Gartner Revises HP’s Server Sales Downward for Q1
July 14, 2008 Timothy Prickett Morgan
Back in May, the analysts at Gartner put out statistics for the first quarter of 2008 for sales and shipments of servers of all kinds on a global basis. And there was much jubilation over at Hewlett-Packard, which Gartner said knocked out IBM as the top server seller globally in the quarter; HP has long-since been the top shipper. But beating Big Blue on its home systems turf is a big deal.
Pity then–and particularly for those HP executives who have bonuses pegged to market share gains–that it apparently did not happen. And hooray, I suppose, for those IBMers who can put their boxes back on their desks and get back to work, since Gartner now says that IBM was indeed the revenue leader for server sales in the first quarter of 2008.
“Based on updated guidance and additional analysis of the data, we have revised the revenue statistics for the first quarter of 2008,” explained Jeffrey Hewitt, research vice president at Gartner who is in charge of the box counting at the company. “We commented in the previous release that HP and IBM continue to vie for market leadership in the worldwide server market based on revenue. This is still the case, but the updated data shows that IBM maintained the No. 1 vendor position in worldwide server revenue in the first quarter of 2008.”
Back in May, Gartner said that HP sold some 683,433 servers and generated $4.01 billion in sales, just edging out IBM’s $3.912 billion in sales, which was pushed by 302,057 unit shipments in the quarter. But after the revision, Gartner now says that HP only generated $3.773 billion in server sales. HP’s shipment numbers remained the same, which suggests that Gartner’s spreadsheet calculating the average selling price of servers had an error in it. Anyway, best that Gartner can figure, IBM got 29.4 percent of the $13.32 billion server pie compared to HP’s 28.3 percent share. The change in HP’s revenue figures by Gartner also knocked down the growth of the overall market from 4.3 percent aggregate revenue growth for servers to 2.5 percent growth, a decline that was solely the result of HP’s growth from Q1 2007 being reduced from 10.3 percent to 3.8 percent. Incidentally, HP did outgrow IBM’s sales increase of 2.1 percent quarter-to-quarter, even after the numbers were rejiggered.
As best I can figure, no other numbers changed in the report from May, which we reported on here, but the relative shares of Windows, Linux, Unix, and other platforms may have shifted a little because of the changes in the number. As you can imagine, $237.6 million is a lot of server revenue for any player, even ones as big as HP and IBM. The word on the street is that average selling prices of HP’s X64-based ProLiant server line were made too large, and that was the error that Gartner is correcting. In the May numbers, Gartner said that HP sold $2.665 billion in X86 and X64 servers, generating 13.5 percent revenue growth compared to the first quarter of 2007, when it sold $2.347 billion in such iron. If all that extracted revenue was pulled out of X86 and X64 boxes, then HP would have sold $2.427 billion in gear and would have only grown sales by 3.4 percent for this type of machinery. That would also kick down the X86/X64 server category to $7.32 billion in sales and overall growth across all vendors in this category to 3.6 percent.
The IBM-HP fight continues. This is far from over.
Maybe HP should buy Sun Microsystems? With a market capitalization of $8.3 billion and Sun’s stock pretty much diving in a straight line from just under $25 a share last October to just over $10 a share now, there is never going to be a better time to buy Sun. Even if HP didn’t pay any premium at all because Sun has little hope of growing itself fast enough to become Wall Street’s darling, that’s still a lot of money. But Solaris 10 can be easily ported to Itanium-based Superdomes (Sun did a Solaris 8 port to Itanium a long time ago, and no one knows Itanium better than HP, probably not even Intel at least as far as operating system porting and tuning goes) and Solaris 10 already runs on ProLiants. All that neat engineering at Sun would find a good home at HP–skinny server designs, big switches, open source Java, middleware, and other goodies. The best argument for such a deal–and I am mostly kidding about this, folks–is that Sun pushed $1.318 billion in server sales in the first quarter by Gartner’s reckoning. That would put HP at just over $5 billion in server sales for the quarter. And if memory serves me, that is about what you would have expected HP to do in 2001 when the dot-com boom was roaring and the company bought Compaq. (It didn’t work out that way, thanks to the dot-com bust, the 9/11 attacks, and a worldwide recession that hit IT extra hard.)
Then again, if IBM wants to keep some water between itself and HP, maybe it should buy Sun. All the same arguments apply. With Solaris being ported to mainframes and Power machines already–well, somewhat tepidly and maybe just for the sake of some press releases and a few intrepid customers–IBM could easily absorb Sun. And gain control of Java and the intellectual property behind Unix. Remember, Sun has all of those juicy rights because it was AT&T’s partner in creating Unix System V. Then IBM might be able to open source AIX and/or possibly merge it with Solaris. You could put Solaris runtimes inside AIX and AIX runtimes inside Solaris and let the customers choose the Sparc or Power iron that they want, and let the market decide which one lives.
It’s the summer, the silly season. All kinds of ideas bubble up from the brain stem. Must be the heat.