Think of it as a partial vindication for IBM chairman Louis Gerstner. When Gerstner took the helm five years ago, he absolutely refused to split IBM into independent companies. It was a move that many industry pundits and analysts said was necessary for IBM shareholders to get full value for their investments and for individual IBM business units to finally be free from each other and therefore better able to compete against the likes of Sun, Compaq and Hewlett-Packard. For the last few quarters, IBM has been struggling to meet its hardware sales targets. The S/390 mainframes are in almost perpetual transition and IBM is not meeting its sales targets for its entry level Multiprise 2000 mainframes or for its larger G4 9672s; the PC business is losing money again and the RS/6000 is in a slump because the Unix workstation businesses of IBM, Sun and HP are being hammered by NT workstations using Pentium II chips. How fortunate that IBM’s AS/400 business is doing well amid all this mess. That’s the partial vindication part. An IBM confederation can ride through such troubles, with a little luck, by having diverse products on different product cycles; but we still think the AS/400 fiefdom, for instance, separated from the larger IBM, would have been better able to compete in the midrange if it did not have to prop up IBM’s other flagging hardware lines at times like these.
By Timothy Prickett Morgan
That said, IBM is happy to spend the money it is coining in the AS/400 market right now. During the first quarter of this year, according to Tom Jarosh, general manager of the AS/400 line, AS/400 unit shipments were up 31% compared to last year, and the resulting revenue grew by 16%. During the second quarter, unit shipments were up 25% and revenue was up double digits, which probably means 11% or 12%. Jarosh says that, so far, third quarter revenue and shipments are still in the double digits growth, and, if the Apache and forthcoming Northstar model 150 entry and model 170 Invaders keep selling like hotcakes, IBM could potentially sell over $5bn in AS/400 hardware and systems software. A lot of that growth is coming from Year 2000 and euro conversion projects, but Jarosh says that about 25% of revenue this year is coming from customers new to the AS/400. These companies are by and large installing relatively inexpensive AS/400e server models and running popular ERP suites. From IBM’s perspective, things could be a lot worse, and they have been. In 1988, when the AS/400 came onto the market, IBM’s aggregate share of the midrange market – which IBM quantifies as share of revenue for servers that sell for between $15,000 and $1m – was only 10%. That included 9370 VAX killers, System/36 and System/38 midrange boxes and a couple of different kinds of specialized servers IBM sold into the telecom markets. In 1992, the AS/400 – especially the D machines, which offered unprecedented price/performance from IBM – had put IBM back in the race. IBM’s share of the midrange market rose to 18% for the year, with 14% coming from the AS/400 line and another 4% coming from IBM’s RS/6000 Unix server business. This put IBM, however briefly, ahead of most of its competitors. Not only was IBM back in the driver’s seat in the midrange in 1992, but its AS/400 business was experiencing doubled digit growth in a midrange market that was growing at between 4% and 5%. More startling is the fact that 60% of the AS/400s sold in the US that year went to customers who were buying an IBM midrange system for the first time. The remaining machines went to existing customers who were adding new applications or converting S/3X equipment. In 1993, as the world economies were struggling to emerge from recession, the midrange market only grew 1% or 2%, and the AS/400 pushed as hard as it could and only grew at between 5% and 6%. The midrange market started to shrink in 1994 as PC servers that sold for under $15,000 began to take over jobs reserved for Unix or proprietary servers; the AS/400 division took its hits again. By 1995, the Advanced Series AS/400s, which offered a big price/performance advantage compared to prior generations of AS/400s, got the AS/400 business growing at a 25% clip for a few quarters, but the new PC-dominated midrange server market was growing so much faster that IBM stopped talking about the AS/400’s share of the take. In 1996, which was a stellar year for IBM’s AS/400 division – the AS/400 was back up to 16% of midrange market revenue and growth was double digit again thanks to the RISC AS/400s. Last year, IBM lost some momentum as it transitioned to the AS/400e Apache models late in the year, but now seems to have gotten the wind back in its sails.
Shrinking market
The overall AS/400 business – by which IBM means AS/400 hardware and software sales, maintenance, peripheral sales and CRT, NC and PC sales into AS/400 accounts as well as services rendered to AS/400 customers – may have been growing for the past decade, but between 1992 and today the core AS/400 processor business has been shrinking, and there’s not much IBM can do to stop it except to try to make it up in volume. In 1993, IBM sold about $3.25bn of AS/400 processors and processor upgrades; the following year, processor sales slipped to $3.1bn. Two years later, as it was shipping very fast RISC processors that had extremely low prices (at least for AS/400s), processor revenue dropped to about $2.6bn. What saved IBM’s top and bottom line was the shift from external AS/400 disk subsystems, which were sold by IBM’s San Jose disk division, to internal disk subsystems, which are designed, built and sold by IBM Rochester. As CPU revenue shrank at the AS/400 division, disk subsystem revenue increased radically. This allowed IBM to say that AS/400 hardware sales were steady, down or up each quarter over the past few years. Even still, in 1996, AS/400 hardware sales didn’t fall as far as they might have because there was pent-up demand among 9406 customers, who bought lots of RISC machines once V3R6 was stable. But the first two quarters of 1997 were pretty weak, and though AS/400e models sold better than expected at the end of the year, it was still not enough to offset revenue declines early in the year. Last year, IBM sold $4.2bn in AS/400 hardware, of about $2.5bn was for processor sales. This year, as IBM closes many deals involving thousands of small AS/400s – Allstate just bought 18,000 machines – IBM will probably sell about $5bn in AS/400 hardware, about $3.3bn of which will come from processors. If Big Blue does this, it will be a record year, at least as far as shipments are concerned, for the AS/400 division. IBM says that the AS/400 is the market leader in the world’s emerging (and right now, receding) economies for servers that sell for $100,000 or more. In Poland, three quarters of the banks run on AS/400s, not mainframes, not Unix servers. Russia’s Stolichnaya bank is using AS/400s, too, and so is the Russian conglomerate Gasprom. So is Bank Islam in Malaysia. Over a thousand AS/400s are installed in China, and the number is growing very fast. The customs bureau in Yugoslavia (or what’s left of it) has a hundred AS/400s in its depots to keep track of what comes and goes in the region. In Shanghai, the stock exchange ditched its HP 9000 Unix servers for AS/400s because they could get better service and applications for the IBM midrange equipment. Customers who are far removed from the NT and Unix talent pools available in the US and EC will likely buy AS/400e machines. The key for IBM’s success with the AS/400 in these markets is to pitch IBM’s worldwide reach for service and to make sure that there is enough AS/400 programmers and systems analysts in these markets to meet demand. The first part is easy, while the second part IBM has only begun to address with its recent education and business partner certification initiatives. And if IBM helps get business partners pushing AS/400s to support the new e-business and groupware workloads at AS/400 sites even while they are trying to keep up with their customers’ Year 2000 conversion efforts, IBM may even be able to push as much as $6bn in AS/400s in 1999, which is good news for the AS/400 division in particular and for IBM in general. IBM is going to need all that extra money to prop up its flagging mainframe sales and to buy back stock to artificially inflate its stock price to keep shareholders happy.
Content from our partners
