This joint effort will affect Unix desktop system markets since it includes IBM, the number three market share supplier of workstations for 1991. The key factors in this impact are the Power PC and enhanced AIX. Since the delivery timeframe of product is 1993-1995, this impact is definitely long term in nature. The Power PC chip (see Table 3), will create a low-cost RS/6000 system and a very high-performance Apple product. InfoCorp expects IBM to be a large volume workstation supplier, second to Sun and even with or slightly ahead of Hewlett-Packard in market share by that time since a Power PC type of chip was expected from IBM before the joint effort. The Power PC on an Apple product will put the company in a position to compete for market share in the Unix workstation arena, something we had not anticipated. If the chip is available on schedule with the right performance (see Table 3), the result will probably be an Apple plus IBM share greater than Sun and make the RS/6000 CPU micoprocessor the leading volume RISC CPU by 1995 – a position the Sparc CPU maintains today. In the unlikely event that Apple switches its entire product line of Unix and non-Unix systems to the Power PC we would expect the RS/6000 to be the computer industry’s number two volume CPU behind Intel’s X86 family, supplanting Motorola’s 680XO family. Considering the AIX enhancements we see reinforcement for the market share changes due to the Power PC chip development. Given the timeframe of the AIX enhancement (to have Apple’s A/UX-Macintosh functionality) this joint effort appears really to be an enhancement to OSF/2 (which will subsume AIX by 1993-1994 in our estimates). A full conversion of A/UX to OSF/2 plus Macintosh would create an operating system that could run AIX and OSF/1 applications (now nearly 2,000 in number) plus all Macintosh applications (close to 6,000 in number) with no alteration to binary or object code or user interface. If this top end version is on IBM workstations and Apple systems it would at least guarantee IBM number two position behind Sun (and possibly make IBM number one in Unix workstations) given the suddenly richer Unix plus Macintosh application set available to IBM users. Of course this top end version will also allow Apple to offer all AIX/OSF applications (unaltered) on its systems and pull Apple into the engineering/design segment for the first time, making it a major – top five in market share – Unix workstation supplier. If Apple sticks to its current distribution channels, which don’t conflict much with those of current Unix workstation leaders, this market pull will enlarge the Unix desktop market. On the other hand, attempts by Apple to utilise only the same direct sales and OEM channels used by Sun, Hewlett-Packard, IBM and DEC for Unix/VMS workstations won’t enlarge the Unix market, but will distribute market share differently, at the expense of Sun and Hewlett-Packard primarily and IBM secondarily. InfoCorp considers the former channel strategy to be Apple’s dominant method although some direct sales will also be used. As a result we expect the Unix workstation/desktop market to expand above our current forecasts if and when the Motorola-IBM-Apple joint effort delivers hardware and software. InfoCorp’s UMSMET1991 presents estimates of this market impact. At best we would expect this joint effort to add 30% to 40% to Unix desktop shipments by 1996-1998, over and above InfoCorp’s current forecast.

Table 4: 1995 Impact of IBM-Apple Success

OSF Acceptance/Share +++ OS/2 Usage 0 RS/6000 Success/Share + 88000 Usage — RS/6000 Chip Share +++ Micrsft Applns Bsns ++ Motif Usage + Microsoft OS Busns Apple PC Share 0 Sun, Sparc Mkt Share – Apple Wkstation Share ++ Mac/Unix Integration ++ IBM PC Share 0 Mac/SAA Integration +++

0 no impact; – negative; + positive; number: degree

We do not expect the object-oriented user environment discussed by IBM-Apple (Item 1, Table 1) to have a big impact on the Unix business owing to its poor definition and uncertain timeframe. This development will have an impact on th

e PC market and is discussed in InfoCorp’s Microsystems Infoflash, are Apple, IBM and Motorola Taking AIM? dated July 10, 1991. The final Unix-related impact of this joint effort is on Motorola’s 88000 RISC microprocessor. We believe this joint effort signals that Apple will not use the 88000 in any future products. As a result the 88000 is not a viable CPU for desktop systems and is likely to be dropped from multi-user systems by 1994 – the end of the 88110’s life cycle (we expect the 88110 to ship in volume by January 1992 and offer up to 100 MIPS of performance). Although this agreement is the long-term death knell for the 88000 as a top end CPU performance leader its production life will be quite long given such design wins as that at Ford Motor Co that will put an 88000 in every Ford sold. Since the demands for such embedded applications are very different than for desktop CPUs, we expect future 88000 microprocessors (after the 88110) to diverge from desktop or multiuser system requirements and eventually not be used at all.

The big ‘if’ that is the key to success – can it be delivered in time?

The IBM-Apple joint venture has a big ‘if’ associated with it whether the three major new products can be delivered on time. If the promised products are delivered there will be a few short term effects – by 1992, but the long term effects will be as summarised in Table 4. The object-oriented technology product has the biggest market window (and lowest impact on the Unix systems market); 1995 is an acceptable first customer ship date. The AIX enhancement (AIX+Mac) has the shortest time requirement and biggest Unix market impact; 1993 is the acceptable ship date. The Power PC has a moderate Unix system market impact and its timeframe is flexible, based on performance delivered (see Table 3) – the later it is the faster it must be.

Obsolete

There is a good chance (we estimate a 70% likelihood) that the three products can be delivered by 1994 and make the joint effort successful. There are two major risks in this work, however. The first is whether the object-oriented user environment can be created and be cost effective for PC and workstation systems in terms of CPU performance and memory size requirements. Until more details are available on product direction this risk cannot be assessed any further. The second risk is whether the Power PC chip can actually be manufactured with acceptable yields and performance. We believe that at least 0.5 micron feature sizes are needed and the resulting chip would have 3m to 5m transistor equivalents – both requirements necessitate major advances in semiconductor process technology that as a worst case may not be production ready until 1995-1996. If this worst case transpires the Power PC is at risk of being too slow or obsolete. These are serious risks, but the rewards of a successful joint venture to IBM, Apple and Motorola are significant also. IBM stands to maintain its strong market share in the face of tough competition in the PC market, gain share in the Unix workstation arena and make an IBM CPU the de facto RISC standard. Apple can also maintain its market share in PCs, gain a top five position in Unix workstations and make the Macintosh user interface a de facto standard for PCs and Unix workstations. Motorola may be the biggest winner – it will become the number one RISC CPU manufacturer and whether Apple chooses to stay with the 680XO CPU in its products or switch to the Power PC for all systems, Motorola will have a strong number two position behind Intel for CPU microprocessor market share for complex and reduced instruction set combined.