State Of The Power Systems Base 2025: The Systems

February 17, 2025 Timothy Prickett Morgan

In last week’s issue, we talked about the upgrade cycle for the IBM i operating systems over more than a decade based on the annual IBM i Marketplace Survey done by Fortra and pretty heavily massaged by us to reflect what we think is closer to reality for the IBM i market at large.

No matter if you look at the raw data from the survey or the more elaborate model we have built derived from the past eleven surveys, one thing is clear: There is a regular pattern for Power Systems upgrades among OS/400, i5/OS, and IBM i shops both for the operating systems and for the underlying hardware that supports them. Because of backwards compatibility – a beautiful thing that users demand of the IT sector from the beginning, allowing for new software to run on older hardware – the operating systems tend to get ahead of the hardware.

Based on our model, for instance, only about 9.3 percent of Power Systems machines running IBM i are on Power10 hardware as of the end of 2024, when the Fortra survey was performance and which is revealed in the 2025 report. That is what the “2025” means in the charts and table below. This is not a projection for what the Power Systems installed base will look like among IBM i shops at the end of this year. So do bear this in mind.

In our model based on the raw data in the 2024 report, Power10 iron represented 5.3 percent of the installed base, and in our model of the 2023 report, Power10 was around 1.2 percent of the base. This data is significantly lower than the raw data that came from the Fortra report, which looks like this:

Believe me, we would much rather believe that the entire installed base was as far ahead as this survey data indicates. But we have talked to far too many customers, and far too many business partners with lots more customers, who tell us all that they are on much older iron and much older operating system licenses than the annual Fortra survey indicates. Our model is intended to rectify this impedance mismatch between the survey data and the anecdotal evidence about the IBM i installed base and the related Power Systems installed base.

We think there are IBM i laggards as well as IBM i leaders, and Power Systems laggards as well as Power Systems leaders. We think further that of the 120,000 unique IBM i customers worldwide that are estimated to run one or another variant of IBM i on Power Systems, about 30,000 of them keep relatively up to date with their hardware and systems software, and the remaining ones are hanging back on older servers running back releases – many of them that have long since gone off standard support and are even out past extended support.

In our model, we take the raw data in the IBM i Marketplace Survey, we build a model out of the raw data and try to adjust for the leader/laggard split. We assume that the operating system and hardware levels of the laggards are four or five years back from the current leaders and apply it to three-fourths of the installed base of iron that we think represents the laggards. (We get a machine count by looking at the machine count distribution of the installed base that is in the demographics of the Fortra survey, which gives us an average number of machines at each customer, which we then apply across a base of 120,000 unique customers.)

The column chart above in the latest Fortra report doesn’t go back across the full dataset, showing the share of the machines by Power processor type, and so we have gathered all of the data up for the past eleven years and also provided some granularity on the older machines that have been lumped into the “Power6+ or older” category:

Again, remember this data show in the charts is for the year the report was published, not the year the data was gathered for the report. So this is more of a statement about the state of installations in the fall of the prior year to the publication of each Fortra report.

We think that this data does indeed accurately reflect the distribution of machines across what we call the Power Systems leaders, and it would not be hard, given this data, to create a pattern that would probably reflect what happened even a decade earlier this is so regular. It is kinda cool how regular this pattern is, in fact.

But, as we pointed out, we do not think that this pattern of machine distributions over time reflects the complete Power Systems base running IBM i. if you assume that the Power Systems laggards are around five years behind the leaders, and then backcast that data across 90,000 customers and get a distribution of machines across the base (accounting for the fact that many customers have more than one machine) and adjust that further for machine consolidation and the fact that more and more IBM i shops are running larger numbers of partitions on their machines, then you get a machine distribution that looks like this:

We do not think for a second that more than half of the actual machines running IBM i are Power10 and that more than half of the machines have Power9 machines, which the raw data from the latest Forta survey suggests. We do think that Power9 installations are growing, some with new iron in the reseller channel and others with second-hand gear that is taken out as companies move to Power10 machinery. We do believe that Power8 machinery peaked about two years ago and that Power7/Power7+ machinery peaked six years ago. And we similarly believe that there is still a fair amount of Power6+, Power6, Power5+, and Power5 iron in the field, as hard as that might be to believe.

And that is because we know if customers who are trapped on IBM i 5.4, IBM i 6.1, or IBM i 7.1 because they can’t move their applications forward either because their application vendor doesn’t exist anymore, they have gone off maintenance for their applications and they can’t afford the money or the disruption to get those applications current, or they write their own applications and they do not have the resources to update them for a newer OS and the newer hardware it would require.

If history is any guide, the Power10 iron will comprise about 20 percent of the installed base as 2025 comes to a close, even with the Power11 machines coming later this year, and will rise to be about 30 percent by the end of 2026 and maybe hit 40 percent by the end of 2028.

Here is our model data for Power Systems machine counts at the 120,000 sites using IBM i worldwide, with a total of just under 285,000 machine in the field.

We think that Power11 will have a tiny uptake this year, considerably more in 2026, and quite a bit in 2027 and 2028 because history is indeed a strong guide when it comes to the Power Systems and IBM i markets.

This Issue Sponsored By

Table of Contents

Content archive

Recent Posts

Subscribe

Pages

Search