Programmable Notes

Wall Street Speaks, But Who Listens? A Wall Street investment report on programmable logic is solid, but not favorable to all the players. by Larry Waller

Because of the infamous "not invented here" factor, people on the inside of any business or field seldom pay much attention to what an outsider has to say. Semiconductor people especially are proud of ignoring outside business advice, since the industry continually reinvents itself in ways that often contradict so-called conventional wisdom.

Ignoring the recent report Wall Street powerhouse Smith Barney compiled on programmable logic, however, would be a shortsighted mistake for those wise enough to ponder informed external opinions. For openers, it is an excellent, up-to-date summary of markets, technology, and competition in only 25 pages--an accomplishment in itself in this mercurial field. Don't look for market data that breaks much new ground, however, since it's mostly what available to any well-informed programmable or EDA company executive, from the usual market research sources.

The real value stems from who the authors are, and the point of view they bring to bear on what is, after all, still a very new business. Semiconductor analyst Hayley W. Kissel did much of the work, along with veteran James L. Barlage, who is the ranking Wall Street chip analyst in terms of both time spent in the business and earned respect of his peers. Barlage has followed semiconductors for almost three decades.

While many of the Smith Barney observations sound remarkably similar to those expressed by company executives not exactly noted for restraint in predicting a rosy future, there's something about judgments coming from an independent vantage point that have the ring of authority. Too, some of the opinions have an edge to them, and will hit nerves with especially the proponents of non-proprietary software and antifuse technology. Also, in these days of the roaring Bull market, the aura of picking investment winners, always the target of ace analysts such as the Smith Barney duo, carries much weight. In other words, to slightly warp the old slogan, "When Wall Street talks, people listen." And they listen even though many industry executives say they disdain Wall Streeters for their well-known obsession with short-term results.

Not surprisingly, the Smith Barney report calls programmable logic "one of the more dynamic growth opportunities in the semiconductor industry over the next four years." The high-density PLD segment in particular is where the action is, showing a 30.4 percent compounded annual growth rate through 1998. These higher-density devices will soar on the strength of replacing simpler logic parts, along with fulfilling the promise of making a big dent in the low end of the gate array business.

Penetrating this low-end gate array segment, estimated at some $1.2 billion a year, does not demand the absolute price parity between arrays and PLD parts that often has been stated as a requirement, according to the report. Rather, a 3:1 ratio of PLD price to gate array price is viewed in the report as "the threshold at which a designer will switch to programmable logic." The well-recognized benefits of PLDs in fast design turnaround and lack of stiff NRE costs bode well for PLDs in outweighing the cost advantage of gate arrays. PLD prices will continue to drop at about 30 percent a year, which itself should bring near parity by decade end.

The Smith Barney presentation is among the first to predict that new designs done on the Japanese home ground of big gate array producers could open up major potential for U.S. PLD vendors. The numbers tell the story here. Japan now uses 37 percent of all gate arrays and only 12 percent of PLDs and has a low programmable market presence. Moreover, it is predicted that with the design of new generations of electronic equipment in the offing, fab capacity tight and growing costs of capital, the situation is ripe for important PLD gains in Japan.

A point made over and over is one that will gladden the hearts of industry pioneers such as Bernie Vonderschmidt, CEO and founder of Xilinx (San Jose, CA), who have long preached it: "While competition between the existing participants ... is intense, there exist substantial barriers to entry, making it difficult for new entrants to gain a strong position or disrupt the competitive landscape."

The fragmentation of PLDs itself, says the report, where vendors sell slightly different architectures (with proprietary tools to match) makes the going especially tough for newcomers to fit in. The only surefire way to get established is what has proven virtually impossible to pull off since the early years. The strong technology and pricing positions resulting from the nose-to-nose competition between Xilinx and Altera (San Jose, CA) in particular pose a daunting barrier for new entrants.

The Smith Barney take on software tools also is somewhat different than current thinking, which assumes that proprietary tools are on the decline, and the move toward architecture-transparent third-party packages is an irresistible wave. Not so, they say. For one thing, designers "have attained a certain comfort level with the tools they are using and are reluctant to switch to a new vendor."

This reluctance to change vendors, in no small part influenced by software issues, was not supposed to be the case in a commodity business marked by the high-competitive level of PLDs, but has become what Smith Barney terms "characteristic of the PLD market." This fact is one reason for the change in plans for heavy-weight chip firms with vast resources that were expected to exploit the mass PLD market. Intel (Folsom, CA) and Texas Instruments (Dallas, TX) both have withdrawn by selling their PLD operations, and Motorola (Mesa, AZ) is experiencing a slow start with its FPGA line.

The result is all to the good for early birds Xilinx , Altera, Advanced Micro Devices (AMD, Sunnyvale, CA), and Lattice Semiconductor (Hillsboro, OR). The proof of this view is market success. Together, these four firms presently command about 75 percent of the total of CMOS PLD revenues and are showing little sign of giving much of it up.

The semiconductor analysts therefore greatly favor the companies that continue to regard software development as equally critical to silicon. Altera, for example, has a leading technological position strengthened by its proprietary software offering. By contrast, there is AMD's decision to spin out tool development to third-party EDA house Minc (Colorado Springs, CO) "This may prove to be controversial," says the report, "since in our view, proprietary software and control of development represent a competitive advantage."

Xilinx also gets kudos for having proprietary tools, but it is noted that "they are apparently more complex to use versus Altera's products." (Some skepticism is probably in order regarding the Smith Barney position vis-à-vis the hot Altera/ Xilinx rivalry.) While the report is largely evenhanded in its evaluations, the investment firm appears to favor Altera's prospects, no doubt as a result of having recommended the PLD company as a top investment choice.

As for spotting new technology trends, the Smith Barney duo are far from ready to jump on the antifuse bandwagon. The technology developed by Actel (Sunnyvale, CA) in the '80s and used since then has small cell features and much high-density potential. Xilinx licensed it a few years ago and is scheduled to bring out its own antifuse line this summer. The report points out, however, that Actel had fabrication troubles in the past, which could signal manufacturing difficulties that might be a real drawback against antifuse taking off. In their view, one-time programmability also is a limitation. "Antifuse has yet to prove itself as a mainstream PLD technology," the report concludes.

Larry Waller is a contributing editor to Integrated System Design.

To voice an opinion on this or any Integrated System Design article, please e-mail your message to: michael@asic.com.

integrated system design  August 1995