MAS 9.1 Rolling Upgrades, AppPoints Math, and the Real Cost of Standing Still
There is a date that should be in every Maximo leader's calendar: September 30, 2026. That is the end of Extended Support for Maximo 7.6.1.x. After that, only Sustained Support (no new security patches, no new fixes) remains through 2030. If you are still running classic Maximo on WebSphere today, you are running on borrowed time — and borrowed budget.
Meanwhile, IBM's MAS 9.1 release train keeps moving. The May 27, 2026 monthly drop (Manage 9.1.18, MAS Core 9.1.18, Optimizer 9.1.12, Predict 9.1.8, Collaborate 9.1.11, IoT 9.1.11) is the latest in a cadence that now ships monthly. The 9.0 line is also still receiving updates (Manage 9.0.26) for customers who have not yet cut over to 9.1.
This article does the boring but necessary work: turns the May 2026 release data into an upgrade plan you can actually defend in front of IT finance, infrastructure, and the business.
What shipped in May 2026 — and why it matters
The full MAS 9.1 release table tells a story. Every application now has a security patch, a known-issue fix, or a feature-channel drop on roughly a monthly cycle:
| Application | May 27, 2026 release | Notable fix / capability |
|---|---|---|
| MAS Core | 9.1.18 | Security updates, suite-wide stability |
| Manage | 9.1.18 | APAR rollups across WO, Inventory, MIF |
| Optimizer | 9.1.12 | Builds on April FC: what-if RBF algorithm, MO on SNO support |
| Predict | 9.1.8 | Model accuracy and runtime improvements |
| Collaborate | 9.1.11 | Continuing the Assist → Collaborate rename cycle |
| IoT | 9.1.11 | Connector maturity |
| AI Service | 9.1.15 | Powers Maximo Assistant + Reliability Strategies AI |
| MVI | 9.1.17 | Visual Inspection model packaging |
A few things to notice:
- 9.1 is the strategic line. New capability (Condition Insight, Maximo AI Service, agentic workflows) lands on 9.1 first. Staying on 9.0 means you can patch, but you are not getting the AI roadmap.
- Feature Channel is alive. Optimizer's April 2026 Feature Channel delivered the upgraded what-if analysis using Radial Basis Functions (RBF) — a real algorithm upgrade, not a cosmetic change.
- 9.0 is in maintenance. It is safe, but it is also the line where new AppPoint add-ons may not be eligible.
The AppPoints model, demystified
AppPoints is the single biggest "this is a different platform" moment for anyone coming from Maximo 7.6. The 7.6 model was named users and modules. MAS replaced that with a shared consumption pool that flexes by user tier and access pattern.
The published entitlement values (per user, per access model) look like this:
| User tier | Concurrent AppPoints | Authorized AppPoints |
|---|---|---|
| Self-Service | 0 | 0 |
| Limited (≤ 3 modules) | 5 | 2 |
| Base (standard access) | 10 | 5 |
| Premium (full access + advanced) | 15 | 10 |
| Administrative (system config) | — | 3 |
| Administrative (with industry solution) | — | 5 |
The Authorized model bills the user continuously, whether they are logged in or not. It is the right choice for a planner or reliability engineer who lives in the system. The Concurrent model only bills while a session is active — better for occasional users, contractors, or role-based app access that spikes weekly.
What is included in the base pool (no extra AppPoints): Maximo Manage, Maximo Mobile, Maximo Scheduler (no longer separately licensed), the Reliability Strategies Library, and the role-based applications.
What costs extra AppPoints: Health, Monitor, Predict, Visual Inspection, Maximo AI Service, and Maximo Optimizer. In 9.1, Maximo AI Service carries a limited-use watsonx.ai license but is an additional 10 AppPoints per use.
A worked AppPoints example
Take a 500-user utility: 30 power users (reliability engineers + supervisors), 200 standard technicians on Manage + Mobile, 200 occasional users on a self-service role, plus 20 admins.
- 30 Premium authorized × 10 = 300
- 200 Base concurrent × 10 = 2,000
- 200 Self-Service × 0 = 0
- 20 Admin × 3 = 60
- Subtotal: ~2,360 AppPoints
- Add Health for 30 reliability engineers (~30 points), Predict for the same group (~30), Optimizer for the dispatch desk (~10): +70
- Total: ~2,430 AppPoints baseline
Compare that to 7.6 perpetual + annual SA pricing for a similar footprint, plus a separate Mobile license, plus a separate Scheduler license. The honest answer is that AppPoints can land 15–30% cheaper — but only if you do the entitlement work up front. Customers who blindly map 1:1 user-to-Premium and authorize the whole roster end up paying more than 7.6.
A rolling upgrade that does not burn a quarter
The MAS 9.x upgrade path is the cleanest it has ever been, but it is not free. Two facts shape the plan:
- Direct upgrade from 7.6.1.2 or 7.6.1.3 to MAS 9.x is supported. You do not need to stage at MAS 8.x.
- Channel Subscription is the recommended upgrade method for 9.x. Manual installers still work for MAS 8.9 and earlier, but in 9.x you should be using the IBM Operator Catalog in OpenShift.
The proven sequence is:
- Assess. Capture Maximo version, DB platform/version, Mobile, Spatial, Scheduler, industry solutions, cron tasks, automation scripts, Java customizations, BIRT reports, MIF objects, doclinks, identity provider, certs, and filesystem dependencies. Confirm 7.6.1.2 or 7.6.1.3 as the source.
- Size. Use the IBM MAS sizing calculator as the floor. Adjust upward for: actual user concurrency (not licensed count), integration volume (Kafka TPS, file-based exchanges), cron density, Mobile inspectors, document storage growth.
- Build OpenShift. Even-numbered OCP versions are required for Health, Predict-Utilities, and Collaborate because of the App Connect / Cloud Pak dependencies. Don't try to be clever on OCP version.
- Deploy Manage with add-ons and industry solutions first, then connect to a cloned Maximo database. Do not activate the upgrade on production data first.
- Activate. Database activation is the critical path. Run Integrity Checker in report mode. Confirm zero pending changes in MAXOBJECTCFG / MAXSYSINDEXES. Disable custom DB triggers. Make sure your DB version is in the MAS 9.1 compatibility matrix.
- Validate. Pods green is not "done." Validate: cron tasks, MEA bundles, BIRT, REST/Kafka integrations, API keys, doclinks, Mobile login + offline sync, search, start centers, and performance under realistic load.
- Cut over, then hypercare for 60–90 days.
Things that bite people in step 5
- Direct SQL is restricted or prohibited in managed/SaaS MAS deployments. This is a deliberate architectural change, not a bug. Any custom report, integration, or data fix that relied on ad-hoc DBA access needs a redesign — usually as a REST call, MIF object structure, or BIRT parameter.
- Custom database triggers must be disabled before activation. Document them. Some can move into automation scripts or outbox events; some cannot survive the move.
- Customization archive hygiene matters. Pull Java, XML, web.xml, DBC scripts, and third-party JARs into a clean archive. "It worked in 7.6" is not a manifest.
Java 17, OpenShift, and the platform underneath
Two infrastructure truths shape MAS 9.x:
- MAS 9.x requires Java 17. Custom Java code that targeted Java 8 needs to be revisited. Nashorn (used by the JS automation script engine) is no longer present in JDK 17 the same way it was in JDK 8 — script authors need to be aware of how this affects their code paths.
- OpenShift is the only supported runtime. Whether you run customer-managed on Red Hat OpenShift on AWS, on IBM Cloud, on Azure (via Azure Arc / Red Hat OpenShift Service on AWS), or fully managed SaaS, you are running containers. There is no more "install on a Windows server" path.
The implication: your upgrade is not just a Maximo project. It is an OpenShift project. If your infrastructure team has not run OCP 4.14+ in production with persistent storage, ingress, cert-manager, and operator lifecycle management, you are doing both jobs at once.
The cost of standing still
Doing the rough math: an organization paying 7.6 extended support in 2026 is buying one year of patch coverage with no new features. From October 2026 onward, they are on Sustained Support — which is usage questions and existing fixes only, with no security patches after 2026. For a regulated asset-intensive business, "no security patches" is rarely a defensible position to a CISO.
The other cost is opportunity. The 9.1 features — Condition Insight, Maximo AI Service, agentic workflows in MAS 9.2, the new Monitor/Health integration with RCM, the rollouts of reliability-strategy AI recommendations — are not backported. They are not coming to 7.6. They are not coming to 9.0. They live on 9.1.
A defensible plan for the next 90 days
If you are still on 7.6 and reading this on June 6, 2026, here is a defensible 90-day move:
- Week 1–2: Lock down the source-level assessment. Confirm 7.6.1.2 or 7.6.1.3. Capture the customization archive. Inventory integrations. Confirm DB version.
- Week 3–6: Build the OpenShift landing zone. Stand up MAS 9.1.18 in non-prod. Deploy Manage with the add-ons and industry solutions you actually use.
- Week 7–10: Clone the production DB, run the upgrade activation in non-prod, validate the suite of regression scenarios. Rehearse rollback.
- Week 11–13: Run a controlled pilot: one site, one asset family, one set of integrations. Hypercare with a small war-room.
- Day 90: Cut over. Decommission the 7.6 environment within 30 days.
That is not a guess. It is the pattern that has shipped in the field, on time, on budget, through 2025 and into 2026.
Bottom line: the 9.1 line is healthy, the upgrade path is well-trodden, and the AppPoints math is workable if you do your entitlement homework. The window for a calm, planned migration closes on September 30, 2026. After that, you are migrating under duress.
