Drilling Down Into IBM’s System Group
January 23, 2017 Timothy Prickett Morgan
It is a new year, and we are all looking forward to a growing economy and some much needed competition for compute in the datacenter as there is a resurgence of interest in the Power processor and a growing probability that ARM server chip makers, particularly Qualcomm and Applied Micro, are going to see some traction this year.
It is with this in mind that we ponder IBM’s fourth quarter financial results, which were announced last Thursday, and more specifically think about what may be in the cards for the Power Systems platform in 2017 and beyond. First, let’s go over the numbers.
The most obvious thing is that IBM’s revenues and profits continue to shrink, but the downside is getting smaller and smaller, and we think that IBM’s core systems business will start to level out this year and maybe even grow by the third or fourth quarter, depending on when Power9-based Power Systems and z14-based System z mainframes hit the market. In the final period of 2016, IBM’s overall revenues were $21.77 billion, down 1.1 percent from a year ago, and net income rose by nearly a point to $4.5 billion. This is sure a lot better than a year ago, when IBM’s revenues fell by 8.4 percent to $22 billion and its net income fell by 18.6 percent to $4.46 billion. For the full 2016 year, IBM’s revenues were off 2.1 percent to $79.85 billion, but its “real” systems business, which includes servers, storage, switching, systems software, databases, transaction monitors, and tech support and financing for its own iron, fell by 8.3 percent to $26.1 billion. (That’s our estimate; IBM does not break out sales this way, but we have some pretty good guesses on how it all breaks down.)
The good news is that IBM’s declines in this core systems business are slowing, even if it is not at quite the same pace as the overall business. IBM’s efforts to promote SoftLayer cloud and Watson cognitive computing, mobile and social and marketing software and tools, and security wares – what it calls its strategic imperatives – are almost filling in the gap left behind as the core businesses shrink. IBM wanted these strategic imperative businesses to reach $40 billion and 40 percent of revenues by 2018, and in this quarter it already hit the 40 percent mark, with $33 billion in revenues for 2016–as much because of its overall revenue decline as for the growth in these businesses.
Here is a table below that summarizes how IBM did in the past four quarters in the new divisions and groups it set up in early 2015 after it sold off the System x server business to Lenovo.
In the Systems group, this includes Power Systems and System z mainframes and their operating systems plus storage hardware. This presentation does not include internal revenues for servers shipped to other IBM divisions and groups. For the full year, IBM sold just over $8 billion in Systems products, and brought $934 million to the middle line as pre-tax income. That profit level is more or less in line with its Global Business Services group, and about a third of what it derives from its Cognitive Solutions group, which sells its database, data warehousing, data analytics, and transaction processing wares. Calling this “cognitive,” which implies machine learning or some form or artificial intelligence, is an obvious stretch. This group largely sells batch and online processing software, to tell the truth. And no matter how you cut it, the services portions of IBM account for two thirds of the revenue stream, more or less. IBM is still, after 20 years, largely the services company that former CEO and chairman Lou Gerstner envisioned it could be.
As for Systems, IBM chief financial officer, Martin Schroeter, did not offer much insight into how the Power Systems business was doing and provided a little more insight into how the mainframe was faring. Schroeter said that Linux-based Power Systems machines now drove 15 percent of revenues, and that is pretty good considering that two years ago it was a few percent of sales. Some of that increase is due to the ongoing decline in the Unix market, which has pulled Power Systems down, of course. But some of that is IBM actually selling Linux-based Power Systems iron to customers who might otherwise have bought X86 iron. The Linux server portion of the Power Systems business grew in the double digits during the fourth quarter, and the remaining portions of the Power Systems business declined. IBM’s top brass never talks about the IBM i platform, so we basically know nothing. On the System z front, revenues rose 4 percent and IBM saw double digit shipments of MIPS capacity into the base compared to the prior year’s fourth quarter. Systems hardware sales were off 12 percent in the quarter to just under $2.1 billion, and operating system revenues were off 12 percent as well to $467 million. Pre-tax income at Systems group was 21.6 percent of revenues, almost exactly the same as the prior year when IBM sold more iron and OS software, which is pretty good.
We expect to see Power9 boxes ship initially around the middle of the year, with z14 machines later in the year, with volumes ramping into early 2018. The comparisons to 2016 and 2017, when the Power8 and z13 were at the tail end of their product cycles and revenues had been sliding for both. With IBM no longer selling X86-based servers, it has to ride out its own product cycles, and it is unfortunate for the company that the z and Power lines are on about the same upgrade cycle. The difference is that mainframe customers buy machines with fully populated systems and activate cores as needed, so at the end of the mainframe cycle when companies are just activating cores, the profit margins are actually higher because there is nothing to manufacture or ship. You flip some bits and send a bill.
That is something that all Power Systems machines should have, we think, and hopefully this capacity on demand capability, which is already in some midrange and all high-end Power Systems servers will be extended down to the smallest machines IBM sells.
As we have pointed out before, we think IBM has a much larger real systems business, and in the fourth quarter, this aggregate of products had $7.43 billion in sales by our estimates, down 7.7 percent. In addition to the Systems hardware and operating systems mentioned above, this included around $1.56 billion in transaction processing software of various kinds (mostly CICS and other things running on mainframes), $1.27 billion in integration software (including myriad WebSphere products), $1.37 billion in tech support, and $682 million in financing.
Past data is no guarantee of future data, of course, but it sure looks to us like the components of IBM’s real systems business have found their floor as 2016 came to an end. There may be a few bumpy quarters between now and when Power9 and System z14 machines ship, but if IBM gets clever with marketing and promotions, it may be able to fill in some hardware gaps. As we have said in the past, the vast majority of IBM i shops are not on Power8 iron, and if IBM has a lot of this gear sitting around in the barn, it might make sense to have a fire sale and get large numbers of customers moving ahead rather than letting it sit and collect dust. This will better position IBM i shops for the Power9 future while at the same time preserving whatever volumes IBM needs for its cutting-edge Linux clients in analytics and supercomputing. An aggressively priced Power8 machine with innovative packaging of software and services will be just as good as a shiny new Power9 machine without those things to an IBM i shop.
The mantra for 2017 should be: Get current, and stay current.