IBM's server safety net has HP at a loss
Not looking good, Carly
IBM's server business is booming - in a big way - and it couldn't come at a worse time for the HP camp.
The server divisions at IBM scattered around the globe deserve high praise for a solid second quarter. IBM marched back into first place as the worldwide leader in server sales, taking 30.4 percent of the market compared to HP's 27.8 percent, according to IDC. In Intel servers, IBM was the sole major vendor to gain market share, and in Unix kit IBM punished main rival Sun Microsystems. In a business marked by enormous pricing competition, IBM managed to grab everything it could and make heaps of revenue in the process.
These gains are bad news for all of IBM's competitors, but they should really scare the life out of a certain company over in Palo Alto. You know. The one that likes to Invent.
HP is telling the public a couple of things. It says that customers are abandoning midrange Unix/RISC systems in favor of Intel-based kit. The Intel servers have improved over the last couple of years, and they can get the job done on a lot of enterprise tasks. Unix is, however, still important on the high end, Fiorina says. Lucky HP's Superdome servers have been selling so well of late.
On the Intel side of the house, HP wants to beat Dell on price for low end servers and beat IBM on technology with midrange to high-end gear. This means HP has to keep volumes high, shipping as many servers as possible, while still maintaining enough focus to squeeze profits out of the higher margin low volume hardware.
For HP's customers, this strategy sends an interesting message. The Unix users are hearing that they are important, but that HP is really putting all of its focus on the high-end. Fiorina tends to talk as if midrange Unix has already gone by the wayside. So for the chosen few with millions to spare per system, everything is going to be all right. The rest of the users can deal with a decimated HP hardware team that is busy elsewhere - probably answering calls in India. That's comforting.
A large class of Intel users are then being told that HP will continue to beat Dell on price in the long haul. HP assures customers and analysts that this Compaq acquisition thing has paid off in spades. The party line seems to ignore the fact that Dell has chipped away at HP's market share since the merger closed and that Dell's server business is profitable, while HP's is not. It doesn't take an Itanium engineer (They have lots of time on their hands - Ed) to figure out that HP will have a real hard time fending off Dell on the low end. Just ask the former Compaq folks.
This leaves HP battening down the hatches and putting its blood, sweat and tears into the midrange and high-end Intel server market. It is here where HP wants to compete. It's with these precious Intel boxes that HP can show it still has some engineering mettle left. These are the systems replacing all that Unix kit and bringing in the big bucks.
On four processor and higher Xeon servers, HP actually seems to have a pretty strong case. HP and IBM can argue all day long about who scales better or who has the most intelligent servers, but we're not seeing any clear winner here. IBM's xSeries boxes tend to receive more favorable billing from analysts, but so it goes. Big Blue probably has a better "analyst outreach" program.
But just when all looks bright and cheery, we run into another problem, and it's a big one. Itanium.
For Itanium to survive as a viable product, it has to become an industry standard. That's the whole reason HP and Compaq stopped making their own chips. Remember? The market needs something everyone can agree on.
Sadly the most recent numbers from IDC have Itanic coming up well short in the industry standard category. Believe it or not, Groupe Bull out-shipped Hitachi, Unisys and NEC - combined - by selling all of 17 servers. Maybe the French know something the rest of us don't. Well, maybe 17 or so French people, that is.
It's HP's Itanium servers that have to replace a lot of those low-end and midrange Unix sales. Carly says the Unix market - Superdome excluded - is drying up fast. Someone better start shipping something that can handle 64bits then because Itanic doesn't seem to be getting the job done.
How long will the software makers stand by HP's side, waiting for the Itanium momentum to begin? Another decade? Probably not.
Does anyone get the feeling that IBM is stringing HP along with Itanium? IBM supported the original Itanium, then it held out on the second chip. It does have a pair of Itanium servers now, but the thing is that IBM doesn't talk a whole lot about them. In fact, we hear that IBM has some pretty nasty words for the chip when Intel server chief Mike Fister isn't around.
IBM does, however, talk a whole lot about its Power processor and that little x86-64bit guy called Opteron.
This gives IBM three, well two and half, 64bit options compared with HP's one. IBM will sell you an Itanium server if you really, really, really want one. But for most of its customers, and HP's, there are plenty of PowerPoint slides on the benefits of Power and/or AMD's Opteron. The HP slides, by contrast, are full of arrows pointing to various dates in the middle of this decade with one big, red hot arrow pointing at its Integrity Itanium servers.
IBM is covering all of its bases and letting HP live or die on the Itanic. That much is clear.
All told, HP's server business appears on shaky ground. Fiorina made a huge bet on market consolidation, but she might well have played her cards too early. Maybe the time wasn't right to become a Microsoft VAR. Don't you have to prove that you can beat Dell at that game first before committing your company and Compaq's future to the premise? Don't you need to make sure giving up your chip, your precious intellectual property, is a safe road to take before betting your company on it?
HP is taking huge risks with its future and those of its customers. IBM just keeps plodding along scooping up market share and nicking revenue. ®
Sponsored: Transform Your IT Infrastructure