Is the iSeries Next, After PCs?

by Timothy Prickett Morgan

Like many of you, it was with some shock that I learned that IBM would actually consider selling off its PC business. The PC business is coterminous with my adult life, and it just seemed impossible, even after all of the company's woes in this very tough business, that IBM would throw in the towel and no longer try to compete. As I ponder the future of the iSeries in this special edition of our Four Hundred newsletters, I can't help wondering if the iSeries is next.

I am not certain whether selling off the iSeries would be a good thing or a bad thing for the business and the 225,000 or so companies worldwide that use the box for their corporate computing. But if IBM is willing to let go of its ThinkPad business, which has the highest brand awareness of any product, and is happy to sell off a barely profitable $11 billion business for $1.25 billion, anything is possible.

More than a decade ago, IBM had the highest market share of any of the clone PC makers. With the advent of the direct PC business, with Gateway in the early 1990s, and the drop in component prices and the PCs themselves that turned the PC from a tool for businesses desktops and for rich people to play with at home to a device that every consumer wanted and could afford. The PC racket, like the server business, is a volume game. And once you start backing out of the volume market--as IBM did when it exited the retail PC business in 2001--the situation gets harder, not easier. Once IBM backed out of the retail PC business, the volumes it had in its PC factories in Raleigh, North Carolina, and Greenock, Scotland, were not going to be high enough for it to make money, even on the supposedly more profitable PC business. That's why, in early 2002, IBM outsourced manufacturing of PCs to Sanmina-SCI. That was pretty much IBM's last cost-cutting maneuver. Last year, Sanmina-SCI got a similar contract from IBM to build its xSeries servers (except for the high-end "Summit" xSeries 440 and 450 servers and the BladeCenter blade servers). This is also pretty much IBM's last move to cut costs in the X86 server business. Luckily, IBM's high-end xSeries and blade servers are profitable enough to cover the costs of being in the entry X86 business, but it is not inconceivable that IBM will someday, when it can't make anymore money, ditch the X86 server business. Remember, International Business Machines is less concerned with machines these days and is more concerned with business: business process outsourcing is the biggie.

All of the competitive pressures that IBM could end up facing with the xSeries, it has been facing with the iSeries for quite some time. The difference is the OS/400 operating system and its integrated relational database management system, DB2/400. This system software accounts for the vast majority of profits that even a substantially diminished iSeries business generates. In 2003, IBM probably sold something on the order of $1.8 billion in hardware and operating systems, and assuming that about half of that is booked as software, the basic iSeries business has a gross profit of about $900 million. (These are very rough numbers.) However, some of the costs of the iSeries, such as processor development and manufacturing, server engineering, and so forth, are shared with the pSeries. In 2004, development costs went up and sales were off around 25 percent (depending on how the fourth quarter goes) to around $1.4 billion. So gross profits might have dropped to between $400 million and $500 million. After allocating sales, general, administration, and taxes to this gross profit, the iSeries business is probably a little less profitable than IBM as a whole, which had a gross profit of 37 percent in the first three months of 2004. But these numbers are estimates--and wild estimates at that. I supply them only as a thought experiment.

The question you have to ask yourself is, how much lower do Power-based server margins have to go before IBM decides to outsource production of the Power servers, both the eServer i5 and p5, to a company like Sanmina-SCI? And after that, how intense does the competition have to be before IBM decides to be the Power chip supplier and the AIX and OS/400 operating system supplier on Power platforms and leave the manufacturing to others?

The idea sounds pretty far-fetched, I know. But a decade ago, the idea that IBM would exit the PC business would have sounded a little whacky, too. It's something to think about, and in the coming months I will be pondering exactly what an independent iSeries business might look like. Don't be shy. Send me your thoughts as well.