The Hundred Thousand Plus on the Four Hundred
October 18, 2010 Timothy Prickett Morgan
How big and healthy is the base of customers for the IBM i operating system and its related Power Systems platform? Those are two questions that are very difficult to get answers to except in very special moments when some IBMer somewhere makes a slip of the tongue–or intentionally lets the data out to try to comfort the multitudes in AS/400 Land. Depending on how you want to look at it, the latest numbers tossed around by IBM can be chilling or comforting.
First, let’s get the number out there for the size of the AS/400-iSeries-System i-Power Systems running IBM i installed base: easily 100,000 enterprises. At least. That number comes compliments of Steve Will, the chief IBM i architect at Big Blue. Here’s the direct quote, which came from a blog post where Will writes over at IBM Systems Magazine.
“I can tell you, factually, that IBM i has more customers than any other IBM platform,” Will writes, adding “that we still have easily over 100,000 enterprises of one size or another actively using IBM i and its immediate predecessors, that we’re still sold and have active customers in more than 115 countries in the world.”
Personally, I find that number–easily 100,000 enterprises–a bit of a shock. But that is because I have a very, very long memory. But let’s take a walk down short-term memory lane first, and I will remind you that back in May, when the AS/400 Large User Group–you know, the biggest hotshots in IBM i land–put up a presentation by Will given to the LUG about the future of the IBM i platform. And in that presentation, it said “100,000’s Customers, 100,000’s Systems” when referring to the IBM i and predecessor installed bases. So the plural was perhaps a typo on the customer count?
No matter, I had private discussions with people within IBM who should be in the know, and they believe that the installed base of unique customers is somewhere just north of 100,000 companies these days, and other ISVs I have talked to recently have said the same thing.
As best as I can figure, the AS/400 went from zero to 100,000 customers in the first two and a half years of its life, exiting 1990 breaking that barrier. The introduction of the D Series, the first dual-processor AS/400, and the E Series, the first three-processor machine, in 1991 and 1992 helped the business stabilize at around $4.8 billion in system and storage revenue and the customer count kept growing, despite the first Bush Recession. The AS/400 box peaked, in revenue terms, around this time thanks in part to a shift from mainframes to midrange boxes in the early 1990s.
The competition from Unix servers in the mid-1990s ate into revenues, but the customer base kept growing, and according to Bill Zeitler, who was AS/400 general manager at the time of the AS/400e launch in Rochester in 1998, the base peaked at around 275,000 customers and the revenue stream from hardware and systems software kissed $4.7 billion after drifting downward a few years early. That bubble was thanks in large part to the ERP and Y2K booms, which hit IBM shops earlier than the rest of the IT market, I think.
That number fell to 250,000 in 2002, and by 2005, IBM was talking about somewhere on the order of 220,000 unique customers in the base. After that, IBM stopped talking about it. At a bleed rate of 25,000 customers per year over the past five years, plus 2,500 new customers to the base, you would end up somewhere north of 100,000 customers.
That’s a big change from IBM, which still insists that the IBM i platform is still its biggest customer base, as Will said in the above quote. In 2002, my best estimates for the size of IBM’s four platforms in terms of unique users were: iSeries, 250,000 customers; pSeries, 40,000; zSeries, 20,000, and xSeries, somewhere between 200,000 and 400,000 customers. To make the numbers work out to 500,000 unique customers, which I knew was the total IBM enterprise customer base, then xSeries had to be the smaller number. And of course, across those four different groups, there was a lot of overlap. Probably half of IBM’s iSeries shops had xSeries servers, and a smidgen had pSeries.
Today, there are probably something on the order of 200,000 Power Systems shops, with a slim majority by number being from the IBM i side of the Power Systems house, but with the AIX and Linux side accounting for most of the midrange and nearly all of the high-end systems by shipment count. AIX accounts for the bulk of revenues, I think. I have a hard time believing that the IBM i hardware and systems software business is much above $1.5 billion a year, but that is just a wild guess on my part since IBM does not break out i-versus-p sales any more. It may only be $2 billion a year including maintenance and other tools and software.
The best numbers out there for mainframes put the installed base at 6,000 to 7,000 unique customers, and I have no idea whatsoever about how many System x and BladeCenter customers there are. But for what Will says to be true, a massive customer exit from IBM X86 and X64 servers would have had to been happening over the past decade. IBM could be selling more gear to fewer customers, as it did with System z mainframes until the z10 machines petered out last year.
The message here is not to freak out, but to take a look at the numbers and think about why this has happened and what it might mean. There has been a massive amount of consolidation in the global economy, and every time a mainframe or an AS/400 (meaning any i-related box) comes into a merged company, these venerable machines are on the defensive and Unix, Windows, and Linux platforms are on the offensive. We’ve had a decade and a half of Java and Windows and two decades of Unix grinding against the AS/400 and its progeny, and IBM has been the big beneficiary in Unix in the past decade. You, AS/400 shops, paid the marketing bill for that with the interactive software tax, artificially high hardware prices, and other shenanigans that have been documented in this newsletter for decades. IBM should send you a thank you note and maybe a couple of shares of stock. Or cut you a break on your next upgrade.
The other thing about that 100,000-plus number to consider is that IBM probably doesn’t really know how many customers are using the boxes, any more than it knew how many System/36s were out there in the field in the mid-1990s when the CISC-to-RISC AS/400 transition was under way and IBM was doing the last of its many pushes to try to get System/36 customers onto AS/400 machines. One of the funniest phone calls I ever had was Big Blue calling me up to ask how many System/36 customers it had, to which I could only think: “Good heavens, you don’t know?”
The comforting thing is, no business really knows who its customers are, which is why everyone is so CRM happy these days. They really don’t know. In any business, the 80-20 rule–well–rules, and before too long, the 80 percent of the customers who are not doing much fall by the wayside and are lost. Or misplaced. Maybe they don’t go anywhere, but a company no longer can say for certain that they are still using their product. Ford doesn’t know my 1995 Taurus station wagon died in June and I junked it. Hewlett-Packard still thinks I have ProLiant servers, and I don’t, even though the maintenance contract is still live. But a friend of mine has the machines, and they are being used. HP doesn’t know that, either.
Here’s another thought: How many customers did Sun Microsystems have? How many customers made it worthwhile for Oracle to spend $7.6 billion ($5.4 billion net of Sun’s cash hoard) to take control of the company? How does 35,000 Solaris platform shops grab you? That’s the number, as you can see here. I realize that the main reason Oracle bought Sun was to get control of Java, which I discuss in this issue elsewhere, but at least some of that dough was for the Solaris base. By the way, other Oraclers say the number is 50,000 Solaris shops, which just goes to show you that Oracle and Sun don’t know for sure, either.
The important question for you and me and IBM is this: If there are only 100,000-plus IBM i shops in the world, are a lot of them active and vibrant? If you take 20 percent of 275,000 customers, you get 55,000 vibrant shops circa 1998. Maybe what IBM has shed are a lot of the shops who were not really investing in the OS/400 and i platform anyway. A lot of the crunch comes from the consolidation of financial services, insurance, telecommunications, manufacturing, and distribution firms. When an i shop merges with an i shop, you still only have one i, like a Cyclops.
Like you, I wish the number of IBM i customers were larger. I also wish I had my own wings so I could fly everywhere, too. I’ll take a sailboat and a little red wagon and a pony, too. In the end, I happen to think even 100,000 customers is a big number. And we plan to keep on serving their needs each and every week to the best of our ability.