ERP Management GroupERPMG.AI
Menu
Field Notes

On premises, cloud or hybrid: what changes for a complete ERP

By John Rae · Released November 4, 2026 · Posted November 7, 2026

People often treat the deployment question as if it settles the ERP question. It does not. Whether your ERP runs in the vendor's cloud, your own data center or some mix, my advice is the same: one complete ERP, with every function on the vendor's own modules. What the model changes is who controls the upgrades, how the cost behaves, how much control you keep, and how long the road to one platform is.

I run an Oracle Fusion program at a regional water utility, and I spent the 27 years before that on other ERP systems, starting with Infor/Lawson implementations in the late 1990s. So I have lived on both sides of this line. The facts below come from what I could open, and I flag where the source is thin.

Upgrades

This is where the models differ most, and the vendors do not all work alike.

Oracle's Fusion development team says it plainly on its Fusion Insider blog: quarterly updates are mandatory and cannot be skipped. Test environments are updated on the first Friday of the update month and production two weeks later, on the third Friday, and that two weeks is the whole testing window. New features that change what users see arrive disabled, but some switch on by themselves at a published opt-in expiration date. The quarterly shape matches the cycle I run.

SAP S/4HANA Cloud public edition releases every February and August, per SAP's training, with customers and partners still running regression testing. Private edition keeps the on-premise cycle: Feature Pack Stacks every six months, a requirement to install at least one upgrade every seven years to stay within mainstream maintenance, and SAP executing the technical upgrade. NetSuite rolls its upgrades out in waves. A partner's schedule shows the current one running from August through October 2026 and the next expected in spring, and Oracle's documentation gives you a Release Preview account to test your own workflows before your production account moves. For Infor CloudSuite, a third-party directory says feature releases come every six months with monthly fixes. I did not open an Infor page saying so, so that is to confirm.

On premises, the calendar is yours. So is the work. For a complete ERP, that distinction matters more than it first appears. In the cloud, each module on the suite arrives on one vendor calendar. A side system adds a calendar of its own, whichever way the ERP is hosted.

Support dates: less urgent than you may have heard, depending on the vendor

Oracle's March 2026 announcement extends E-Business Suite 12.2 Premier Support through at least 2037, along with Hyperion, JD Edwards EnterpriseOne, PeopleSoft and Siebel CRM. Oracle's Lifetime Support policy describes that date as discretionary, extendable in one-year increments, and says Oracle currently has no plans to end Premier Support on the EBS Continuous Innovation releases. An Oracle shop on premises has time.

SAP's clock is firmer. An SAP spokesperson told The Register that the dates remain the end of 2027 for mainstream maintenance on SAP Business Suite 7 and the end of 2030 for extended maintenance, with a transition option from 2031 to the end of 2033 tied to the RISE path. The Register, citing Gartner, reported that only 37% of ECC customers had licensed S/4HANA.

Infor's date is the one closest to my own history. A 2023 post on the LawsonGuru user forum quotes Infor's support site: mainstream maintenance for all Lawson version 10 products runs until December 31, 2030, extended maintenance follows from January 1, 2031 for on-premises deployments only, and single-tenant cloud customers are told to move to CloudSuite before that date. ERP Research reports the same date. I have not seen it on an Infor page myself, so confirm it with Infor. The same guide warns that moving from on-premises Lawson to its CloudSuite successor is a re-implementation. That is the thing to plan around, not the date alone.

Cost

I wish this section were stronger. The only total-cost study I opened is Nucleus Research from 2016, which put on-premise ERP at 2.1 times the cost of Oracle ERP Cloud over three years, with personnel as the main difference. It is ten years old, covers one vendor and uses a short window.

Forrester's study for SAP and AWS, paid for by those vendors and modeled on a composite, found the largest savings came from eliminating data centers, with $6.1 million in reduced costs from phasing them out, against $11.5 million in total costs, for a 45% ROI. Its Infor study, commissioned by Infor, describes IT teams spending significant time maintaining heavily customized on-premises systems. A Gartner survey of 383 finance leaders, as reported by ChannelPost, found on-premises ERP the most discontinued technology in that survey.

Read together, these say that a cloud move shifts money out of data centers and staff time and into a subscription and a migration. They do not prove it is cheaper for you. The honest test is your own hardware, staff and upgrade project costs against a real quote.

Control

On premises keeps real control: when to upgrade, how far to customize, where the data sits. The cloud gives some of that up, and critics say so plainly. Third Stage Consulting, which sells ERP advice, argues that "clean core became popular because vendors needed it" and that private cloud can give managed infrastructure without surrendering flexibility the way pure SaaS often requires. I do not dismiss that. I would only add that customization is often how a shop ends up unable to upgrade at all.

Hybrid

Hybrid usually means a cloud ERP with some systems still on premises. A Gartner research director wrote in 2015 that hybrid would be the norm within five years, under the "postmodern ERP" idea of breaking suites into loosely coupled applications. TechTarget's 2022 discussion of composable ERP also describes "a hybrid of on-premises and cloud applications." I treat hybrid as a fair place to be during a move and a poor place to stay. Mixed estates are the thing a complete ERP exists to remove, and a hybrid held as a permanent design keeps every interface alive.

Counterpoints

Some reasons to stay put are real. Cloudian, which sells on-premises storage, reports from 212 IT decision-makers that 45% have seen new cross-border data restrictions in two years and 75% moved at least some workloads back from public cloud in 24 months. A Citrix-commissioned poll, reported by IT Brew, put the figure at 94% of 350 leaders. Those are vendor-paid and about infrastructure rather than ERP SaaS, and "at least some workloads" is a low bar. Sovereignty is the better argument. Forrester has warned of sovereignty constraints in SAP's cloud push, although The Register's report on Fusion Applications in Oracle's EU Sovereign Cloud shows vendors answering inside SaaS. I found nothing I could open on FedRAMP or US-specific regulated workloads, so that is to confirm. If a regulation or contract forces you on premises, stay there and consolidate there. And cloud does not end testing: a vendor survey reported by IT Brief found 92% saying cloud adoption increased theirs.

What I would do

  1. Check your support date first, from the vendor's own page. That tells you how much time you have.
  2. Decide the destination before the hosting: one complete ERP, and which family.
  3. If you are on premises with time left, begin by retiring systems beside the ERP that the on-premises suite already covers. That shortens the later move.
  4. If you are going to the cloud, plan it as a re-implementation, and move module by module instead of ripping and replacing.
  5. If you are hybrid, write down the date each on-premises piece goes away.

Wherever it runs, the recommendation holds: all modules under one ERP umbrella.

Sources

Opened:

Not opened, to confirm before publishing:

← All Field Notes and Decoded · Four ERPs, one update calendar