Resource Library
Legacy AI-on-IBM-i articles and Power hardware research. New infrastructure decisions are organized in the workload-placement and Power inference guides.
Which IBM AI Hardware Do You Actually Need?
Most IBM i shops do not need to start with an accelerator. Match the job to the Power processor you own, the Linux runtime beside IBM i, or a Spyre card only after the workload proves it needs one.
From 2 nm to 0.7 nm: How IBM Actually Makes AI Chips
IBM sold its chip factories and kept the laboratory. In June 2026 that laboratory announced the first transistor smaller than a nanometre. Here is what that means for anyone budgeting AI hardware over the next decade.
IBM Granite: The Open Model That Comes With the Hardware
An accelerator card is an empty lorry. Granite is the cargo, IBM gives it away under an open licence, and the memory savings in the newest versions decide how many cards a project actually needs.
IBM Put Claude In Its Own Developer Tools. Does That Change Your Server Plan?
IBM partnered with Anthropic in October 2025 and reported a 45 percent productivity gain across 6,000 internal users. It is a genuinely large number. It is also not a reason to buy hardware, and the announcement never mentions IBM Power.
InstructLab: Teaching a Model What Only Your Shop Knows
Every AS/400 shop has one person who knows why the month-end job runs in that order. InstructLab is the open source attempt to get that knowledge out of their head and onto your own hardware before they retire.
NorthPole, the AIU, and Where Your Spyre Card Came From
The card you can order for a Power 11 server has a family tree, and reading it tells you what IBM thinks AI hardware is for. Four chips, one stubborn argument: keep the model next to the data.
The Spyre Card Will Not Start Without Red Hat. Here Is the Whole Stack.
IBM's own documentation lists Red Hat AI Inference Server or Red Hat OpenShift AI as a requirement for running a Spyre Accelerator on Power 11. Most budgets we see have the card and nothing else.
How On-Chip Matrix Math Acceleration Changes AI on IBM i
IBM Power10 and Power11 processors build Matrix Math Acceleration directly into every core. Here is what that means for running AI inferencing workloads alongside production IBM i partitions.