Can IBM really get 50,000 of the new AS/400 Advanced 36s away by March as has been suggested (CI No 2,518)? While it is credible that IBM has sold at least 208.000 System/36s over the many years, it is less credible that there are that many still out there. Where are they? How many in the US? Europe? Are there a zillion of them in the third world? If half the base (whatever it actually is) is in places with a median personal income below $1,000 a year, the System/36 is the IBM 1401 of the 1990s. IBM won’t displace them because it can’t find them. Customers don’t go to IBM for service. They are far more likely simply to copy and keep using the software, however illegal that might be. They don’t care about power bills when the savings are more than offset by the IBM service costs they are not now paying – and with the new box they will have to pay service bills until there is a base of old iron from which to cadge cheap or free spares, as there is now in the 36 market. Consider a comparison between the new kludged PowerPC 620-based micro36 and a Pentium P90 machine: such a contest is reasonable if not necessary if one really wants to examine the prospects for the new box. A P90 typically can have up to 128Mb and some motherboards, such as those from Micron Technology Inc, run to 192Mb, and those ones also use 3.3V chips.
$10,000 slack
A P90 can have more disk space than a micro36 and there are plenty of RAID 5 disks available for less than IBM charges for its non-RAID files. In any event, a comparable configured P90 probably costs less than $5,000 if you are willing to examine models other than the priciest IBM or Compaq Computer Corp server units. For instance, the Hewlett-Packard Co P90s in the US (which are now beginning to be sold here in Europe too) qualify, and don’t they have a local network attachment on the motherboard? For such a P90 there is Santa Cruz Operation Inc Unix, Solaris, NT and OS/2 plus Novell Inc software. There are many databases, just about all of them more sophisticated than the data structure required to meet or beat an S/36. There are also related word processors, spreadsheet packages and applications galore. And of course consultants by the thousands. Granted, you can’t just move your applications and data from an S/36 to a personal computer but if you have $10,000 slack compared with the $15,000 that the micro36 costs, you certainly have some options, including getting full personal computers instead of green screens… or even a personal computer board that will drive your old green screens. Note that IBM’s price assumes you have terminals and a printer. If you need a new printer, are you not far better off in the personal computer world? The same applies for new terminals, which for under $2,000 might come with Windows and enough MIPS to stretch a P90 well beyond the capacity of a micro36.
By Hesh Wiener
On that basis, it seems likely that IBM will sell nearer 15,000 than 50,000 of these new babies during the next two years, and at the same time stimulate some other 36 users enough to get them to move to an AS/400. And if the PowerPC-based 400s are launched before IBM is most of the way to 15,000, it seems likely that the bottom machine in that group might be more attractive in every way than the micro36, which by that time had better cost a third to half what it does now. That last point also contradicts the view that the new 36 will have good residuals: it is very doubtful, at least by comparison with a 36 viewed a decade ago. Granted it will carry much better residual value than a P90, which is likely to be worth zilch in three years. A lot of the value of a micro36 is in the disks and those disks are not going to be worth much in the future because IBM will move to 2Gb and 4Gb and maybe 5Gb units in the same form factor without appreciably increasing the cost to itself or the price to the user. Moreover some of the micro36 buyers might be those that would otherwise move to a more costly AS/400, so the net impact on IBM revenue and profit could be somewhat less rosy than that produced
by an incremental sale of 15,000 or 50,000 of any multi-user system. IBM’s finance terms on the new machines are structured in such a way that the company is in effect discouraging anything but purchase of the smallest of the configurations. Only the largest of the three has an interest rate that is reasonable (in the US) and the terms are likely to be far worse in Europe – and it is widely acknowledged that 36 shops are poor and won’t easily find $12,000, let alone $15,000 or more. It would be hard to prove, but it is reasonable to assume that most if not all the 36 shops in the advanced countries and a significant number in poor countries also have personal computers of one sort or another, so the concept of migration may not be as far-fetched as IBM would like to think. And a key factor in all migrations and all software development is the changing environment. If users have to rewrite software, the migration issue largely disappears because a rewrite for the 36 and a rewrite into something running on another host may amount to the same cost. Remember the old Datamaster? Also known as System/23? A strange little IBM machine that slotted in below the System/34, was built around the 8085 microprocessor. Suppose you had one of these things, and somebody came into your shop with a very fast Datamonster with a proprietary 16-bit 8085 in it, when your old Datamaster was clearly ill.
Supermaster
All your familiar software would work perfectly. You might have to change to 3.5 diskettes but even that could be prevented. The new Supermaster might use much less electricity and come with a three-year warranty. One part of your personality would prefer it to a standard personal computer – you’d learned to love to hate it. But in the end what would you do? Particularly if the new machine cost twice as much as the PC? And if you were told that every possible future upgrade of this machine, barring a very unusual change in the market, would be to a machine running Datamaster/OS under Windows 95 or Windows NT, but that the emulation would be perfect, faster than what you have and of course easy to use. Now let’s add a wrinkle: in the odd event that the new machine fails, you have no other back-up around. But a regular PC, with all its problems, is nothing but a software platform essentially identical to a lot of others in your office and, if for any reason you need outside support, it is identical to machines used all over the place. There is a big 36 base. It’s happy. The 36 is a great computer. But it might be a mistake to confuse the inertia of the 36 base with the momentum that gemerates computer sales. Momentum does not favour the 36, and although it does still seem to favour an AS/400, there is a lot of data out there indicating that IBM will need another price-performance boost – the PowerPC ought to do it – to restore momentum in the AS/400 segment.
Copyright (C) 1994 Technology News of America Co. All rights reserved.
