Some Thoughts on the 22 Percent iSeries Sales Decline in Q1

As I reported elsewhere in this issue, IBM announced its financial results for the first quarter of 2006 last week, and in those results, Big Blue said that the System i, formerly iSeries, server business was off 22 percent.

This, obviously, is not great news, but IBM, like other server vendors, blamed this decline, as well as the 18 percent decline in the fourth quarter of 2005, on product transitions. In the fourth quarter, IBM explained, customers were expecting revamped machines in the first quarter of 2006, and at the end of January they came out. The expectation slowed sales last fall, and the new machines, which offer price/performance improvements, stalled sales in the first quarter.

While product transitions are always troublesome for any server maker, they are exacerbated when it comes to specific, almost niche, machines like the iSeries and System i boxes. As we all know, the OS/400 platform has been extended such that it can support infrastructure workloads running on Windows, Linux, and AIX partitions or co-processors, but these infrastructure workloads, which make up a large percentage of the world’s server shipments, do not drive an iSeries or System i sale. The need to add performance or upgrade software behind back-end business applications does, and the much-touted infrastructure consolidation benefits of the System i line are secondary, even if they are important.

Moreover, IBM’s iSeries and System i sales are the victim of two opposing forces–forces that hit the broader server market, too, but which are masked by the server sprawl that is still the rule rather than the exception in the data centers of the world. When IBM gradually shifted most of its sales to the reseller channel over the past decade, that, of course, caused its revenue to decline. And now, it seems to be seeing a decline that is related as much to model mix as to a decrease in box and upgrade shipments. In 2005, over 96 percent of the iSeries boxes IBM shipped were i5 520 machines–those with one or two processors. And many of these were deployed with geared-down processors at that. Basically, OS/400 workloads are not growing fast enough to pump up sales, and every couple of years, as companies deploy new OS/400 servers, they seem to be downgrading to smaller, less costly boxes. This is good for the customer, but it is problematic for IBM.

While the mainframe base has been buoyed by deploying Linux and Java on mainframe engines that cost one-quarter the price of a real mainframe engine, the real reason the mainframe is doing relatively well is because the big banking, insurance, manufacturing, and distribution companies that have always depended on mainframes have seen their businesses improve in the past several years. They are churning through more mainframe transactions, so they need more mainframe MIPS. I am not convinced that the small manufacturers, distributors, and financial services companies that rely on OS/400 platforms have seen their fortunes rise so much. The Fortune 1000 is doing much better in this economic recovery than the other million or so mid-sized businesses of the world. That’s a hunch, not a statistic. But I think this is part of the reason that the iSeries and System i business has been hit hard by product transitions, harder than in years gone by. A few years back, the exact same thing happened in the Unix midrange businesses of Sun Microsystems and Hewlett-Packard, and IBM swooped in with 45 to 50 percent discounts and just pounded its Unix rivals to a pulp–and became the market leader because of it.

The problem that IBM has with the iSeries and System i business, as far as I can tell, is that while there may be in excess of 200,000 customers worldwide using the OS/400 platform, not enough of them are in desperate need to buy new capacity–or IBM’s pricing for Power servers and i5/OS software are so prohibitively expensive that if they do need to add applications, they do it on Wintel platforms. As I have said many times before, IBM must see that the only way for the iSeries to produce consistent results, quarter to quarter, is to become a volume solution for SMB customers. And to do that, IBM is going to have to make the System i so attractive for those 200,000 customers that they just keep coming back for more and more capacity, just like the top mainframe shops keep doing.

Let’s try an analogy that maybe we can all understand, maybe even IBM. You can try to sell me a very expensive bottle of champagne every once in a while, and I will go for it on special occasions. The wife and I like fine French champagne–but not too often, that’s for sure. Or, you can try to sell me vast quantities of cheap American beer. I don’t want to drink that cheap beer, of course. Who does? But when I was in college and, after that, when I was a young writer living in New York City, cheap beer was my only budgetary option. So Bud and Coors were my drink of choice. Now, as a grown man, I prefer a fine microbrew, something that is a little more expensive–I said a little, but so little I don’t really notice–and it has a lot more character, hops, and kick to it. IBM has got to stop selling the System i like it is champagne and start realizing that it is actually a microbrew beer–and price it accordingly. Because, I gotta tell you, many people are getting sick on the cheap Wintel beer. The hangovers are terrible, and it tastes like dishwater. But, they can’t afford to drink champagne, either. So, being human, they drink the bad beer. And, iSeries sales drop 22 percent and Bill Gates laughs all the way to the bank.