Will The Turbulent Economy Downdraft IBM Systems Or Lift It?

April 28, 2025 Timothy Prickett Morgan

We have always contended that recessions accelerate technology trends rather than slow them down. And if we start heading into a recession either in the United States or around the globe – it is hard to imagine one without the other – it is reasonable to assume that companies will be looking very aggressively to take automation up another level to cut costs further, to generate new lines of business, and to push profits as hard as they can in what will probably be a deflationary environment.

There are a lot of assumptions in that paragraph, so let’s pick it apart a bit by reviewing the timing of various computing revolutions and the economic difficulties that were coterminous with them. Some recessions are mild, some are extreme. People often start feeling recessional for some time before the economy starts to shrink, which is one of the factors that causes the shrink – recessions being self-fulfilling prophecies as much as economic growth. Given this, people argue about when recessions really start. For the purposes of this story, let’s just talk about when things started to get bad and a general sense of how long it stayed bad, and what change in computing was accelerated at the time. This is not necessarily an exhaustive list, but an illustrative one.

Here is the overlay of computing innovations and recessions:

Computing was pretty nascent at the time the United States went into recession in 1960, which extended into 1961. The mainframe launch was still three years away and we don’t think there was much of an impact here. Computing had really not gone commercial as yet, so there wasn’t the catalytic and enzymatic effect that we see between business and computing. These days, it is hard to tell one from the other.

The System/360 mainframe in 1964 and the System/3 minicomputer (the sixth great grandfather of the IBM i on Power Systems platform) in 1969 came out during periods of relative economic wellbeing. The System/3, which was very popular, launched into the belly of the 1969 recession, which was fairly mild and caused by an intentional tightening of monetary policy by the Federal Reserve and a pullback in government spending by the Nixon administration.

The PC revolution followed the 1973 recession, which started during thanks to an oil embargo of the United States in November of that year (among other factors). The PC held out the promise of increased efficiencies for businesses. This was largely a transformation in the front office at the time. Believe it or not, people had secretaries to do correspondence and reporting before the PC became affordable and commonplace. Offices were designed with secretaries gatekeeping managers as well as having secretarial pools. Within a decade and a half, executives had word processors and not long after that, they had email. It didn’t take long before we all became our own secretaries.

In the wake of this economic downturn, Digital Equipment launched the VAX 11/780 and IBM launched the System/38 in 1978, and both capitalized on the growth in the early 1980s and stole jobs that might have otherwise been run on IBM mainframes. DEC VAXen were far more popular in terms of numbers, and particularly among the technical computing crowd.

With the recession that spanned from 1979 through 1982, the minicomputer revolution started to take off, and at the tail end of that the wildly popular System/36 came out, replacing the System/32 and the System/34 and laying the foundational base for the future AS/400 that was launched in 1988 (also not a time of economic expansion, being in the wake of the stock market crash of 1987, but we are getting ahead of ourselves).

Black Monday, which happened in October 1987, followed a five-year bull market on Wall Street, and had numerous causes. All of the stock markets around the world declined, and the downdraft from these declining stock markets over the next two and a half years eventually triggered a recession in 1990. We think that recession actually started in people’s minds in December 1988, at least in major cities. I moved to New York City to try to get a job after graduating from college at the time, and the feel of the economy is still fresh in my mind. I got a job on the inauguration day for President Bush (the senior one), and by May I was unemployed as companies started making cuts. In July, I became founding co-editor of The Four Hundred.

The stock market crash made people jittery and caused minirecessions like I saw in NYC in 1989, but it was the Gulf War that finally pushed the world into a full-blown recession in 1990 and 1991. The economic turmoil of the late 1970s and early 1980s and research into minimalist RISC computing spearheaded by John Locke of IBM and David Patterson of the University of California at Berkeley set the stage for the RISC/Unix revolution. RISC/Unix machines were initially aimed at what we now call HPC workloads today but which eventually RISC/Unix machines showed such price/performance advantages that people ported enterprise software stacks to them and started moving applications off proprietary mainframes and minicomputer platforms. Sun Microsystems was founded in 1985 and gained steam over the next few years, and IBM followed suit in 1990 with the RS/6000, which of course is the forebear of the current Power Systems line as is the CISC-based AS/400s that came on 1988.

The dot-com boom in the middle 1990s and the desire to have cheaper computing by the hyperscalers and web titans essentially drove the architectural rise of the X86 processor in the datacenter, and the pressure on all other proprietary and RISC/Unix platforms was masked by the advent of Java at the same time, which gave customers a measure of application portability that they never had before in exchange for less efficient processing that helped prop up revenues for some time.

The dot-com boom went bust in March 2000, and a bunch of economic scandals (Enron was the big one) helped caused a recession, which obviously got a lot worse after the September 11 terrorist attacks in 2001. We think there was another minirecession in 2003 and 2004. And throughout this time, server virtualization on all platforms became the rage as companies tried to wring more work out of their machines while at the same time making them more flexible and manageable.

Throughout this time, we saw the rise of the hyperscalers and cloud builders and a push to do more data analytics, particularly for front office functions to drive engagement with customers – what is often called the digital revolution. The desire to do this better and to reduce the cost of computing really took hold after The Great Recession started in December 2007, which lasted until late 2009 or early 2010, depending on who you ask.

It was another decade before the coronavirus pandemic caused another global recession, and this time the revolution really happened at the application level, not at the system level, as we all learned how much we could use all kinds of software to do our jobs without going into the office.

There always seems to be a recession or economic downturn that drives the next IT cycle to change faster than it might otherwise would. It is hard to say for sure because there is no control group against which to measure. . . .

And so, here we sit, in early 2025, with trade wars and tariffs possibly setting us up for another recession just as IBM is trying to get its System z17 mainframes out the door and is working to getting its Power11 systems launched sometime in the second half of this year. It is hard to say how this all might affect Big Blue’s Systems group.

In a conference call going over IBM’s first quarter financial results, Arvind Krisna, chief executive officer at IBM, explained the current situation. We quote him at length so you see the full breadth of his thinking.

“Technology remains a key competitive advantage, allowing businesses to drive cost efficiencies, productivity, and preserve their balance sheets,” explained Krishna. “In the near-term, uncertainty may cause clients to pause and take a wait and see approach. However, the value of hybrid cloud, automation, data sovereignty, and on-premise solutions becomes even more critical in volatile windows. Recent conversations that I have had with clients reflect this view of the current environment. These conversations vary by industry, business, and geography. For example, our containerization and virtualization pipeline continues to grow, with clients focused not only on near-term costs, but also longer-term savings driven by our modernization capabilities. There are also areas of our business where volatility acts as a catalyst for demand, driving increased capacity requirements, particularly across our mainframe environments. This played out over the last couple of weeks amongst our financial services clients. However, for clients with a more direct impact from current policy, the slowdown may be more pronounced. Consulting is also more susceptible to discretionary pullbacks and DOGE-related initiatives. While no one is immune to uncertainty, we enter this environment from a position of relative strength and resiliency. Our clients run the world’s most essential processes. Our diversity across businesses, geographies, industries, and large enterprise clients position us well to navigate the current climate.”

With that, let’s talk about how IBM Systems, and Power Systems in particular, did in the first quarter of 2025.

Given that IBM is at the tail end of the System z and Power10 cycles, it is reasonable to expect some softness in system revenues. Everyone has known for a long time that new systems are coming, and those who can wait do wait because a new generation always carries an implicit promise of improved price/performance compared to the prior generation.

In the quarter, IBM posted sales of $14.54 billion, up a smidgen from the year ago period. IBM had a $502 million tax benefit this time last year, and this time around had to pay $103 million in income taxes, so net income shrank by 34.3 percent to $1.06 billion.

IBM’s Infrastructure group – what we still call IBM Systems as many executives and Big Blue do as well – had sales of $2.89 billion, down 6.2 percent. But thanks to capacity on demand upgrades in currently installed systems, pre-tax income rose by 13.3 percent to $248 million. In our model of IBM’s financials, server and storage revenues fell by 10.2 percent to $1.63 billion, and tech support revenues were essentially flat at $1.26 billion.

IBM gives growth rates at constant currency and as reported for Systems z and then Power Systems and storage together, but it does not give revenue figures. So we have to build a model for revenues. Here is what we think Power Systems sales look like since the Great Recession ended:

It’s not a pretty picture, of course, but it has been stabilizing and growing in recent years, and due in part to the IBM i platform as well as continued use of AIX and increasing use of Linux, which underpins SAP HANA in-memory databases that drive a lot of sales of Power Systems in recent years.

We had been expecting Power11 announcements about now, but these have been pushed out to later this year. So the revenue bump is probably going to take some time. But as IBM is embedding AI processing into the Power hardware platform, both natively in the Power10 and Power11 chips and externally through the “Spyre” AI accelerators that came out of IBM Research, there is a good chance for Big Blue to push Power Systems sales – including storage arrays based on Power and Power Virtual Server cloud capacity – to $2 billion and beyond.

It is not that much of a stretch. In Q1 2025, we think IBM sold $173 million in Power Systems iron and then booked another $75 million for storage based on Power. That alone is $248 million. In our model, IBM sold $1.59 billion in servers and storage based on the Power architecture in 2024. If you add in some Power11 upgrades this year, driven in part by AI projects, and PowerVS rentals, plus systems software and tech support, it is not hard to get well above $2 billion.

This Issue Sponsored By

Table of Contents

Content archive

Recent Posts

Subscribe

Pages

Search