Xref: utzoo comp.ai:1882 comp.lang.lisp:919 comp.misc:2521 comp.sys.xerox:227 misc.invest:1872 Newsgroups: comp.ai,comp.lang.lisp,comp.misc,comp.sys.xerox,misc.invest Path: utzoo!utgpu!jarvis.csri.toronto.edu!csri.toronto.edu!tjhorton From: tjhorton@csri.toronto.edu (Tim Horton) Subject: Re: Symbolics stock Message-ID: <8806060457.AA06947@sheppard.csri.toronto.edu> Summary: reprint Organization: University of Toronto, CSRI References: <57@daitc.ARPA> <33064@linus.UUCP> <692@esl.UUCP> <870@papaya.bbn.com> <1115@gmu90x.UUCP> Distribution: na Date: Sun, 5 Jun 88 23:37:22 EDT In article <1115@gmu90x.UUCP> heldeib@gmu90x.UUCP () writes: >I read in a recent article that many of the AI companies including >Symbolics, of course LMI, Xerox, and software oriented companies >were losing money. I can't find the article at the moment but It was probably the one in the IEEE "The Institute" news supplement to IEEE Spectrum, May 1988, Page 1. "AI: lower costs, new markets sought", by Glenn Zorpette Until a year and a half ago, business was brisk... High technology giants Texas Instruments of Dallas and Xerox Corporation in Stamford, Conn., were among the handful of companies pushing (Lisp machines). But for months, sales of TI and Xerox machines have been slow, and market leader Symbolics Inc. of Cambridge Mass., has reported losses for six straight quarters. Symbolics lost over $25 million on its fiscal 1987 sales of $104 M, and the first quarter of its fiscal 1988 brought a loss of $4.5 M, with another $2.7 M gone in the second quarter. "Artificial intelligence is clearly not the broadly based technology people thought it was three or four years ago," says Thomas J. Martin, director of the AI program at Arthur D. Little Inc., in Cambridge, Mass. "And the amount of training required to make effective use of the technology was vastly underrated. It's not a simple, straightforward activity to sit down and use a Lisp machine." All the same, both Symbolics and TI, the second biggest supplier of Lisp machines, are out to beat the downturn in sales by finding new AI markets. They are aiming first at the gap between high-end $50K units and ordinary workstations running Lisp software. On March 3, TI unveiled the microExplorer computer system, which will sell for only $15K to $25K depending on memory and monitor. It is essentially a Macintosh II computer system from Apple Inc., of Cupertino, Calif., outfitted with a board based on a custom Lisp microprocessor designed and produced by TI. Next August, Symbolics plans to introduce a Lisp microprocessor at the meeting of the American Association for Artificial Intelligence (AAAI) at St. Paul, Minn., according to Neil Weste, a company spokesman. At the same time, Symbolics hopes to introduce a board incorporating its Lisp chip, also for the Macintosh II, to be followed later by boards for other machines. The moves are seen both as an effort to stem the loss of business to cheaper general-purpose machines and as heralding a new phase for the AI computer industry. [slow death :-)] This phase will embrace a wider variety of machines, often inexpensive by todays standards and more narrowly focused on particular groups of users... The drop in sales of today's special-purpose machines was partly due to stiff competition from technical workstations, which are also often used to create Lisp programs. For example, in the last two months, Hewlett-Packard Co. in Palo Alto, Calif., and Digital Equipment Corp. in Maynard, Mass., both announced artificial intelligence software for their technical workstations. Although outperformed by the best Lisp machines, workstations are far cheaper than TI's Explorer II and the Symbolics 3650, which cost more than $50K. TI's new microExplorer, however, offers better Lisp performance than comparably priced general purpose workstations. With the Machintosh II's Motorola MC68020 microprocessor, the microExplorer can run a version of the Unix operating system... TI also introduced a $10K package so that an owner of a Macintosh II can upgrade it to a microExplorer. The microExplorer offers roughly two and a half times the performance of TI's $40K Explorer AI Workstation, introduced in 1984, but about half that of the Explorer II. Though the Explorer II and the microExplorer use the same custom Lisp microprocessor, the more powerful machine is optimized thoughout for Lisp and so, like the Symbolics 3650, is more suitable for intensive AI development work -- for example, in writing programs for personal computers, robots, or superminicomputers. [?] The Macintosh II is coincidentally based on NuBus, like all the Explorer products. Developed at MIT, the 32-bit bus technology is now owned by TI. Familiarity with NuBus simplified the job in integrating the Lislp chip into the Macintosh II, which TI engineers did in less than a year, Smith said. (NuBus is draft IEEE bus standard number 1196). Symbolics long-awaited Lisp chip, called Ivory, is expected to be the most powerful ever. The company is working on boards not only for the Macintosh II but for the IBM AT personal computer and compatibles, as well as for a "popular" workstation whose name Weste declined to disclose. The offerings will be part of a three-tier strategy. "Lisp is going from a research base to a more mixed group that wants Unix or MS-DOS," Weste noted. "For them, lower performance is OK," so a personal computer or workstation equipped with Lisp processor is good enough. This group forms the middle tier identified by Symbolics. Below this group are users with even more modest AI needs, best served by software that runs on conventional processors. Symbolics intends to address this group, too. The highest tier encompasses users doing heavy developmental work or pushing the state of the art in such fields as pattern recognition. They will still need the power of full-fledged $50K Lisp machines, in Weste's view. Word length is one main difference between the Symbolics and TI Lisp microprocessors... certain features of the Lisp language favor longer words... [much stuff deleted] One AI specialist who uses both TI and Symbolics machines noted that the American National Standards Institute is about to adopt a formal standard for object-oriented programming in Lisp... [much more stuff deleted]