Mad Dog 21/21: Missing Inaction
June 4, 2007 Hesh Wiener
IBM, which is neck-and-neck in the server trade with Hewlett-Packard, formerly made typewriters, disk drives, printers, and personal computers. Over the past two decades it shed many of the products it used to manufacture, by continued to grow by creating a huge services group and absorbing a number of software companies. Except for its typewriters, none of the products IBM dropped have disappeared, the way DB Cooper, Jimmy Hoffa, or Judge Crater vanished. Today, IBM still manufactures some servers and storage devices, and perhaps it doesn’t have to manufacture anything at all.
IBM acts as if its vision of the future includes its present manufacturing operations, but that’s hardly proof of anything. The company has plans for future developments in all of its product lines, and, I presume, more than one set of such plans. Like every other computer company, Big Blue has a long planning cycle. Its processors, software, and support offerings can take quite a while to develop. Yet even a company as big as IBM has limited resources. If it develops systems that don’t please its customer, it can find itself far down a path that is difficult and time-consuming to retrace. And if it cannot get satisfactory financial returns on products it makes, it has to examine every alternative. Consequently, it always has to consider Plan B and maybe Plan C, too, but probably not plans D, E, or F.
The computer market can pack surprises, so IBM can’t always make choices about product development and stick with them, but neither can it develop lots of what-if alternatives. The high quality of IBM’s business is evidence that predictions it made years ago have generally turned out to be pretty accurate. Chances are, its future generations will be on target, too. But as stockbrokers have to say in their advertisements, past performance is no guarantee of future results.
One way computer makers attempt to shape the direction the industry takes is to talk up forthcoming development that promise attractive benefits. Like most every other aspect of the business, this practice carries risk.
IBM can tell customers a little about what it’s planning, enough to keep them excited about their affiliation with Big Blue and its various products, but it can’t really share everything because competition is so fierce. If IBM told the world all about what it planned to offer two or three years from now, some of its rivals just might beat it to the market with very attractive alternatives. This is particularly true in its System x line, which adheres to industry standards that are also met by servers from HP, Dell, Sun Microsystems, and others. It’s also harder for IBM to plan in this segment than in its other areas of concentration, where it is risky to make promises, because Intel and Advanced Micro Devics are apt to zig where they have said they will zag. When that happens, every server maker’s plans will gang agley. The chipmakers’ shift of emphasis from clock speed to power efficiency is one example of a technology change that threw every server maker off balance.
Even where IBM is operating in a proprietary architectural universe that it defines and controls, as is the case with its System z mainframes and System i servers, rivals can build machines that offer competitive functionality even if they cannot deliver plug-compatibility. In IBM’s proprietary systems markets, the balance between its urge to prime future markets and its need to manage competitive risk may fall in a different place than in the industry standard server segment, but there’s enough danger in excessive disclosure to keep IBM cautious.
In the Unix market, where the System p competes with alternatives that are not fully compatible but nonetheless enjoy a much lower barrier to migration, IBM would be taking some serious risks if it revealed its plans too early. So, while IBM can promise future support for emerging standards such as connectivity protocols, it has to keep a lid on ideas that are bound to give it a competitive advantage . . . until rivals catch up. Yet these ideas are the ones that can attract customers of other vendors and also retain existing customers who might otherwise be tempted to leave IBM for another vendor.
If IBM has trouble managing expectations about future products, it also has to think about the impact of worry on the part of customers that it will decide to cease manufacturing some of its products. IBM’s customers usually don’t think that their chosen computers, or the systems’ descendants, will suddenly disappear. But that is in fact something that can happen. It’s the nature of the computer business, and it can be the IBM way, too.
One day in 1991, the Selectric was an IBM product; the next it was a Lexmark product. One day in 2002, the UltraStar disks in Big Blue’s servers and storage subsystems were IBM brand, the next they were Hitachi. One day in 2005, IBM supplied the PCs to its corporate customers, the next the PCs came from Lenovo. One day this year a lot of big laser printers came from IBM, the next they were sold by Ricoh.
In each of these cases, IBM did what it felt was in the best interest of its shareholders. Customers who are sentimental about Selectrics, IBM 3380 or SCSI disks, their old PC AT, or, against the odds, their big IBM page printers might wish things had turned out otherwise. But that’s now what happened.
IBM has been pretty good at facing the facts when it comes to operations that make money and operations that don’t. Still, only insiders, and not too many of them, really participated in the decisions that sent various divisions to the IBM archives (and the assets of those divisions into the hands of subsequent owners). Whenever IBM has announced its sale of a product line, its withdrawal from some part of the business where it had long been prominent, quite a few customers suffer trauma. Users have to review their plans. Executives who sold their management on a product strategy based on the IBM name have had to ride out disquieting circumstances and endure questions about their judgment. In isolated cases, the computer executives ultimately had to disappear, but not quite as dramatically as other have made their exits.
When Dan Cooper, more widely (if incorrectly) known as DB Cooper, hijacked a Northwest Orient airplane back in 1971, nobody guessed the outcome. Cooper let the passengers go in return for two hundred grand and a bunch of parachutes for himself and the crew of the plane, a tactic that increased the chances he would end up with a working parachute. He then let the passengers go and directed the plane to fly from Seattle towards Mexico. Along the way, Cooper jumped out. Nobody found him. Nobody found his bones, either, so he might have made it. Years later, some of the money turned up, but nobody is quite sure what happened to Cooper. In 1971, he looked like some kind of rogue hero; today nobody would have much admiration for an aircraft hijacker, even if nobody got hurt except in financial terms. Still, Cooper remains a role model for chief executives who hijacked their companies and got golden parachutes as part of their dismissals that make DB’s take look like peanuts.
Then there’s Jimmy Hoffa, whose actual middle name is Riddle, once the boss of the Teamster’s Union. He vanished in 1975. Lots of people think he was bumped off. Hoffa had plenty of enemies, and he had friends who might have been even more dangerous. It’s remotely possible that he got away from it all and lived on under a new identity, but nobody ever claimed to have discovered him. If he was murdered and buried in a secret grave or beneath some construction project, nobody actually knows that either, although from time to time the authorities dig here and there in a vain search for his bones. Lenovo, after going through a period of considerable uncertainty, now seems to be having an easier time finding a market for the former IBM PC line than the FBI has had finding Hoffa’s long lost remains . . . if, in fact, there are any remains to be found.
Another big shot who vanished was Judge Joseph Crater. He was a political powerhouse, a philanderer, and very possibly a crook, too. In 1930, he had dinner with friends (including a mistress) and then left on his own, hopping into a taxi, presumably headed for a Broadway show. He was never seen again, at least not by anyone who talked about it. Crater, born in 1889, won’t ever turn up alive, and at this point it seems he won’t turn up dead, either. His fate and that of the Selectric have a lot in common, except that everyone loved the Selectric, although apparently not enough to keep buying them. The heir to the Selectric’s ubiquity is Microsoft Word, about which the range of opinion more closely resembles the mixed, at best, way contemporaries felt about Judge Crater. The heir to the former IBM division that offered typewriters and low-end printers, Lexmark, seems to be doing pretty well, but no company dominates the small printer market the way IBM did the quality end of the typewriter trade.
What would happen if IBM stopped making its four lines of servers and the collection of storage subsystems? (IBM’s System x servers are already made by contract manufacturer Sanmina-SCI, and have been for a few years.) Would the products continue to play a major role in computing, the way Hitachi disk drives, Lenovo laptops and PCs, and Lexmark’s line of office machines do? They might. Microsoft doesn’t make computers and the major Linux distro companies don’t even write the software they sell, but that doesn’t keep the hardware running their operating environments from selling to the tune of billions of dollars a year. The software that defines IBM’s storage systems far more than does the disk and controller hardware could remain popular and respected if IBM got out of the disk array game business.
It’s even possible that IBM would be better off not merely getting out of mainframes, System i, and System p servers while opening up the market in platforms that could support its proprietary software. Intel and AMD build reference platforms and more, but that doesn’t seem to get in the way of companies that want to develop and market machines that run Windows, Linux, FreeBSD, and other X64-related environments. It is conceivable that IBM could offer its Power chips and glue circuits to anyone who wanted to build AIX machines (or machines that could run AIX, Linux, and any other operating system that was ported to Power) and find that it had a very good business indeed.
Would customers run away from IBM systems, by which we mean systems defined and supported by IBM, if they didn’t include IBM hardware? It’s hard to say. Hewlett-Packard seems to have survived the transition to Intel and AMD server chips pretty well. Sun defines server chip architecture, farms out chip fabrication, shares the market for architecture it has developed and still retains the loyalty and respect of a very large customer base. Dell doesn’t seem to worry about not making the processor chips, storage subsystems, printers, and other products it sells in volumes that give it an eleven-digit business.
When IBM farmed out the manufacturing of the z800 mainframes to Hitachi, it didn’t suffer any consequential loss of market share. If anything, the z800 has proved to have the staying power of any mainframe built by IBM; it remains in wide use despite IBM’s push to upgrade users to z890 and now z9 processors. The machines Hitachi built for IBM met Big Blue’s stringent engineering standards and enjoy a reputation for reliability that is indistinguishable that of from the rock solid mainframes that came off IBM’s lines in Poughkeepsie.
What might cause IBM to rethink its involvement in hardware manufacturing? One thing that comes to mind is a change in the fickle electronic games business, without which IBM’s chip fabrication facilities would have to operate under a whole different and far more difficult set of economic rules. If Sony and Microsoft said “Game Over” and migrated away from the respective Cell and Xenon Power chips in the game consoles to chips from a different fab–most likely Intel or AMD–then Big Blue might have to say the very same thing to its own hardware divisions.