IBM Power Pricing FAQ

The Power Value Index is this site's own independently developed 0-100 composite score ranking IBM Power8 through Power11 on six weighted inputs: performance (25%), remaining useful life (20%), capacity (15%), software economics (15%), market price (15%), and value retention (10%). It is not an IBM-published metric. See the Power Value Index page for the full methodology and every generation's score.

By the Power Value Index, Power10 currently scores highest on value (76 of 100) even though Power11 scores highest overall (86) on raw capability. Power10 was just withdrawn from new-unit marketing (31 July 2026), so its secondary market is forming now, its price is softening, and it still carries several years of expected support runway -- a combination that outscores Power11's zero-discount new-build pricing on the value measure specifically.

It depends on your time horizon. A used Power9 scores well on acquisition cost and current performance, but its standard IBM support already ended for scale-out models (31 January 2026), so you are buying a short, defined runway at a discount. A new Power11 costs more up front but starts a full support lifecycle with no announced end date. For a workload you plan to keep 5+ years, Power11 (or Power10) usually wins; for a 1-3 year bridge or a small, non-critical partition, a used Power9 can be the more rational buy.

IBM i licensing is set by software tier (P05 through P50), independent of which Power generation you run it on. As of the most recent reporting, P05 runs about $2,356/core/year on subscription or $3,315/core as a one-time perpetual purchase; P20 runs roughly $14,500-$16,000/core/year on subscription. See IBM i Tier Economics for the full tier table and how the hardware generation you choose changes how many cores you actually need to license.

IBM i tier pricing is per core, and each generation delivers more CPW per core than the one before it. A newer generation can often clear the same performance requirement with fewer licensed cores, which lowers your effective software cost even though the published per-core tier price itself does not change by generation.

Annualized ownership cost spreads a system's full multi-year cost -- amortized hardware, ongoing maintenance, and software licensing -- into a single per-year figure, making generations with very different up-front prices directly comparable. This site models it over a representative 5-year hold using IBM's published ~15% hardware maintenance baseline. See Annualized Ownership Cost for the full model and its assumptions.

Power8's standard IBM maintenance ended across the full line by October 2024, so any Power8 still in production today is on extended or third-party support, which typically escalates faster than standard-support pricing and carries growing parts-availability risk. Combined with Power8's bottom-ranked Power Value Index score (40 of 100, driven mainly by near-zero remaining useful life), most Power8 workloads are past the rational keep-vs-replace breakeven point already. See Upgrade Economics for the decision framework.

Yes, and by this site's model it is currently the single best value-scoring generation. Power10 was withdrawn from new-unit marketing 31 July 2026, meaning IBM will no longer sell it new, but that is precisely when secondary-market pricing typically softens fastest while genuine support runway (likely several more years, though no end-of-service date is announced yet) remains intact.

For Power8 and Power9, RUL uses IBM's own published end-of-service dates. For Power10 and Power11, no end-of-service date has been announced, so this site models RUL from the historical ~8-10 year gap IBM has kept between general availability and end-of-service on the last two generations (Power8: 2014 GA to 2024 EOS; Power9: 2018 GA to 2026 EOS). This is disclosed as a Krisada Estimate, not an IBM date, everywhere it appears.

No. Every model has a configuration zone where cost per unit of usable capacity is lowest. Past that zone, added cores can push an IBM i partition into a higher software tier, or added memory can exceed what IBM i can actually use on a given processor feature code, so cost rises faster than usable throughput. See Configuration Efficiency for where that point falls per model class.

It depends heavily on workload, IBM i dependency, and how you value operational control versus elastic scaling. IBM Power Virtual Server offers pay-as-you-go pricing for IBM i, AIX, and Linux workloads without on-premises hardware ownership, while x86/Windows or Linux alternatives avoid IBM i tier licensing entirely but cannot run IBM i workloads natively. See Competitive TCO for the framework and what data is and is not publicly verifiable on this comparison.

For a small, single-partition P05 or P10 workload, an entry scale-out system (Power11 S1112-class, or a used Power10 S1012/S1014-class system) typically delivers the best value: these entry models are sized specifically to stay inside the smallest software tiers without paying for scale-out headroom you cannot use. See Workload Value for sizing guidance by workload profile.

No. The Power Value Index is an independently developed analytical framework built by this site, not an IBM-published or IBM-endorsed metric. Every input is sourced where a public figure exists (IBM benchmark data, IBM end-of-service announcements, third-party pricing reporting); inputs without a citable public source are explicitly labeled as a Krisada Estimate on the Power Value Index page.

Hardware acquisition ranges are modeled from publicly observable secondary-market dealer listings (Alta Technologies, Spectra Logic, Flagship Technologies, EssentialTech) and marketplace data. IBM i software tier pricing is sourced from IT Jungle's ongoing coverage of IBM's published subscription and perpetual pricing changes. Every dataset page on this site cites its specific sources and flags any modeled (non-cited) figures.

Datasets are reviewed and republished as IBM pricing, end-of-service announcements, or observable secondary-market conditions change materially. Each dataset page and JSON file carries its own updated_at date; check that date against IBM's current announcements before using any figure for a purchase decision.

The configuration tool asks about your workload size, budget priority (lowest cash cost vs. longest runway vs. best raw performance), and planned hold period, then weights the same six Power Value Index components against your answers to surface which generation fits best. It runs entirely in your browser against the published PVI dataset -- no data is sent anywhere.