fee you paid, prorated across the entire lapsed period, plus the fee for the new support period. The arithmetic makes a round trip hand back exactly half of what was avoided: two skipped years on a $2M line avoid $4M and return for $6M.
Can we drop Oracle support on just some licenses?
Only at the license set boundary: matching service levels requires the whole set, the program plus its options and management packs, or programs sharing source code, to carry the same support level.
Drops scoped below the set trip the clause, and partial drops reprice the surviving licenses at the smaller volume's discount with a floor.
Does third party support pause the reinstatement penalty?
No. The reinstatement meter runs from the day Oracle support ended, regardless of who supports the estate in between, so a third party period counts fully toward the prorated 150 percent.
The route works as a permanent destination and fails as a parking space, and it should be evaluated on the permanent case.
Is Sustaining Support a way to cut Oracle support costs?
No, it removes rights rather than fees: no new versions, patches, or certifications, while the invoice continues. It is the correct posture for genuinely frozen estates and a rights decision everywhere, not a savings lever.
The savings levers are the set separations and negotiated positions a level above it.
What is the difference between terminating support and terminating licenses?
Reversibility.
Terminated support can be reinstated at the published price.
Terminated licenses are destroyed assets, repurchasable only at then current list and discount, and every partial move in our reviews that terminated licenses to shrink a support set paid twice, in survivor repricing and in irreversible loss.
When does dropping Oracle support actually pay?
Where the return scenario is genuinely off the table: cleanly separable license sets, estates frozen on stable versions, platforms with dated decommission plans, or permanent moves to third party support.
Priced against all three clauses at the license set boundary, those cases bank real savings; everything else is a loan at 50 percent.
