The Low-Down On IBM’s Power Systems Sales
March 7, 2022 Timothy Prickett Morgan
Late last year, IBM, under the new regime of Arvind Krishna and in the wake of the acquisition of Red Hat and the spinout of the Kyndryl managed services business, reorganized its businesses and the financial reports it gives to Wall Street investors. Generally speaking, the way Big Blue is talking about itself, from a financial perspective, is more accurate than the reorganization it did after Ginny Rometty had been at the IBM helm for a bit. Which is good. But the one downside is that we lost visibility into the Power Systems platform.
We lamented this when looking at the restructuring of financials when they were announced in October 2021, and were a little surprised when IBM’s top brass didn’t say anything specifically about Power Systems at all during its discussion of its financial results for the fourth quarter of 2021, which we covered in January. This bothered us, and while we had a model to try to figure out Power System sales based on its old reckoning of sales of its Systems group, it was not immediately obvious how to do this with the current presentation.
But we stared at the screen for a while, and then the pie charts and tables IBM has put out, and made some assumptions about Power Systems sales and storage sales from the old data, and have come up with a model for Power Systems sales that covers the full years of 2019, 2020, and 2021. We do not have quarterly data because of the nature of the information that IBM has supplied with the new categories, but we have been able to make a pretty reasonable guesstimate for annual results for these years. We also make the assumption that technical support for each IBM platform – that would be System z mainframes, Power Systems, and storage – is proportional to the revenue streams from that, and have allocated technical support revenues for Power Systems based on this. With this assumption, which is reasonable, Power Systems is about a quarter of revenues for systems and support.
So without further ado, here is IBM’s annual revenues in its new groups and divisions and what we did to pull apart Infrastructure revenues to come up with a Power Systems platform revenue, which includes servers, operating systems, and support:
Despite the fact that the Power9 systems were at the tail end of their cycle and due in part to some high-end Power10 system sales as 2021 came to a close, it looks like Power Systems server revenues (including operating systems, which used to be calculated separately) were essentially flat in 2021 at just a tad over $2 billion. If you do the ratio math on Infrastructure Support, then we estimate Power Systems support represented another $1.47 billion in sales, also essentially flat in 2021. So add it up, and overall Power Systems revenues were up four-tenths of a point to $3.48 billion, and representing 24.5 percent of overall Infrastructure sales for Big Blue.
An important way to think about this is that over the past four years, during the era of the Power9 processor, IBM has probably raked in somewhere around $15 billion in revenues, and even if the Power9 platforms burned $2.5 billion in investment for processor, systems, and software development, the return on investment has been a factor of 6X.
The trouble is, at a $3.5 billion annual business, we don’t think IBM is profiting at an operating level, and that it needs somewhere north of $4 billion a year in revenues to be breakeven and somewhere closer to $5 billion to show a reasonable profit. To be fair, when you add in systems software, database software, middleware, and financing for Power Systems gear, it is quite possible that the Power Systems line is profitable overall. It is hard to say for sure.
Of course, IBM could eliminate all of the confusion and just tell Wall Street and its customers. . . . But it won’t.