Red Hat OpenShift AI
Red Hat's add-on platform for serving and governing AI models on a Red Hat OpenShift cluster, supported self-managed on ppc64le, the IBM Power architecture that also runs IBM i. It lets a Power shop host open weight models on its own hardware instead of calling an outside service.
What Red Hat OpenShift AI does
It is an add-on you install on a Red Hat OpenShift cluster. Red Hat describes it as a platform to deploy open weight models and autonomous agents at scale. An open weight model is one whose trained numbers you can download and run yourself, instead of calling someone else's service over the internet. Think of it as owning the engine rather than renting a ride. The short name for the product is RHOAI.
Does it run on IBM Power?
Yes. Red Hat says OpenShift AI Self-Managed is supported on any supported OpenShift Container Platform configuration running on x86_64, ppc64le, s390x and aarch64 architectures. ppc64le is the IBM Power architecture, the same hardware line that runs IBM i. Two version pairings matter:
- OpenShift AI operator 3.3, 3.4 and 3.5 are the versions listed for ppc64le.
- They run on OpenShift Container Platform 4.19.9 and later, 4.20, 4.21 and 4.22.
OpenShift AI parts and their status on IBM Power
| Part | Status on ppc64le |
|---|---|
| Dashboard | Generally available |
| AI Pipelines | Generally available |
| KServe model serving | Generally available |
| Red Hat AI Inference | Generally available |
| TrustyAI and LM-Eval | Generally available |
| NeMo Guardrails | Generally available |
| AI Hub | Generally available |
| Feature Store, also called Feast | Technology preview |
| KubeRay | Technology preview |
| AutoML | Technology preview |
| MLflow, MLServer and Spark | Limited or no support |
From Red Hat's supported configurations article for OpenShift AI, read 18 October 2026. Generally available means fully supported for production. Technology preview does not carry that promise, so do not build a production job on one.
What about the Spyre Accelerator?
Red Hat supports single-host deployments for IBM Spyre AI accelerators on IBM Z and IBM Power with Red Hat Enterprise Linux AI 9.6 and later, using the IBM Spyre Enablement Stack 1.1.1. A single host means one machine serving the model, not a pool of them.
Plan around this gap. Red Hat lists cluster deployments for Spyre on IBM Z only, as part of OpenShift AI 3.0 and later. On IBM Power the listed option today is the single host.
Where does IBM position it?
How Red Hat OpenShift AI landed in the Power story
- 8 July 2025IBM's Power 11 announcement names Red Hat OpenShift AI as part of how Power 11 servers run AI work.
- 25 July 2025Power 11 becomes generally available.
- Q4 2025IBM's stated availability window for the Spyre Accelerator.
Dates from IBM's Power 11 press release of 8 July 2025.
For an IBM i shop the appeal is location. The cluster can sit on the same Power hardware family as the production workload, so the model and the data stay on gear you already own and already know how to run.
Related Terms
IBM Spyre Accelerator
A PCIe4 AI inferencing card IBM ships for IBM Z, LinuxONE and Power11 systems to offload generative and agentic AI workloads from the main processor. On Power11 it is driven from a Red Hat Enterprise Linux or Red Hat OpenShift AI partition, and as of September 2026 IBM has not documented a way to use it from an IBM i partition.
IBM Power 11
IBM's current generation of Power processor and server hardware, announced July 8, 2025 and generally available July 25, 2025, carrying forward Power10's on-chip Matrix Math Acceleration and becoming the first Power generation to support the IBM Spyre Accelerator card.
IBM i Merlin
IBM i Merlin was IBM's browser-based development and modernization workbench for IBM i, built from IBM, Red Hat and ARCAD Software tools running in Red Hat OpenShift containers. It converted fixed-format RPG to free format, analyzed application impact and ran Git and Jenkins pipelines, and IBM withdrew it from marketing and service on October 7, 2025.
Matrix Math Acceleration (MMA)
An on-chip acceleration feature in IBM Power10 and Power11 processors that speeds up the matrix math operations used in AI inferencing, without requiring a separate GPU.