Migrating From EBS Supply Chain to Fusion SCM Cloud: What the Licensing Shift Costs

Moving supply chain off E-Business Suite is not a swap of one bill for another, it is the permanent surrender of a perpetual asset in exchange for a repricing subscription, and Oracle's support policies are engineered so that switching off EBS SCM support saves you far less than your business case assumes. This page quantifies both sides of the ledger, exposes the Matching Service Levels and repricing clauses that eat the savings, and gives you the CSI restructuring and contract moves to make before you sign the Fusion order.

What Your EBS SCM Entitlements Are Actually Worth Today

Before you model a single Fusion SCM subscription line, fix the baseline, because most migration business cases start from the wrong number. EBS Supply Chain is not one product. It is a set of separately licensed modules (Inventory Management, Order Management, Manufacturing, Procurement) sold perpetually on Application User or Custom Application Suite metrics. Inventory Management lists at $4,595 per Application User, the same list as Financials, and Manufacturing modules typically start at $5,000 to $7,000 per Application User. Nobody paid list. Real-world discounted net on SCM Application Users lands around $2,500 to $3,500 per user with first-year support included. That distinction is the whole argument: annual support is 22% of net license fees, not list, so an enterprise that accumulated $10M in net EBS license fees over fifteen years pays $2.2M per year, indefinitely, and that $2.2M is the number a Fusion SCM subscription has to beat. Large EBS estates in the $3M to $15M net range sit at $660K to $3.3M annually. If your business case compares Fusion list pricing against an EBS list base, you have manufactured savings that do not exist.

The second baseline problem is documentary. Most large EBS deployments were assembled over a decade or more through multiple purchase orders, acquisitions, and expansions, which produces a partially documented license position: nobody can cleanly reconcile entitled Application Users to deployed ones. Oracle License Management Services treats exactly that condition as a revenue opportunity, and the moment you announce a Fusion SCM migration you hand Oracle both the trigger and the timeline. Reconstruct your entitlement, order by order and CSI by CSI, before any Fusion conversation begins, and cross-check how your current user population would map under Fusion SCM's named user, employee, and transaction volume metrics, because a compliance gap discovered mid-negotiation costs you the discount you were about to win.

The 2037 Deadline That Does Not Exist

Every Fusion SCM proposal we see arrives with an urgency narrative attached, and it is almost always false. Oracle has extended Premier Support for EBS 12.2 year after year: from 2030 to 2033, then 2034, 2035, 2036, and in March 2026 to at least 2037. Oracle said so on its own EBS Technology Blog, citing Continuous Innovation. The mechanism is a rolling commitment, documented in Oracle's own 12.2.15 announcement, to give at least ten years' notice before ending Premier Support, adding a year to the horizon as each year passes. Meanwhile at least one 2026 vendor guide still prints "Premier Support through December 2031," and Oracle sales reps happily repeat stale dates when it suits the quarter. Treat any December 2031 claim as a negotiating tactic. Demand the Oracle-published date in writing, from Oracle, before you accept a single time-pressured Fusion SCM quote.

Any 2031 end-of-support claim is a sales tactic, not a fact: Oracle published 2037 on its own blog in March 2026.

Context matters here too. EBS 12.2 sits inside Applications Unlimited alongside JD Edwards EnterpriseOne, PeopleSoft, Siebel CRM, and Hyperion, all carrying Continuous Innovation and long-dated Premier Support. EBS SCM is not an orphaned line being quietly starved. That is real leverage: you can credibly stay put for another decade while you evaluate scope, metrics, and price, and the same dynamic shapes negotiations across the family, as we cover in the Siebel to Fusion CX licensing analysis. Be honest about the downside, though. Falling out of Premier is not a comfortable place to land. Market Driven Support runs roughly $25K to $225K per year and still leaves coverage gaps, so it is a contingency, not a strategy. The point is not that you should never move. The point is that you should move on your timetable, at your price, with a supported EBS estate as your walk-away position rather than a cliff edge Oracle invented for you.

Fusion SCM Subscription Economics Against a Net Support Base

Oracle sales will frame the business case against the cost of *repurchasing* EBS SCM at list, which is a fiction, because nobody in your position is buying EBS again. The only comparison that matters is the annual Fusion SCM subscription against the incremental annual EBS support you would genuinely stop paying, and that second number is far smaller than the first because your EBS support base is historically discounted. Oracle charges 22% of net license fees, so the enterprise that accumulated $10M in net EBS license fees over fifteen years pays roughly $2.2M per year (Vendor Benchmark, April 2026). If SCM modules represent, say, 40% of that net base, the theoretical stop-paying number is around $880K, not the $4,595-per-user list arithmetic Oracle's slides will show. Then add the asymmetry at the end of the term: EBS support buys maintenance on an asset you own, while the Fusion subscription buys nothing residual and must absorb renewal uplifts indefinitely. In our negotiation experience, uncapped Fusion SCM renewals land in the 4% to 8% band per cycle, which compounds a $1.2M subscription past $1.75M by year ten. Cap it in the original order, not at renewal, and read how to right-size modules and cap Fusion SCM uplift before you sign.

The second under-modeled cost is metric translation. Your EBS entitlement is counted in per-Application-User licenses for Inventory, Manufacturing, and Order Management. Fusion counts differently: employee-based metrics for some modules, named user for others, and pure transaction volume for Order Management order lines. A 900-user EBS SCM footprint can translate into a 6,000-employee Fusion metric, or into an order-line tier that Oracle sizes off your peak year rather than your average. Model both the user-based and volume-based options with your own transaction extracts before Oracle sizes them for you, using the Fusion SCM metric comparison and the order-line volume pricing mechanics as the reference points. Where the counts diverge by more than 20%, that gap is your negotiation target, not an accounting rounding error.

The Repricing Clause That Eats Your Savings

Here is the governing language, verbatim from Oracle's Software Technical Support Policies dated 07-Aug-2026: "In the event that a subset of licenses on a single order is terminated or if the level of support is reduced, support for the remaining licenses on that license order will be priced at Oracle's list price for support in effect at the time of termination or reduction minus the applicable standard discount." Read that twice. Terminating your SCM lines does not simply subtract them from the bill. It resets the price of everything left on that order to current list, less whatever Oracle deems "standard discount" today, which will be a fraction of the discount you fought for a decade ago. Work it: a customer with $10M net license across one CSI pays $2.2M in support. SCM is $3.75M of that net, 37.5% of the estate. Drop it, and Oracle reprices the surviving $6.25M of Financials and Procurement at today's list minus a standard discount. The survivors' support can rise from $1.375M to roughly $2.2M, and the net annual saving is zero. The "we will just stop paying SCM support" line in your business case evaporates.

Terminating 37.5% of your estate can cut your support bill by exactly nothing, which is why the "we'll just stop paying SCM support" line item is fiction.

Two limits sit inside that clause, and buyers almost never enforce them. First, Oracle's repricing right is confined to licenses on a single order, normally identified by CSI number. Terminations on one order cannot legally reprice licenses sitting on a separate order (House of Brick, February 2026). If your SCM modules were bought on their own PO with their own CSI, Oracle has no repricing hook at all, and your job in the renewal call is to make the account manager show you the order document that supposedly links them. Second, there is a hard cap most customers never invoke: the new support price "will not exceed the previous support fees paid for both the remaining licenses and the licenses being terminated or unsupported." That means the worst case is you keep paying what you already paid, minus nothing, and Oracle cannot use repricing to charge you *more* than the pre-termination total. Also note Oracle never defines "standard discount," which in practice should sit at least 10% to 25% below list. Demand the arithmetic in writing before you authorize any partial termination, and if the CSIs are commingled, restructure them before the Fusion order is signed rather than after.

Matching Service Levels and Why Cherry Picking Is Blocked

The clause that kills most EBS SCM migration business cases is not a price, it is a rule about sets. Oracle's Software Technical Support Policies require that all qualifying licenses in a support set sit at the same service level, Premier, Extended, or Sustaining. You cannot leave EBS Financials on Premier Support while quietly dropping support on Inventory, Order Management, and Manufacturing because those SCM users have moved to Fusion. Oracle's own policy language and the repricing provision dated 07-Aug-2026 make the alternative explicit: to stop paying on the SCM lines you must terminate them, not merely unsupport them, and once terminated the entitlement is gone even though your ordering document calls it perpetual. Buyers routinely discover this three weeks before renewal, at which point the only options are pay for everything or surrender licenses permanently.

There is a second detail that decides whether you are arguing from a strong or weak position: which version of the support policy binds your order. The policy in force at the original ordering date generally governs, unless later paper (a new order, an amendment, a cloud agreement with an incorporation clause) pulled in a newer version. For SCM orders signed a decade ago, that distinction is worth real money, because older policy documents were less aggressively worded on repricing and set composition. Pull the actual ordering documents and the referenced policy version before you accept Oracle's characterization of what you can and cannot drop. In our experience, the version Oracle's renewal desk cites is the current one, not the one your paper incorporates.

The structural countermeasure is CSI architecture, and it is only available before you sign the Fusion order. Oracle's repricing right is limited to licenses within a single order, usually identified by CSI number, so terminations on one CSI cannot legally reprice a separate CSI. That makes the pre-migration task obvious: identify every SCM line you intend to drop, and get them grouped into a distinct CSI while Oracle still wants your Fusion signature. Once the Fusion SCM order is executed, you have no commercial lever left to demand a restructure and Oracle has no reason to grant one. Treat CSI separation as a contractual deliverable of the Fusion deal, negotiated alongside the uplift cap and module right-sizing terms, not as an administrative request you file afterward.

Run-Off Costs During a Multi-Year Migration

Every EBS SCM business case we review understates the overlap. A phased SCM rollout, warehouse by warehouse or plant by plant, runs 18 to 36 months in practice, and for most of that window you are paying EBS support and Fusion subscription simultaneously. The EBS side does not shrink proportionally as sites cut over, because Matching Service Levels and the repricing clause mean partial reductions rarely produce partial savings. Meanwhile Fusion is billed on committed subscription from the contract start date, not from go-live. On a $2.2M annual EBS support base (the Vendor Benchmark example of $10M accumulated net license fees at 22%), a 30-month overlap carries roughly $5.5M of run-off support that no one modeled, before uplift. Most contracts we see carry 4% to 8% annual support uplift caps; uncapped estates saw 7% to 12% increases in 2026, which on a long tail compounds into real money.

Third-party support at 50% to 70% below Oracle list is the number that changes the negotiation. It is not only a cheaper run-off floor, it is the credible alternative that disciplines Oracle's Fusion SCM pricing. Say it out loud in the room: the tail estate can move to third party, which means Oracle's revenue from the EBS side is at risk regardless of whether the Fusion deal closes. Understand the asymmetry, though. If you lapse support and later need to restore it, Oracle charges a 150% reinstatement fee on the back-support amount. If you terminate the licenses, reinstatement is not available at any price. A failed or delayed Fusion SCM phase, and manufacturing and order management phases fail more often than financials do, means repurchasing EBS capability at current list. Sequence terminations behind verified production stability at each site, and validate your Fusion scope against which SCM modules you actually need to license before you cut anything off.

What to Do First: The Pre-Signature Sequence

Do these in order, and do not let Oracle sequence them for you. First, reconstruct your own EBS SCM license position and CSI map before License Management Services offers to do it for you. Most large EBS estates were assembled across a decade of separate POs, so the license position is partially documented, and that ambiguity is treated as revenue opportunity at renewal. Second, kill the urgency: get Oracle's March 2026 confirmation that Premier Support for EBS 12.2 runs through at least 2037 stated in writing by your account team, because stale 2031 dates are still being quoted. Third, make CSI separation of the SCM lines you intend to drop a condition precedent to signing the Fusion order, not a follow-up request. Oracle's repricing right is limited to licenses on a single order, usually identified by CSI, so separation is the only structural defense against Matching Service Levels forcing you to keep paying for everything. Fourth, model the Fusion subscription against support you will actually avoid after repricing, not against your gross EBS support line. In our experience most first-draft business cases overstate avoided support by a wide margin because they assume clean cancellation.

• Put the annual uplift cap in the Fusion SCM order at 4% or lower, ideally CPI-linked with a hard ceiling, and cover renewal terms as well as the initial term. See negotiating a Fusion SCM renewal and capping uplift.
• Get written acknowledgment of the policy cap that repriced support "will not exceed the previous support fees paid" for both remaining and terminated licenses.
• Fix module names, metric definitions, and included-versus-separate functionality in the order itself, using the SCM metric definitions as the drafting baseline.
• Strike any language tying the cloud discount to audit findings, license reconciliation, or a compliance settlement.
• Secure fixed pricing for known ramp volumes across the full term, not a first-year rate.

Post-signature, the risk moves to renewal repricing and to any residual on-premise audit exposure, both of which need their own governance track.

Related advisory: the full topic guide.

Frequently asked questions

Do I lose my perpetual EBS SCM licenses when I move to Fusion SCM Cloud?

Only if you terminate them. Perpetual licenses survive as long as you keep paying support or accept dropping to Sustaining, but Oracle's Matching Service Levels policy can force termination of licenses the ordering document calls perpetual if you try to reduce support on a subset within the same order. Once terminated there is no reinstatement path, you would repurchase at current list. Decide deliberately, and never let a Fusion order document terminate EBS entitlements as a side effect.

When does Oracle actually stop supporting EBS 12.2?

As of March 2026, Oracle publicly committed to Premier Support for EBS 12.2 through at least 2037, extended annually under a rolling ten-year-notice policy. Several vendor guides and Oracle sales materials still cite December 2031, which is stale. Ask for the current published date in writing before accepting any deadline-driven Fusion SCM proposal.

Why does dropping EBS SCM support not reduce my bill proportionally?

Oracle's support policy reprices the remaining licenses on the same order at list minus the applicable standard discount at the time of termination. If your historical discount was deep, the survivors reprice upward and can offset the entire saving. The limits are that repricing applies only within a single order or CSI, and the total cannot exceed what you previously paid for remaining plus terminated licenses combined.

Should I restructure CSIs before or after signing the Fusion SCM order?

Before, always. Separating drop-candidate SCM lines into their own CSI is the only reliable way to create a partial termination path, and Oracle has little reason to agree once your Fusion signature is in hand. Make CSI separation a condition precedent in the negotiation, not a follow-up request.

Is third-party support a real option during an EBS to Fusion SCM migration?

Yes, and it serves two purposes. Providers typically price at 50% to 70% below Oracle list, which cuts your run-off cost during an 18 to 36 month overlap, and the credible option itself is leverage on the Fusion subscription price. The tradeoffs are limited Premier feature scope and no new regulatory patches, so scope it against the modules you are actually still running.

How should I compare Fusion SCM subscription cost to what I pay for EBS today?

Compare the annual Fusion subscription against the EBS support you would genuinely avoid, net of any repricing on surviving licenses, not against EBS list prices or the theoretical cost of rebuying. Then extend the model over ten years with 4% to 8% subscription uplifts and no residual asset at the end. Most business cases look favorable only because they compare Fusion pricing to an inflated EBS baseline.