Crazy Idea #483: A Leveraged Buyout Of IBM i
April 6, 2015 Timothy Prickett Morgan
Sometimes the crazy ideas come from us, sometimes they come from you. Sometimes we are all pondering the possibilities at the same time and it seems like an idea is almost a meme, spreading like a virus, hopping from brain to brain, or as I sometimes suspect might be possible, vibrating in the cosmic ether for others to reflect on if they can only hear the image. Whatever scenario this one is, a bunch of you have reached out and asked me to write about the possibilities of a leveraged buyout of the IBM i business. And so, here we go.
I have taken up that idea more than a few times in the two and a half decades that I have been watching the market for the AS/400 and its progeny. Back in the early 1990s, when IBM was having a lot of its financial troubles because it did not understand that the nature of the data center was changing and that RISC/Unix and then X86 machines running Windows and then Linux would radically change the way most computing would be done, more than a few of us talked about spinning out the AS/400 business separate from mainframes and the rest of Big Blue. In fact, quite a number of people said that the way to fix IBM back then, when it was sustaining heavy losses and firing enormous numbers of people, was to make a bunch of Baby Blues, with the AS/400 and mainframes being but two of them. There was also a separate X86 Baby Blue focused on PCs and servers in most scenarios, and spinoffs focused on software, services, and storage, respectively, as well. The idea was to free these IBM business units from each other and let them compete wholly and openly with each other where they should and to cooperate where they could.
This never came to pass, of course. IBM brought in Louis Gerstner from American Express and RJS Nabisco fame and he concluded that what Big Blue needed to do was focus on services and keep itself together long enough to reap the benefits of the enormous bloodletting of jobs that had taken place before he came on board on April 1, 1993, and in the months after he took the helm. As I pointed out two months ago, however deep IBM’s layoffs are these days, they are nothing compared to the swinging and singing axes that were whirling around back then. IBM has divested itself of its X86 server business, so it doesn’t have to worry about internal competition from X86 iron anymore, and it has similarly sold off its IBM Microelectronics chip making business, which was the real money loser in Systems and Technology Group. I think the former System x division takes a bad rap for not being profitable, and what really happened is that IBM needed to sell something to balance out the price it actually paid GlobalFoundries to take the IBM Microelectronics division off its hands. IBM chairman and CEO Ginny Rometty and the captains of Big Blue who put this strategy together no doubt hoped to make more money than they did from Lenovo Group–the word on the street was $5 billion to $6 billion when rumors of the deal first circulated back in early 2014, which is a lot less than the $2.3 billion that IBM got–and Big Blue no doubt didn’t think it would have to actually pay GlobalFoundries $1.5 billion to take its chip fab business over.
Them’s the breaks. Especially when the economics of the chip business are what they are. The minute that ARM ascended in the embedded space against PowerPC, the minute that the PowerPC partners could not keep Apple as a client not only for PCs, but for future products like iPhones and iPads, the minute that IBM lost the game console business from Sony, Microsoft, and Nintendo, and the minute that the Unix market took a severe dive, it was inevitable that IBM would look to start selling off less strategic units.
The irony is that IBM is more committed than ever to its Power architecture, and through the OpenPower Foundation it is revitalizing the Power platform, only this time not for Apple devices, but for datacenter uses, particularly for hyperscale datacenter and supercomputer applications. It is a kind of Power redux, since the Power architecture’s success was largely funded by the supercomputer centers of the world that ran big federated AIX machines and then, later, massively parallel BlueGene Linux machines. Now, the mantra as readers of The Four Hundred well know is to open up the Power chip and compete better against X86-Linux machines in HPC and hyperscale, while making just enough investments in the Power architecture to keep IBM i and AIX current and useful.
With the selloff of Microelectronics and all of the noise coming from the OpenPower Summit about third party machines, it is natural enough to ask why IBM wouldn’t just spin off the IBM i and/or AIX businesses and be done with them. The answer has to be–and we will see if soon enough if IBM provides financial results for its Power Systems division–that even though this Power business is shrinking, it is profitable. Especially when services and software as well as hardware are taken into account. It is hard to say for sure how much revenue is actually being pushed by IBM i shops. Back in 1993, when IBM was trying to get itself off the rocks, I took a look at the AS/400 ecosystem, and it was a whopping $14 billion worth of business to Big Blue, and it broke down like this:
That is a very old style chart, and it is hard to read, I realize. We went a little retro there, and if you want to read the reprint of that original story, you can do so at this link. You can see my update on the ecosystem in 2005, which was, yes, a decade ago, at this link. My guess a decade ago was that the OS/400 ecosystem was about $4.5 billion of business for IBM and another $2 billion or so for the 3,000 or so ISVs that peddle applications on top of it.
If I had to guess, the ravages of Moore’s Law and virtualization have radically compressed that IBM i ecosystem. Sure, there are around 130,000 or so customers worldwide and they do a lot of work on their systems, but transaction processing and web serving and graphical front ends are not driving growth in the same way that analytics and other workloads are. As was pointed out somewhat humorously in our April Fool’s story last week, IBM i shops don’t need more performance as much as they need lower-cost hardware and software on which they can do more work. IBM wants customers to spend more money, not less, and they want to spend less money, not more. And so there is a kind of dÃ©tente across a demilitarized zone where customers only spend when they have to.
If I were going to argue for a spinout of the IBM i business, it would be because I thought they would have a chance to radically change that IBM i business before it is too late. Getting those IBM has around 30,000 shops that are engaged, but another 100,000 or so that are not. They are on old systems and old systems software, just getting by and doing as little as possible on the machines. This is not a customer base that is going to spend a lot of money. If anything, I would argue that IBM should port IBM i to some of these third-party hyperscale machines it is showing off, slash the prices and make them subscription-based like cloud software, and get those 100,000 customers up to modern iron. I don’t care if that is System/36 or System/38 or RPG IV applications running in a very inefficient emulation mode, because to put it bluntly, those Power8 machines have plenty of excess capacity.
Now, if that is the plan for an IBM i buyout–getting customers current and getting them all running in some way, shape, or form on modern hardware and software and positioning them for the future, I can get behind that. But as I have said before, I think the answer is to build a massive cloud and just have customers pay monthly to use the new iron and new software. A few strategically placed datacenters around the globe, an aggressive price to attract those customers would do the trick. How about free for the first year? You can bet those vintage AS/400 and iSeries shops would come out of the woodwork for that.
Aside from an aggressive business plan like this, I think a spinout of the IBM i business would actually be weaker at this point than leaving it inside of IBM. The IBM i software stack cannot possibly cost Big Blue much money to maintain, even if it cannot possibly generate as much money as it did a decade ago. If the IBM portion of the ecosystem was $14 billion in 1993 and $4.5 billion in 2005, I shudder to think of what it is in 2005. Is it maybe as low as $1 billion across all categories of IBM revenues? With maybe another $1 billion in ISV software, all told?
The time to break the IBM i business free was back before 1993, when there was a chance to radically change the trajectory for the PowerPC hardware and OS/400 software. IBM could open source IBM i, but the techies in the IBM i community are not low-level programmers like those in the Linux, so they are not going to take the bull by the horns even if IBM i was opened up.
Having said all of that, I would love to be a fly on the wall in Martin Schroeter’s office so I could know, for sure, how much revenue and profits the IBM i business makes for Big Blue. The fact that IBM has not spun it off tells you that it must be enough that it wants to keep it. With the Power business being a dominant platform for IBM, and the company wanting those profits to increase, IBM i perhaps matters to IBM more than it has in many years. The question becomes does IBM want to grow this business or just manage its decline? If the answer is the latter, that obviously doesn’t make us all happy. I don’t know how hard you can push to increase the IBM i business without some pretty big changes in that business, ones that will not result in more profits even if it does drive more revenues. And that is the rub, and perhaps the best argument for an IBM i spinoff. A new company can and would live on lower margins, but would perhaps be focused on getting all 130,000 customers active and happy and modern, and that might double or triple the ecosystem.
Tell us your thoughts about this. I am curious what you think.