AS400 Hardware Guide

It depends heavily on whether the buyer is looking at new, refurbished, or upgraded existing hardware. An entry Power server typically starts in the high four figures to low five figures, while multi-core enterprise configurations and full new deployments can run well into six figures once storage, networking, and infrastructure are included.

Refurbished and used systems usually land 40 to 70 percent below comparable new hardware pricing, depending on generation, condition, and warranty terms. The most useful first step is sizing actual workload requirements, since price comparisons only make sense against a specific configuration.

The honest answer depends on remaining IBM support runway, budget, and how critical the workload is. New hardware carries the longest support and warranty window but at the highest price. Refurbished hardware can deliver most of the same capacity at a meaningfully lower price, provided the seller tests and warranties the unit properly.

Buyers should compare total cost of ownership over the realistic hardware lifespan, not just sticker price, and should always confirm the refurbished seller's testing process before assuming price alone makes the decision.

Support windows vary by generation, but IBM's standard hardware support typically runs several years past a model's initial release before moving to limited or end-of-service status. Once a generation exits standard support, buyers depend on independent service providers and secondary market parts to keep systems running.

Buyers running legacy hardware should check the current lifecycle status of their exact model, since running past end of support is a real operational risk, not just a compliance formality.

If the server's processor and memory still have headroom and the bottleneck is genuinely storage-related, expanding DASD or adding SAN attachment is usually the faster and cheaper fix. Pull current ASP disk busy statistics before assuming storage is the actual cause of a performance issue.

If storage, processor, and memory are all constrained at once, that is usually a sign the whole server has reached its practical ceiling, and a server replacement conversation makes more sense than a series of piecemeal upgrades.

If the save window is straining against current physical tape throughput but the underlying save scripts still work well, a virtual tape library often solves that problem without a script rewrite, since it emulates physical tape to existing jobs.

Buyers who need offsite tape for compliance or disaster recovery should confirm the virtual tape appliance supports exporting to real physical media, since not every configuration is set up for that path by default.