From Brian JefferyManaging DirectorInternational Technology Group4984 El Camino Real, Suite 119Los Altos, CA 94022-1433
In an article in Computergram (CI No 1273), you wrote a highly critical review of a report published by International Technology Group, IBM Strategies for the 1990’s. This piece is by way of rebuttal. We would like to thank the editors of Computergram for their fairness in allowing us to respond in this manner. We offer the following resons-es to your article. You argue that the figure of a $1,000m IBM AIX mainframe business cannot be credible because Amdahl has a better Unix operating system than AIX/370, and so could be expected to grow faster than IBM in this market. We have three observations: 1. Within the timeframe of our projections, IBM will be offering a substantially more powerful mainframe Unix environment; 2. Developments in IBM architecture and technology can be expected to make IBM large systems more viable in computer intensive applications; 3. IBM’s mainframe business is approximately 15 times larger than Amdahl’s. Given equality of mainframe Unix environ ments (which IBM should achieve well before 1993), it could be expected that IBM would be doing a great deal more business in main frame Unix than Amdahl. You take exception to the proposition that IBM will be doing less business in Unix technical workstations in 1993 than Sun Microsystems now. We have three observations: 1. IBM’s new RISC systems will be targeting a market whose start-up growth phase has already passed, and which is already begin ning to mature; 2. Eroding Sun Microsystem’s embedded base would be a difficult proposition for IBM and decelerating growth in this marketplace could be expected to constrain IBM’s ability to increase revenues; 3. IBM RT sales in 1988 were $115m. However, no more than 16% were for engineer ing and scientific installations. The re mainder divided between large account comm- ercial applications and small/medium busin ess users. Our projection for IBM growth scarcely seems unreasonable in this context. You argue that our projection for IBM AIX mid-range systems sales cannot be accurate because IBM will not allow Unix supermicros to undercut the AS/400. Unfortunately for your argument, this is already a done deal. IBM management appears to have decided recently to reposition the AS/400 as an SAA mid-frame for corporate accounts. The small/medium business and industry vertical markets originally targeted by the AS/400 have now been assigned to new multi- user AIX PS/2 and RISC systems. One of the primary reasons why IBM rolled back availability of these from October to early 1990 was to prepare RPG support under AIX and S/36-to-AIX migration tools. After February 1990, the primary migration path for S/36 users will be to AIX. A programme of converting third-party S/36 SSP applicat ions to AIX has already begun in the US. It is also worth noting that despite its lack lustre performance as a technical workstat ion, the RT PC has already done remarkably well as a low-end multi-user system. Over 69% of RT PC sales to date have been in S/36-like markets. In the US, the RT PC is IBM’s primary new account seller. IBM has at least 260 value-added resellers for the line, the vast majority targeting small and medium businesses. The base of applications software for AIX is also revealing. Out of 1,992 AIX/RT and AIX PS/2 products available or under development from 466 vendors, 72.8% are industry verticals and a further 9.4% are general accounting and business manage ment packages used predominantly by small and medium businesses. You argue that IBM’s top management is not suppor ting AIX. This is simply not true. IBM president Jack Kuehler has been a major supporter of the AIX program since its inception, and the weight of resources that IBM has thrown at AIX and RISC technologies since 1987 (over $300m in conversions alone) is incredible even by IBM standards. We will, however, leave it to IBM to respond to your argument. We shall very shortly see a massive, widely-supported launch for a wide range o
f AIX hardware and software products. By February of 1990, it is unlikely that anyone will be questioning the seriousness of the IBM AIX effort. You argue that IBM will not set up AIX as a contender for SAA. This is correct, although not in the way you suggest. AIX and SAA are being established as parallel environments which employ the same underlying architectural features – co-operative processing, distributed processing, common SQL database architecture – and common applications development interfaces. Recent IBM public statements have made it very clear that AIX and SAA will not only coexist, but will be highly interoperable. We strongly urge your readers, and the Unix community as a whole, not to underestimate either the technical sophistication of IBM AIX strategy or the seriousness with which it is being pursued. This effort has strong backing at the highest levels of IBM corporate management. The IBM strategy is a highly sophisticated one. AIX forms a parallel environment to SAA in the same way that the company’s OSI network scenario parallels traditional SNA. Interfaces, conventions and protocols may differ, but underlying architec tures are the same and both will be integrated under IBM’s scenario for Enterprise Information Systems. IBM is, at $60,000m, still by far the world’s large computer company and if there are fools in IBM top management, I haven’t met them. Love them or hate them. But IBM is going to be a major force in Unix, and it could be unwise to underestimate what it could achieve. *IBM’s Unix effort, Mr Jeffrey fails to answer the central point of our piece, which was that IBM will need to be doing vastly more than his forecast of $4,100m in Unix systems by 1993 to compensate for Unix systems being such a low-margin business.
