What Is IBM Going To Do With Its Systems Business?
July 15, 2013 Timothy Prickett Morgan
I don’t pull the intuition card very often, but in this case, I can feel it in my skinny little bones. Something is up at IBM, and I think we can expect to see some big changes in its systems business in the coming months. This is not just a hunch, but the reading of the executive tea leaves combined with a hunch and observations of IBM for over three decades now. You get a feel for when Big Blue is ripe for change if you have been doing this long enough, and we have a new CEO who has to hit her numbers and a rapidly changing IT market.
It doesn’t get any riper than this.
So over the next few weeks, I am going to think this through and try to help us all figure out what IBM might do–and will not do–with its systems business, which has come under pressure in recent years from direct competitors and indirect ones peddling new ways of computing. This is a thought experiment done in the absence of data, and you should think of it that way. As I have said before, you have to be careful of making long-term predictions because the projections tend to follow straight lines, not the curvy, side-to-side and up-and-down stuff that comes from real living.
First, let’s talk about the executive tea leaves within Systems and Technology Group and IBM overall in the past several years at IBM. I think the day that Bill Zeitler retired from IBM in May 2008 as the general manager of STG, all kinds of things were put into play. Zeitler put in 39 years at IBM, so he did not leave under a cloud. Quite to the contrary.
Zeitler had a good sense of systems and customers, and more importantly, saw a need to move Power processors into adjacent markets such as game consoles and embedded devices of all kinds. Zeitler was talking publicly about this custom systems business back in early 2007, which means the plan was probably cooked up in 2005. (The charts and arts from his presentation also used 2005 as a baseline, which is strong circumstantial evidence.) To put it bluntly, IBM did not follow up on this vision particularly well, even after scoring PowerPC design wins in the game consoles of Microsoft, Nintendo, and Sony as well as consumer electronics from Toshiba years earlier. Since then, IBM has basically stopped development on the PowerXCell processors and it has lost two out of three in the next round of game console design wins. If you believe the rumors, the Microsoft Xbox One has an eight-core “Jaguar” chip and matching graphics processor from Advanced Micro Devices, and the Sony PS4 will have a similar CPU and a more powerful GPU, also from AMD. The Nintendo Wii U console has a three-core PowerPC 750 chip called “Espresso” and an AMD GPU, and neither are particular speed demons.
If IBM wants to keep the chip fabrication plant in East Fishkill, New York, from becoming a money pit, it has to make lots and lots of chips. And I don’t think that the rate of embedded PowerPC chips sold into the market (which is seeing incursions by Intel Atom and various ARM processors) plus the System z and Power chips used in IBM’s own systems are enough to justify the investment IBM has to make to go beyond 22 nanometer technology. IBM has already committed to 22 nanometer processes to etch the future Power8 chips, but the investment in chip making equipment has to be spread over all kinds of chips over many years for it to pay off. Every iteration that shrinks chip process technology costs increasingly more to bring to market, and customers are used to getting a lot more bang for the buck, too, with each new chip. So, there is a pincher effect on profits that is unavoidable unless you can expand into new markets and pump up the volumes. I worry about process nodes out beyond 22 nanometers when it comes to IBM’s own fabs.
I think another jarring development for STG was the arrest and jailing of Zeitler’s replacement and one of the heirs apparent of Big Blue when Sam Palmisano would step down. Bob Moffat, who took control of STG in May 2008, joined IBM in 1978 and managed its parts supply chain, something that he did quite well in fact. Moffat ran IBM’s PC and high-end printing businesses, which were sold off to Lenovo and Ricoh, respectively. And if Moffat had not been stung in an insider trading scam in October 2009, he would have absolutely been a contender to follow Palmisano as IBM’s CEO, and he would have been young enough to do the job longer than current CEO Ginni Rometty. It is hard to say what else Moffat would have done.
But in the end, IBM converged Software Group and Systems and Technology Group under Steve Mills, who is not young enough to be IBM’s CEO and probably doesn’t want the job given the fact that he controls so much of Big Blue anyway.
In the wake of Rometty’s appointment as president and CEO nearly two years ago and before she became chairman a year later when Palmisano left the building in Armonk, she tapped Tom Rosamilia, a Software Group hot-shot who started out as an MVS programmer when he joined IBM in 1983 and who ran the database and WebSphere businesses in Software Group and then jumped over to STG to run the System z and Power Systems businesses for a while, to be vice president of corporate strategy and general manager of enterprise initiatives. His bio at the time said that he would be “responsible for the strategic direction of IBM’s future, as well as developing the path that IBM will take as we move forward in this new era of computing.”
Rosamilia spent a year pondering IBM’s future in computing, and back in May, after a particularly lumpy first quarter when Power Systems sales swooned and the System z business didn’t meet its numbers, Rometty moved Rosamilia back to Systems and Technology Group to run the whole shebang. Rod Adkins, who had been running STG, switched into the strategy role. It is not clear just how much reporting Rosamilia does into Mills as he runs STG, but they have a long relationship and presumably a good one. The point is this: Rosamilia spent a year thinking about IBM’s future, and whatever plan Big Blue has, he has probably hatched it by now.
The launch of the PureSystems modular servers to compete against decent-selling Unified Computing System blade and rack servers from Cisco Systems had already been hatched long before Rosamilia was given the Big Think job. But Rosamilia was in that job when rumors surfaced that IBM was working a deal with Lenovo for the Chinese PC maker to buy its System x and BladeCenter businesses for somewhere between $5 billion and $6 billion. The deal has been on-again, off-again for months, and the latest rumors I hear is that the two are still talking and they are haggling about price, employee transfers, and product lines.
Then there is Andy Monshaw, who left IBM a few weeks ago according to his LinkedIn profile. IBM’s press relations people in the PureSystems business confirmed that Monshaw has indeed left Big Blue. Monshaw was in charge of the PureSystems business and a replacement has not been chosen yet. He was a 28-year veteran who ran IBM’s storage business for many years. Monshaw could have left for a lot of reasons: he wants to run his own company as CEO, or he just wants to do something different having seen the PureSystems get launched. We don’t know. But his departure sets off my Spidey sense bigtime, for whatever that is worth, particularly with rumors going around about IBM selling off parts of its server business.
Maybe PureSystems got lumped into the deal, too? Maybe Lenovo is going to make machines and IBM is going to be a Lenovo reseller for its top accounts and just give the rest of its x86 server channel to Lenovo for a few billion bucks. We don’t know. But I am starting to get suspicious.
With Power Systems revenues dropping in synch with the Unix market overall, I am also made a little nervous. I would not be at all surprised if IBM sold off its chip fab in East Fishkill and outsourced Power and System z chip manufacturing to the buyer. I would not be surprised if IBM outsourced a lot of the manufacturing of all of its systems. Then again, another possibility is to do these things and then buy Advanced Micro Devices, thereby gaining a microserver business and an X86 and ARM processor business to boot. And it can tell Intel to go to hell and start competing against it.
There are a lot of possibilities, and it will be fun to explore them, each in their turn.