The Shape Of The IBM To Come
October 11, 2021 Timothy Prickett Morgan
Big Blue is getting ready to spin off its managed services businesses, which represents about a third of its revenue stream and a big chunk of its employee base, into the separate company called Kyndryl and hopes to have this task done by the end of the year. Last week, IBM’s top brass had a virtual meeting with Wall Street to host its Investor Day, and IBM’s chief executive officer, Arvind Krishna, and the rest of the team unveiled a new segment and financial reporting structure that will be put into effect once the Kyndryl spinout is done.
At that time, IBM will no doubt do its financial reporting for the quarter and backcast the new way of talking about product lines and money into prior quarters, and if history is any guide, it will do this for all of 2020 as well as whatever bits of 2021 are done at the time. And everybody, including us, will set about to recast our IBM revenue models.
More than anything else, an Investor Day briefing by a relatively new CEO gives IBM a chance to show how it has been rethinking its positions in the IT market and how it will change to better address the market and drive growth. And for once, we have an IBM that understands that system z and Power Systems infrastructure as well as the IBM Cloud, with all of its X86 iron, is the foundation that the rest of the company is built upon. But make no mistake. Red Hat is the whole first floor above that foundation, and this is why IBM paid $34 billion to acquire it three years ago. Here is the IBM, with its hybrid cloud focus, that Krishna is running:
That is an amalgam of two charts that IBM put out, which we consolidated into one so you can see it all in one place. The Infrastructure segment, which includes server and storage hardware and their operating systems, as you might expect, but also now includes sales of raw IaaS capacity on the IBM Cloud and OEM asset recovery services brought in from the Global Financing arm, which is now just called Financing and just does financing, not reselling of certified pre-owned gear. This really is, for the first time that I can remember, reflective of the true IBM’s systems business, in that it also includes infrastructure support (break/fix and installation support) for all of IBM’s hardware and operating systems as well for the third-party gear that IBM provides support for as well. This represented around $14 billion in sales in 2020, according to Jim Cavanaugh, IBM’s chief financial officer.
By the way, that Distributed Infrastructure category includes Power Systems hardware and operating systems plus all of IBM’s storage arrays and clusters – but not its storage software, which for some strange reason is put into its Transaction Processing software category. (Correct – that makes no sense whatsoever. It is just IBM trying to hide either a decline in transaction processing software or storage software.)
IBM’s financial aspirations are not high, but practical, and this is a good thing for IBM i shops who have made enormous investments in their applications and people over the decades. An IBM that is happy selling Power Systems to them is the IBM that they need right now. One with a mix of legacy and new platforms that can be operated cooperatively, collectively, and coherently. This is the balance that IBM has always tried to strike, and sometimes it gets it right and sometimes it doesn’t. Krishna needs to get the balance just right for Wall Street as well as customers, which is a delicate job indeed. All of the data that IBM dumped on Wall Street didn’t really move the company’s stock price or market capitalization, and it is generally up this year. Which is a kind of good news for a company that is in its 12th decade of existence.
IBM wants to drive mid-single digit revenue growth each year, have a tighter focus on its product lines and efforts, and generate $35 billion in free cash flow over the next three years so it can reinvest in the business and possibly make some acquisitions. The company is also going to have to pay down some of the debt it took on to acquire Red Hat, obviously.
The way that IBM sees it – and particularly, Paul Cormier, who is CEO of the Red Hat division of IBM – is that hybrid cloud and multi-cloud is the reality of enterprise IT, and IBM and Red Hat together are making hybrid, multi-cloud environments “consumable” in much the way that the Red Hat Enterprise Linux stack in the early 2000s made an open source stack – not just a Linux kernel – consumable. Cormier said that 95 percent of the customers polled by IBM have a mix of public and private cloud infrastructure (they also have a lot of standalone host machines, but we are not talking about that right now, apparently), and that the average number of public cloud platforms in use by companies is 2.6 and, oddly enough, the average number of private cloud platforms in use among those polled was 2.7. In IBM i shops, we know for a fact that IBM i is generally being run in a virtualized manner and so is Windows Server, which is by far the dominant second platform at these companies. So that is an average of 2.0 private cloud platforms right there. . . . If you stretch the definition of a cloud the way that people do when they talk about private cloud. (To our mind, it is a virtualized server unless it has metered pricing for the use of that infrastructure in blocks of time. But we can also see that a virtualized IBM i box acquired and used for five years would be essentially the same thing economically speaking as renting time slices on the same machine with reserved cloud pricing.)
The ability to drive that revenue growth and free cash flow starts with the infrastructure foundation, and for the first time in a very long time, IBM is talking about infrastructure in a way that is illustrative and clear about its business. Mostly. Here is the pie chart for how IBM’s revenues are expected to break down for its Infrastructure segment once the Kyndryl spinout is done:
In this pie chart and the two that will follow it, I have gotten out the ruler and protractor from my drafting set to calculate the percentage slices of the pie for those who think in numbers as well as visually.
An important note: The one thing that IBM did not do in its Investor Day briefing is give out any numbers for 2021 projections. But it did backcast this presentation into 2020 data, which allows us to build a better model of IBM’s real systems business. (Well, I hope it does, anyway.). But, like we said, it will do that, and we can build our IBM systems model from there. IBM will not provide finer-grained detail here, we suspect, but it might offer hints about Power Systems sales or System z MIPS shipments, as it has done in the past. The important thing to remember, and Ric Lewis, the new general manager of the Infrastructure business at Big Blue, pointed this out, is that with each of the three last generations of mainframes, the revenue from each generation has grown and the aggregate MIPS shipped within each generation has also grown. The amount of mainframe capacity in the installed System z base is now 3.5X the size it was in 2010 and is steadily growing, albeit at a modest pace. Finally, for every $1 that System z customers spend on hardware and operating systems, they spend another $3 to $4 on additional systems software, storage, and support. So System z represents a very large portion of IBM’s overall revenue stream, and an even bigger share of its profits – as we have been saying for years. The ratio is not as high for Power systems, but my guess is that it is somewhere around $2.50 for every $1 spent on a Power Systems machine with an OS on it.
Over the mid-term of the next three years, IBM expects for the Infrastructure segment to have flat sales, give or take given the product cycles for z and Power iron. Hey – at least it is not down.
Here is how the Software segment will look, based on the 2020 data, for which we added the percent share and the estimated revenues from each line:
OK, let’s break this down a bit. The Red Hat slice of the Software pie includes Red Hat Enterprise Linux, Ansible automation tools, and the OpenShift Kubernetes stack. Things like JBoss middleware are in the Automation slice, which includes IBM’s PaaS platforms, WebSphere and JBoss and other middleware, plus other integration and management tools. The Data & AI slice has all of the databases (except for Db2 for i, which is integrated in the IBM i operating system), as well as statistical analysis and data warehousing tools, Watson stuff and Weather & Media stuff that IBM got through the Weather Channel deal a few years back. Security and Health software are just what they sound like, and Transaction Processing is CICS for mainframes as well as any analytic and integration software running on the System z platform as well as storage software (as we mentioned earlier).
That leaves the Consulting pie, which will look like this after Kyndryl is spun out:
This business is a bit bigger than the Infrastructure business, but with a high single-digit growth rate, it will soon be much larger than the Infrastructure business and within five or six years should be as large as the Software business is today. In a little more than a decade, the Consulting business could rival the software business.
It seems unlikely that the Infrastructure business will keep up, unless IBM buys its way back into it. Lenovo might be a good fit. Har har har. . . . So would Inspur. Neither would pass muster with the Chinese government, presumably. And the U.S. government might have some strong opinions, too, unless IBM did some Made In America guarantees.