AI Inference Partition
A separate supported operating-system partition used to execute an AI model while an IBM i partition may remain the production application and data source.
What an AI inference partition is
A Power server can be cut into partitions, each with its own operating system. An AI inference partition is one of those cuts, set aside to run the model. The IBM i partition next to it keeps being the production application and data source.
What the partition boundary buys you
The boundary lets you size the inference runtime on its own and govern it on its own. Give it the cores and memory it needs without taking the production reserve. Patch it, restart it, or throw it away without touching IBM i. IBM's July 2026 Power S1112 announcement included a customer running Linux partitions this way to explore AI.
What an inference partition does not settle
- Whether the model is compatible with the runtime in that partition
- Whether an accelerator is supported in that configuration
- Whether the partition can reach Db2 for i data securely
Those three need a workload-specific pilot. The partition is the room. It is not proof that the furniture fits.
Related Terms
AI Workload Placement
The decision about which system, operating environment, and partition executes an AI model or agent, separately from the IBM i application supplying its data.
IBM i
IBM's operating system for Power servers running traditional AS/400-lineage business applications, with Db2 for i built directly into the OS.
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.