IBM’s 4381 announcements indicate that the company’s problems in the mid-range may not be solved for another 18 to 24 months. Instead of replacing the 4381 family with new machines that use completely new and more advanced technology, IBM chose to extend its ageing product line by re-engineering the existing logic circuitry and adding 1M-bit memory chips to crate faster models. This suggests that IBM will be unable to announce a succeeding series until 1989 or later. IBM cannot extend the 9370 line upward before that time unless it wishes to sacrifice its relations with trusting 4381 customers. At the same time, the debut of a baby 3090, the model 120E, puts a performance cap on the midrange. This is the last kicker IBM will offer to 4381 users, notwithstanding expectations that a quad-processor machine is in the offing. Having announced the 3090-120E, IBM will not entice users to buy comparable 4381s that cost less… and yield slimmer profit margins. The only beneficiaries of IBM’s strategy are the existing owners of 4381s, particularly leasing companies. Among these, IBM Credit Corp is probably the most important, as it invested very heavily in 4381 leases when the mainframes were new. Now, IBM Credit will benefit as the leases are extended. Independent lessors with 4381 portfolios will similarly do well; their machines will stay on rent for the next couple of years, too. IBM’s decision to enhance the 4381 a second time is almost certainly due to a technology failure within the company, one that IBM is unlikely ever to admit. However, there are clues that something’s terribly wrong in IBM’s engineering departments. The company simply cannot deliver very many 9370s. In all of 1987, IBM admits, it will only be able to ship somewhat over 2,000 of the processors. By contrast, IBM’s rivals in the minicomputer business have announced computers as fast as the 4381s – or faster – that are powered by CMOS processors and other sophisticated technology.

Update their resumes

In short, it is clear that IBM’s mid-range processor effort is presently far behind those of the rest of the industry. DEC, in particular, stands to benefit from its success in engineering, as does Hewlett-Packard, even though it is late with its RISCbased machines. IBM’s prospects in the mid-range will remain dismal until the company can deliver a new generation of superminicomputers. Sales in this segment will be weak despite IBM’s lead in key technologies used in complete systems: IBM’s disk drives are far better than those offered by DEC and other rivals, and its memory chips are at the leading edge of the merchant market’s offerings. Apparently, though, these assets are not enough. After 1990, we suspect, IBM’s rivals will suffer. For when Big Blue falls behind, it almost invariably leapfrogs competition with the products it finally introduces. This was the case in 1979 with the original 4300s, it is the case with its first serious entry into semiconductor memory in several years, and probably will be the case with its eventual replacement for the 4381s. In the meantime, however, the company’s sales reps will have to tighten their belts, its shareholders will have to be patient, and the executives whose decisions led IBM into this morass may have to update their resumes. – Hesh Wiener.