Agile PLM on Third-Party Support: When Leaving Oracle Support Pays

Agile PLM 9.3.6 is the last release Oracle will ever ship, Premier Support ends December 31, 2027, and you are paying 22% plus compounding uplift for a product line that stopped receiving new functionality years ago. This page quantifies the third-party support case, prices the reinstatement penalty honestly, and shows how to structure a partial or full exit without triggering a repricing or an audit.

The Roadmap Cliff That Changes the Math

The decision starts with a fact Oracle does not advertise in renewal letters: Agile PLM 9.3.6, shipped January 2017, is the last release the product line will ever get. Oracle removed 9.3.7 from the roadmap in October 2023. Premier Support ends December 31, 2027, after which Agile drops to Sustaining Support, meaning no new patches, no security fixes, and no critical bug updates. Read that against what your 22% has actually purchased since 2023: Release Update Packs containing error corrections, security fixes, and targeted certifications. Nothing else. There is no functional roadmap to fund, no new modules arriving, no architectural investment being amortized. You are paying a percentage of net license fee for the right to open tickets and download patches on a codebase that stopped evolving nine years ago, and in 24 months even the patches stop.

This changes the math because the usual argument for staying on vendor support, that you are buying access to future IP, evaporates entirely. On a stable Agile estate with frozen configurations and no in-flight implementation, the only genuine deliverables left are break/fix response and certification against new OS, browser, and database versions. Those are exactly the deliverables third-party providers price at half the cost, and they are the deliverables Oracle itself will stop shipping at the end of 2027 regardless of what you pay.

Before you build a case, purge your evidence base. A widely circulated 2023 vendor post (updated November 2024 and still ranking) claims 9.3.7 would arrive in 2024 with Premier Support "guaranteed to mid-year 2029" and support extending "well into the 2030's." That is wrong and has been wrong since October 2023. If your internal business case, your systems integrator's deck, or your CFO's assumption of a 2029 runway traces back to that page, the entire timeline collapses by two years. Verify the current Lifetime Support Policy PDF and the ordering document terms yourself; our Agile PLM licensing guide sets out where those obligations actually live.

There is no future IP to buy, because Oracle stopped building it in October 2023.

What You Are Actually Paying: 22%, Uplift, and the Ten-Year Curve

Price the do-nothing option properly before you compare anything. Oracle's standard support fee is 22% of the net license fee, meaning the discounted figure on your original ordering document, not list price. Some legacy contracts predating 2010 carry lower ratios in the 18% to 21% band, so check the CSI-level detail rather than assuming 22% across the estate. On a heavily discounted Agile purchase, that base can look manageable. The base is not the problem. The escalator is.

Median actual year-over-year uplift runs 6.0%. On uncapped contracts, roughly 44% of estates, Oracle applies its 8.0% default, and even where a 4% cap exists in writing, 38% of renewals breach it. In our benchmarking of 60 to 80 renewals across 2024 and 2025, the letter arrived at the 8% default more than nine times in ten. The cap does not enforce itself; someone on your side has to catch the breach and dispute it before signature. Note also that you are bound to the support policy in force on your original ordering date, not the version attached to this year's renewal quote.

At the 8.0% default, the annual line nearly doubles inside nine years. Applied to a $1.0M Agile support spend, the 2027 Sustaining Support cliff arrives while your run rate is still climbing toward $1.2M, buying steadily less each year. That is the number to put in front of the CFO: not the 22%, but the compounding curve against a terminating product.

Two counter-levers exist before you consider exiting. A structured negotiation cycle typically removes 20% to 30% of the run rate without dropping a single license, and disputing an uplift breach resets the baseline for every subsequent year. Exhaust those first, because they cost nothing but time and they set a lower reference point for the third-party comparison described in our decision framework on when to leave and when to stay.

The 50% Savings Claim, Tested Against Your Stack

Rimini Street's published figure for Agile PLM is up to 50% off annual support fees and up to 90% off total cost of ownership, and its Agile page (updated November 18, 2025) positions that offer explicitly against Oracle's December 2027 end of Premier Support, noting that many customers intend to run Agile well past the cutoff while resisting the push to OCI. Read the qualifier carefully. The 50% is a fee reduction against your current Oracle support line; the 90% TCO number folds in avoided upgrades, avoided customization rework, and avoided migration labor, which are real but only if you were actually going to spend them. Build your business case on the 50% line item and treat the TCO figure as directional. The coverage scope is the part that genuinely earns attention: the offer covers all Oracle Agile products plus the underlying Oracle Database and Fusion Middleware/WebLogic, which is most of what a 9.3.6 stack actually runs on. Agile-specific delivery is not new either, dating to a September 21, 2015 announcement with Bloom Energy as the named early client, so you are not the pilot.

Take the scale and SLA claims into reference calls rather than into the board pack: roughly 2,100 active clients, over 1,000 Oracle licensees, sub-two-minute response on P1/P2, and engineers averaging 10-plus years. In our experience running these calls, the useful questions are how many Agile 9.3.6 references the provider can produce in your industry, who actually answered at 2am, and whether a regulatory validation event has been supported end to end. One coverage caveat matters more than any of the above: Agile PLM was never licensed with a generous database entitlement, and the restricted-use Oracle Database limits under Agile PLM govern what you may legally run before any support question arises. Confirm your restricted-use position in writing before you count database coverage as a free win, because a third party can support a database you are not entitled to deploy, and that gap surfaces in an audit, not a support ticket. For the cross-product view, our 2026 comparison of Rimini, Spinnaker, and Support Revolution sets out how the Agile-specific claims stack up.

Matching Service Levels: Why Partial Exits Blow Up

The single structural blocker to a partial saving is Matching Service Levels. If you remove one part of a license set from support, Oracle recalculates the support on everything remaining in that set as if the original discount had never existed. On a heavily discounted Agile estate, and most are discounted 60% or more against list, this does not just erase the intended saving. It raises your net spend while leaving you with fewer supported licenses. We have seen buyers drop 30% of an Agile module count expecting a 30% cut and receive a renewal letter that came in higher than the prior year. The policy is doing exactly what it was written to do: make the arithmetic of a partial exit unattractive enough that you either stay whole or leave whole.

Design around it rather than arguing with it at renewal. The practical fixes are contractual and structural, not conversational.

• Terminate whole order bundles or whole CSIs, never subsets within a license set, so there is no remaining set for Oracle to reprice.
• Time the cut to the expiration of a complete SKU or CSI, which removes the mid-term repricing trigger entirely.
• Negotiate an explicit written waiver of Matching Service Levels for the specific lines you intend to drop, and get it into the ordering document, not an email.
• On every future order, insist on single-product license sets; if Oracle refuses product-level splits, split the sets by geography or business unit instead, which preserves a partial-termination path.
• Map your CSIs against your Agile PLM module footprint before you model any saving, because sprawl across bundles is what makes clean termination impossible.

A partial exit that ignores Matching Service Levels does not halve your support bill, it raises it while leaving you with fewer licenses.

One contractual anchor is worth more than any negotiation script here: you are bound to the support policy in force on your original ordering date, not the version Oracle quotes in the renewal letter. Oracle has revised these policies repeatedly (the current Software Technical Support Policies are dated 10-July-2026, superseding 07-November-2025), and support renewal teams routinely cite the current text against contracts written years earlier. Pull the ordering documents, pull the policy version they incorporate by reference, and confront any repricing claim with the document that actually governs. If the older policy is more permissive on set composition or termination, that is a live entitlement and a negotiation lever. Our framework for deciding when to leave Oracle support sets out how to sequence that document review before you notify anyone of an intent to reduce.

Pricing the Reinstatement Penalty Honestly

Every CFO who hears "50% savings" asks the same question within thirty seconds: what if we have to go back? The answer is expensive and it is written down. Oracle's Software Technical Support Policies dated 10-July-2026 (prior version 07-November-2025) sets the reinstatement fee at 150% of the net technical support fee that would have been charged if support had been ordered originally, priced under the policies in effect at the moment of reinstatement, prorated from the reinstatement date back to the date support lapsed, and payable in addition to the fee for the go-forward support period. Read that sequence carefully, because three separate multipliers stack: back fees for the gap years, a 50% surcharge on those back fees, and the current-year renewal at today's price book rather than the one you left. Note also that pricing "in effect at reinstatement" means the uplift you avoided did not disappear, it accrued against you in the shadows.

The policy is designed to make termination a one-way door, and in most cases it works. But policy is not law, and 25 years of watching these files says the 150% number moves when Oracle wants something. The documented precedent is a Fortune 500 retailer that in 2023 negotiated the 150% penalty away entirely, worth over $1M, in exchange for a new Oracle Cloud Infrastructure commitment. If you have OCI consumption on the horizon anyway, that is the trade. Support Rewards credits, which reduce on-premise support spend for every dollar of OCI spend, are the related lever for a hybrid posture where Agile PLM moves to third-party support while other lines stay with Oracle. Do two things. First, get any penalty waiver or capped reinstatement right written into the contract, not into an email from a sales rep who will rotate out in eighteen months. Second, model the reinstatement invoice as a genuine liability line, not a footnote, and compare it against cumulative savings: at a 50% cut on a $500,000 line, you bank $250,000 a year and the three-year reinstatement number still exceeds eleven years of savings. That asymmetry is why the decision framework for leaving Oracle support should assume you are never coming back.

The Audit Response, and Why the Legal Ground Is Solid

Keep the legal question and the commercial question in separate columns. Legally, third-party support does not affect license compliance. You own a perpetual license, the right to use the software survives the support relationship, and nothing in the ordering document obliges you to buy maintenance from the licensor or from anyone. Oracle's objection is commercial, not legal, and its representatives will not say otherwise in writing. What Oracle does instead is behavioural: a support termination notice is one of the more reliable triggers for a license review, because an audit is the only mechanism left to recover revenue from an account that has stopped paying 22%. Treat that as a near-certainty in your savings model, not a tail risk. Budget the audit-readiness work and subtract it from year one benefit.

For Agile PLM specifically, three exposures do the damage. First, user metric reconciliation: named user versus concurrent counts drift over a decade of hiring, contractors, and inactive accounts left enabled, and the metric comparison for Agile PLM users tells you which side of the line your entitlement actually sits on. Second, module deployment mapping: Agile's add-ons (PQM, PCM, PPM, Engineering Collaboration, and the integration packs) enable easily and appear in audit scripts whether or not anyone budgeted for them. Third, the restricted-use database license underneath Agile, which permits the database only for the licensed application and breaks the moment someone points a reporting tool, a data warehouse feed, or a second schema at it. That last item is the most common seven-figure finding in this product family.

The sequencing rule is absolute: close the gaps before the termination notice goes out, never after. Once you have given notice, every remediation purchase happens at Oracle's price with Oracle holding the clock, and your negotiating position is a company that just cancelled its support contract. Run the reconciliation, true up quietly at renewal pricing while you are still a paying support customer, document the entitlement position with your own evidence, then serve notice. Nine times out of ten, an estate that walks in with a clean, self-produced position and a defensible module map ends the review with no finding, and the 50% saving lands intact.

Third-Party Support Versus Forced Fusion Migration

Stop comparing support lines. The real question is whether you want Oracle's calendar or your own to decide when Agile PLM gets replaced. Premier Support ends December 31, 2027, and 9.3.6 is terminal, so a do-nothing estate arrives at that date with no runway and a Fusion Cloud PLM proposal on the table. Third-party support at roughly half the Oracle run rate (Rimini Street publishes up to 50% on annual Agile PLM support fees) buys a defined three to five year extension past the cutoff. In my experience that runway is worth more than the cash saving, because a PLM replacement selected under deadline pressure almost always closes on the incumbent's terms, with the subscription term, user metric, and uplift cap all set by the party holding the clock. Use the Oracle third party support decision framework to score your estate before you price anything.

The runway matters more than the discount, because a PLM replacement chosen under Oracle's deadline closes on Oracle's terms.

What To Do First: A 90-Day Sequence

Run this in order. Steps taken out of sequence are how buyers discover Matching Service Levels exposure after they have already signaled an exit, which destroys the leverage they were trying to build.

• Days 1 to 20, pull the paper. Every ordering document, every CSI, every amendment. You are establishing three facts: the actual support ratio (22% is standard, but pre-2010 contracts often sit at 18 to 21%), whether an uplift cap exists, and exactly where the license set boundaries fall. Matching Service Levels risk is determined in those documents and nowhere else. Confirm which support policy version binds you, because you are held to the policy in force on your original ordering date, not the version quoted in the renewal letter. Check the restricted-use Oracle Database underneath Agile PLM here as well, since its CSI placement often decides whether a clean partial exit exists at all.
• Days 15 to 40, model the curve. Build the ten-year do-nothing baseline at both 6.0% (the median observed uplift) and 8.0% (Oracle's default on uncapped contracts). On a $1.0M line that is $13.2M versus $14.5M over the decade. Set the third-party quote against it, then add two line items buyers routinely omit: audit-readiness work, because Oracle frequently responds to a support exit by testing compliance, and reinstatement optionality, priced at 150% of the fees that would have accrued plus the current period fee.
• Days 20 to 70, negotiate Oracle in parallel. Benchmarked 2024 to 2025 renewals arrived at the 8% default more than nine times out of ten, and a structured cycle typically removes 20 to 30% of the run rate without dropping a license. Run that cycle with the third-party quote already signed off internally so it is a priced alternative, not a bluff Oracle can call.
• Days 70 to 90, set the date and write it down. Time termination to whole-CSI or whole-SKU expiry, never a subset, or Oracle recalculates remaining support as if no discount ever existed. Issue notice in writing, and document your reinstatement position in the same correspondence.

Two closing points. Third-party support does not create a license compliance problem: you own the perpetual license and may choose any provider or none, so Oracle's objection is commercial rather than legal. And the 150% reinstatement fee, while policy, has been negotiated away before, typically as part of a broader trade. If you have any OCI spend or intend to, put that on the table before you sign the exit, not after.

Related advisory: the full topic guide.

Frequently asked questions

When does Oracle support for Agile PLM actually end?

Premier Support for Agile PLM 9.3.6 ends December 31, 2027, after which the product moves to Sustaining Support. Sustaining Support keeps your system running and gives you access to existing patches, but delivers no new security fixes, no critical bug updates, and no new certifications. Because Oracle removed 9.3.7 from the roadmap in October 2023, 9.3.6 is the terminal release, so there is no later version to upgrade into for a support extension.

How much does third-party support for Agile PLM save?

Providers publish savings of up to 50% of the annual Oracle support fee, and up to 90% of total cost of ownership when avoided upgrade and migration work is included. Treat the 50% as the ceiling on the support line, not a guaranteed number, and net it against audit-readiness spend and the cost of preserving a reinstatement option. Against a $1M Oracle line escalating at 8%, a flat third-party fee compounds into a materially larger gap by year five than the headline percentage suggests.

What is the Oracle reinstatement fee if we go back?

Oracle's Software Technical Support Policies set reinstatement at 150% of the net technical support fee that would have been charged had support been ordered originally, priced at reinstatement rates and prorated from the lapse date, plus the fee for the new support period. A $500,000 annual line reinstated after three years therefore prices at about $2.75M. The fee is policy, not statute, and buyers have negotiated it away in exchange for other commitments, but you should assume the full number in your base case.

Does leaving Oracle support break our Agile PLM licenses?

No. Perpetual licenses survive the termination of a support contract, and nothing in the license grant requires you to buy support from Oracle or from anyone. The exposure is commercial and behavioural, not legal: you lose access to new patches and to My Oracle Support downloads, and Oracle may open an audit to test your compliance position after you give notice.

Can we drop support on part of the Agile PLM estate only?

Only if the license sets and CSIs are structured to allow it. Oracle's Matching Service Levels policy lets it reprice the remaining support as if the original discount never existed when you remove part of a license set, which frequently erases the intended saving. Terminate whole order bundles or whole CSIs, time the cut to a complete expiry date, and get any waiver in writing before notice goes out.

Should we just migrate to Fusion Cloud PLM instead?

Migration may be the right end state, but the 2027 date should not set the timetable. Third-party support typically buys three to five years of runway at roughly half the Oracle run rate, which lets you scope, tender, and negotiate a Fusion Cloud PLM subscription without deadline pressure. Estates with heavy active development or imminent regulatory recertification are the exception and may justify accelerating.