The processing architecture of IBM Corp’s computers is increasingly focusing around its Power RISC technology as a means of constructing faster systems at lower cost, said William Filip, IBM vice-president and president of the Advanced Workstations and Systems Division. In a presentation in Paris on Wednesday lst week, Filip, whose division handles the RS/6000, said his group’s technology is not restricted at all to the RS/6000 series of workstations. Using processors based on the Power RISC architecture, which has been adopted also by Apple Computer Inc, Motorola Inc and Compagnie des Machines Bull SA, Filip said IBM will go aggressively into the low-end personal computer market with RISC technology next year. Filip’s division also has a joint development programme with the Intermediate Systems Division to design RISC-based versions of that group’s mid-range computer, the AS/400, he said. The AS/400 team has said it will move to the RISC architecture, which will enable it to increase the performance of the current microprocessor technology for a lower cost, he said. Filip’s group is also discussing with IBM’s ES/9000 mainframe developers about possible use of RISC processors in the mainframe family. Filip acknowledged that there will be more competition with the ES/9000 over time, but for now, we’re focused on the Unix market. He denied that the RS/6000 completely overshadows the AS/400. You’d think with the product line today that we’d overlap the AS/400 completely. In terms of performance, we do, but the overlap of interest in the System/36-AS/400 world in the RS/6000 has actually gone down. The AS/400 is based on a highly integrated relational database; it’s a different culture. Our guys in Austin are more interested in open systems, and the fastest screaming performance they can get onto a microprocessor. The ES/9000 guys are interested in building the best data management engine possible. Even though we’re seeing overlaps in product lines, there are still market opportunities for each of these groups, Filip said. The reason for IBM’s interest in RISC is obvious.

Profit margins increased

Our profit margins in the RISC-based systems business have increased substantially. Actually, that’s not a completely fair statement because 1990-1991 were start-up years for us and we invested heavily. But our gross margins will be at or above our market competitors this year, Filip said. Specifically, he said that his group’s margins would be on a par with Sun Microsystems Inc’s and ahead of Digital’s by a large margin. It is difficult to calculate Hewlett-Packard Co’s margins, he added, because they are in a wider-ranging business. We’re still investing more than they are in development, so our bottom line profits are less, but this will be the first year for the RS/6000 that we make at least $1 net profit, Filip said. The profitability margin will continue increasing, said Filip, because IBM is not investing in development at the same rate that sales of the RS/6000 are growing. In 1991, IBM realised $1,500m in worldwide revenue from sales of the RS/6000, which represented a 50% improvement over 1990. Filip expects RS/6000 revenues to grow between 40% and 50% this year. The split between the commercial and technical applications for the RS/6000 are about 50-50 now, says Filip. A recent survey by Datamation and Cowen & Co showed increases in the percentage of both technical and commercial users that indicated they definitely planned to buy an RS/6000 in the next 18 months. The percentage of technical sites was still higher, however; 22% versus 10%. This week, IBM France announces those additions to the RS/6000 aimed at the commercial rather than technical computing market – the two new RS/6000 systems (deskside and rackmount) which operate 25% faster than current models; higher capacity tape and disk storage subsystems; and software including Cobol compilers, the Distributed Computing Environment from the Open Software Foundation, and the CICS/6000 and Encina transaction processing monitors, launched in the US l

ast month (CI No 2,013). Filip said the symmetric multiprocessing project between IBM and Groupe Bull is on target, and that products should arrive towards the end of next year. He also said that IBM’s development group in Toronto would be coming out with the long-promised distributed relational database in mid-1993. They feel they can distinguish themselves with greater transparency between PS/2s, AS/400s and mainframes, with less translation and overhead. I don’t know if they’ll convince the whole industry, but it will be attractive to some customers. But we’ll still be doing a lot of Oracle and Ingres, he said. – Marsha Johnston