|
May 29, 2003 – The struggle for software design firms specializing in microsystems once revolved around the technology itself. But now that they have succeeded in developing tools to help companies progress, they face a new challenge: designing business models to support them until the industry moves into the mainstream.
|
|
There’s more at stake than a few hundred jobs and a handful of companies. Without appropriate design tools, microsystems and MEMS makers risk getting bogged down in development while their competitors race ahead. The solution, some analysts and software designers suggest, involves more cooperation.
|
The leading software toolmakers — Coventor Inc., Corning IntelliSense Corp. and MEMSCAP SA — all were born in the ’90s with government support and spurred by private investment, initial public offerings, or acquisitions. Each company expanded its offerings well beyond design tools, moving into such areas as professional services, and even intellectual property and product development.
With the economy and key market niches like telecom stalled out, the companies recently refocused on making and selling software tools. Their adjustments included cutbacks, consolidations or external collaborations with integrated circuit software designers — and even early talk of cooperation between two of the firms.
Though the moves and methods vary greatly, the overall game is the same: survive to succeed.
“It’s clear that the industry is at another inflection point of sorts,” said James Marchetti, Corning IntelliSense’s director of software business development. “Technically, tools are getting better, and tools have become accepted in the industry. The question is just whether or not people are able to make viable business out of what they’ve done.”
MEMSCAP President Jean-Michel Karam said his cost-cutting plan is not just to get his company on the right track, but also the entire microsystems design industry, with annual revenues of roughly $10 million.
“How can I make MEMS move from specialized technology done by experts in MEMS using software … to any semiconductor system designer? How can I bring to MEMS new opportunities that make it successful?”
Advocates and advantages
Getting over the first hurdle — developing software that can capture a design of a microdevice, model a way to make it and predict how it will perform — was by no means assured back in 1987, when Stephen Senturia spoke at the Transducers conference in Japan. The Massachusetts Institute of Technology engineering professor lobbied for a link between design tools used for integrated circuits and the new world of microsensors and actuators. But the missing link was something that enabled developers to model and simulate three-dimensional devices that combine mechanical and electrical systems.
The idea led to the creation of MEMCAD, an MIT research group overseen by Senturia and funded by the FBI and the Defense Advanced Research Projects Agency. “I think we were the pioneers in trying to create integrated design systems for MEMS,” said Senturia, now founder and chairman of Polychromix Inc., a Boston-area optical MEMS technology startup.
MEMCAD members also were pioneers of the industry. Fariborz Maseeh left the group in 1991 to form IntelliSense, which Corning Inc. bought in 2000; and John Gilbert licensed MEMCAD technology to create Microcosm Technology in 1996, which was renamed Coventor in 2001.
Senturia, who now serves as a technical adviser to North Carolina-based Coventor, said the design firms largely have delivered on the field’s early goals by putting highly sophisticated yet user-friendly tools in suppliers’ hands. But the toolmakers’ biggest challenge is the same facing microsystems manufacturing as a whole: making devices that meet all required market specifications at an acceptable cost.
“Many of the MEMS applications are relatively low volume in terms of unit count,” he said. “If there are not enough successful products built out of a common technology, there’s not enough business to keep a fabrication line full. That makes a fabrication line uneconomical, and the infrastructure required for manufacturing disappears.”
He said design tools can help overcome this challenge by understanding the manufacturability of devices — predicting reliability and the range of strange phenomena that can crop up once they are in use.
“Design for manufacturability is where there’s money to be made,” he said. “If you design something that is robust and can be manufactured, there’s enormous payoff. The tools themselves can’t do it, but they are essential support. You can’t rely on the tools for creativity.”
The design firms need ingenuity on the business side as well, said Jim Walker, an analyst with Gartner Dataquest. “It’s a real challenge … when you don’t have the standard design tools,” he said. “There’s no plug and play in MEMS, and that’s part of the problem.”
Standardization comes with greater business integration, and Walker sees two approaches for design firms: vertical and virtual.
He described Analog Devices Inc. as a vertically integrated firm: The Massachusetts-based supplier of accelerometers to the auto industry does its own design and manufacturing. “As the industry matures, you then start having companies that are vertically integrated.”
In a virtual model, Walker said companies align themselves with other companies through strategic partnerships and agreements short of a buyout — at least initially.
“Companies that align themselves with other companies — they form a supply chain that is basically virtually integrated,” he said. “If you don’t have virtual or vertical integration, you can’t have the industry move forward.”
Some model business plans
Design firms increasingly are moving toward virtual integration, largely through sales and marketing deals with IC designers.
MEMSCAP, launched in 1998, was first to move in that direction. It teamed up with Ansys Inc. in 2000 and Cadence Design Systems Inc. in 2001. It also has a long-standing relationship with Tanner EDA, first marketing Tanner’s MEMSPro advanced analysis software for MEMS, then manufacturing it and ultimately buying the tool suite outright.
MEMSCAP, which is based in both France and Silicon Valley, recently started working with Corning IntelliSense on jointly selling complementary software. The firms won’t reveal details, but MEMSCAP officials see the value of engaging rivals.
“It’s better for MEMS CAD groups to collaborate and grow the market than to compete,” said Mary Ann Maher, MEMSCAP’s general manager of CAD business. “When the market matures, then it’s time to compete.”
Patric Salomon, a Berlin-based microsystems consultant, said there has been little cooperation among the major design players so far. But established partnerships between IC and microsystem designers have been good for the businesses of both: The IC side gets exposure to a new technology without making a major investment, and microsystems can rely on proven tools for the silicon work — bringing a de facto standard to small tech.
Coventor and Cadence announced in February they would sell software that integrates each other’s tools. The partnership was touted as a way for MEMS and IC designers to share design and simulation data to boost development and productivity.
The announcement came after a reorganization by Coventor last year that included laying off about 50 of more than 100 employees, spinning off a biotech unit called Teragenics Inc., and dropping its professional services business that provided design consultation services and support. It also moved away from an effort to act as intermediary between product companies and various foundries.
“We’ve gotten a little bit back to our roots — we had an awakening,” said Gail Massari, senior manager for marketing communications. “We got diverted like everyone else with the promise of high-yield products. … We’re leaner and we’re much meaner, but we’re really comfortable.”
MEMSCAP, which trades on the Euronext exchange, and Corning IntelliSense have not scaled back as far or fast, but each has looked for ways to refocus on software. Marchetti declined to give specific numbers, but said IntelliSense has experienced “relatively small” job cuts and has benefited from the stability of being part of a larger company. MEMSCAP’s reorganization has included cutting about 70 jobs and consolidating five business units into three: design and manufacturing; components, modules and systems; and sensor solutions. Both said most cuts were not made to the software side.
“We really have continuously developed software but not pushed it very hard on the market,” Karam said. “I have said to the company, ‘I want complete engagement.'”
Despite the differences between microsystems and IC design, Karam said his industry must follow IC’s lead. He said design software makes up 3 percent to 4 percent of the $150 billion semiconductor market today, and he believes the percentage should be about the same within MEMS — now estimated at $4 billion — for collective success. But getting there will be as much about business as technology.
“I don’t believe the Ph.D.s of MEMS will make the MEMS market a big market,” he said. “I believe MEMS needs to ride the semiconductor horse. We need to drive … MEMS (to) non-MEMS experts.
“You don’t want to disappear before 3 percent of the market arrives.”