The Frustration Of Not Knowing How We Are Doing
July 23, 2018 Timothy Prickett Morgan
The old adage from management guru Peter Drucker is that you can’t manage what you don’t measure, and it is something that organizations take to heart to run themselves. There is such a thing as taking this all too far, where you spend all your time measuring and managing and not enough time getting actual stuff done, and then there is not having enough information to actually manage well.
The Securities and Exchange Commission, which regulates public companies, does not, to my knowledge, have explicit guidelines as to how companies report their product and divisional categories. And thus, with each quarterly report, every vendor does what it thinks is material to the running of the business and therefore gives Wall Street an idea of what it is doing and how it is doing as it sees fit. In the datacenter racket, Hewlett Packard Enterprise and Cisco Systems give pretty detailed reports, and IBM does a fair job but is not explicit about how various products are doing. To be sure, it is a whole lot better than just getting the normal split of products and services that smaller companies offer up as insight, but we actually do not know how various products such as the System z mainframe or Power Systems hardware and software are doing. We get some generalities, a few growth rates at constant currency, and some very general categorizations that, as we have expressed before, are not exactly reflective of how Big Blue actually sells systems.
I find this particularly frustrating, and got so annoyed in analyzing the company’s latest quarterly results, reported last week, that I pulled out that empty spreadsheet and started building a model of what IBM’s Power Systems and System z business might actually look like. After a few hours of hunting and pecking around public documents from IBM and the box counters at IDC and Gartner, which provide some insight, I stopped about halfway through because this could eat a whole day and other stores needed to be written for this issue. But I will finish the model and share it with you when it is done. I look forward to your input once it is done.
In the meantime, let’s go over IBM’s second quarter financial results and talk about the publicly available trend data that will help us put it into perspective.
First, let’s do the numbers. IBM crested above $20 billion by a smidgen in the quarter ended in June, up 3.7 percent from the year ago period. The company’s net income rose by 3 percent to $2.4 billion, and obviously long gone are the days when net income grew at a faster pace than revenue for IBM. The IBM Systems group, which peddles servers, storage, whatever networking IBM adds in, and operating systems, had $2.18 billion in sales to external customers, up 24.6 percent, plus another $242 million in sales to other IBM groups, an increase of 36.7 percent. Overall systems hardware came to $2.42 billion, up 25.7 percent. If you look at just the hardware, IBM pushed $1.8 billion in iron, up 32 percent, and operating systems brought in another $388 million, down 1 percent. IBM frustratingly only now gives out constant currency growth rates rather than both reported and constant currency, so it is harder to build an accurate model of its sales. The overall Systems group gross profit came to $1.09 billion, up 19.8 percent, and pre-tax income nearly quintupled to $346 million. All is good, right? Well, that depends.
The second quarter of 2017 was a pretty easy compare, and income was pretty low as IBM was investing heavily in the Power9 and z14 processors for its systems. But the Power AC922 CPU-GPU supercomputer node coming out last December, the “ZZ” general purpose Power9 machines coming out in February, and the “Boston” Linux cluster machines coming out in May, we would have expected a pretty big jump in Power Systems sales. IBM had just finished installing nearly 10,000 of the Power AC922 nodes in the “Summit” supercomputer at Oak Ridge National Laboratory and in the “Sierra” supercomputer in Lawrence Livermore National Laboratory, so this should have helped quite a bit, particularly if all of the GPUs and switching were booked as IBM revenue (as they should have been since it is the prime contractor for these two machines). But instead, the Power Systems business only had 4 percent growth at constant currency, which probably translates to around 2 percent or so real growth.
We don’t know, to tell the truth. But this is certainly not what we expected. The 112 percent revenue growth, against a 200 percent increase in aggregate MIPS computing capacity sold in the year ago period that IBM had for the System z line – yes, that is 3X more capacity – thanks to the ongoing rollout of z14 machines, is precisely what we expect. Linux on the mainframe seems to be taking off like white lightning, a second or third boom, including now blockchain and machine learning applications running natively on the platform. Like we have asked Big Blue to do with the IBM i platform for years. To little avail. The z14 boom is bigger and better than the z13 upgrade cycle, but it seems that the Power9 upgrade cycle is weaker than the Power8 one, and for no good reason we can contemplate and in the absence of any real insight from IBM, we are perplexed.
In the interests of helping give out some insight, here is a collection of IBM data on its systems business as well as auxiliary data from Gartner and IDC, who track certain aspects of the server business each quarter. Take a look:
That’s the starting point of any model. It is not easy to reconcile all of this, to be sure. But if you want to mix IDC and Gartner data – which is dubious because they don’t count server revenues in precisely the same way – you can actually get something akin to an IBM i on Power Systems server estimate, which is shown in red bold italics in the table above. It does not insult my intelligence that the IBM i on Power Systems business might be generating somewhere around $500 million a year in hardware sales. But IBM surely isn’t going to tell us this. It used to be almost 10X that back at the peak in 1998 twenty years ago, mind you. (I was told by the IBMers in charge of the AS/400 business in 1998 that they sold $4.7 billion in hardware, and that turned out to be the peak.)
This model will take some time, so bear with me.
Other parts of IBM’s business had their ups and downs in the second quarter, as you can see from the chart below:
IBM has created this vague category called Cognitive Solutions, which includes its middleware and database software as well as data warehousing, analytics, and various Watson services and software. (You can imagine how little I like such lumping.) In Q2, the cognitive software part of this business had sales of $3.13 bullion, up a bit as we reckon it from the data IBM supplies (and not in constant currency as IBM talks about it), while transaction processing software accounted for $1.45 billion, also up a smidgen. Sales inside of IBM to other units and to outside customers added up to $5.28 billion, up 1.3 percent, and but gross profits fell by 3 percent to $3.5 billion; pre-tax income nonetheless rose by 8.7 percent to $1.76 billion.
On the services front, IBM had $4.28 billion in Global Business Services revenues, up 2 percent, and this is comprised of consulting, process services, and application management (what we used to call outsourcing). The pre-tax income here is pretty skinny, only $385 million, but it was up nearly 20 percent, which is good. Technology Services and Cloud revenues are reckoned separately, and came to $8.78 billion, up 2.4 percent, and pre-tax income was down 12.5 percent to $883 million in the second quarter. IBM’s total Global Financing revenues were up 23 percent to $867 million, including internal and external sales, and pre-tax income was up 26.6 percent to $357 million.
If you go through and allocate portions of the divisions that are part and parcel of the real IBM systems business – not including databases, middleware, or application software but including operating systems, tech support, integration and transaction processing software, and financing – then we estimate that IBM’s real and core systems business accounted for $6.54 billion in revenues, up 8.4 percent, and gross profits were around $3.6 billion, up 3.7 percent. That’s a gross profit margin of 55 percent, which is why IBM stays in the system hardware game.