At an EDA Consortium (EDAC) panel discussion March 14, 2013, top executives from Cadence, Mentor, Synopsys, ARM, and EDA startup Nimbus shared their views about a range of business and technology issues facing the EDA industry. Panelists engaged in lively discussions about topics including consolidation, the end of Moore's Law, "verticalization" of consumer electronics companies, IP, cloud computing, and more.
The EDA Consortium 2013 CEO Forecast Survey panel was moderated by Rich Valera, managing director of Needham & Company. Panelists were as follows, seated from left to right in the photo below:
- Simon Segars, president of ARM
- Raul Camposano, CEO of Nimbic
- Aart de Geus, chairman and co-CEO of Synopsys
- Wally Rhines, chairman and CEO of Mentor Graphics
- Lip-Bu Tan, president and CEO of Cadence
As was the case with last year's panel, EDAC submitted questions to its membership in advance of the panel. The panelists were then asked the same questions, and the answers were sometimes quite different from those given by EDAC members. Following are some highlights from the discussion.
Q: Has consolidation helped the level of innovation in EDA?
(EDAC members: 26% yes, 74% no)
De Geus: I completely disagree [with the poll results] because it lacks a definition of "innovation." In the last decade, we have moved from point innovation to systemic innovation. That is why we have done phenomenally well in integration.
Rhines: The dynamic in which startups form, innovate, and are acquired fuels a large share of the innovation in the industry. If that dynamic didn't occur, the share of the big 3 [EDA vendors] would not grow.
Q: Could the EDA industry consolidate from 3 down to two major players?
(EDAC members: 48% yes, 52% no)
Tan: First of all, customers want choices. If you scale it back to two, it would eventually come back to three or four. Secondly, you become a more mature industry when you consolidate. I believe we have a tremendous growth opportunity given the complexity of chip design and the challenges of system design. The pie can get bigger and bigger.
Rhines: For the past 15 years the combined market share of Mentor, Cadence and Synopsys has been almost exactly the same. This strikes me as a very stable industry structure.
Camposano: Will EDA become part of a larger industry? More and more value is going into the system, and into the board, as boards get faster and emit more EM radiation.
Q: Does the Moore's Law cost curve break down at sub 2X nodes? If not, what is the most important factor in extending Moore's Law?
(EDAC members: 48% disagree cost curve will break down. Of these, 30% say FinFETs are most important factor in extending Moore's Law.)
De Geus: I don't think Moore's Law is broken at all. On the contrary I think it's morphing into the next episode of life... FinFETs are a "fabulous discontinuity." For the first time in a long time we see a discontinuity in the power/performance ratio.
Rhines: Moore's Law is an exponential and no exponential can continue forever. But whatever happens, we will continue down the learning curve and the cost per transistor will be reduced.
Q: Is "verticalization" in consumer electronics, as demonstrated by Apple or Samsung, a net positive or negative for EDA?
(EDAC members: positive 29%, negative 38%, neutral 33%)
Segars: I think it makes no difference. People approach design from different directions and solve problems in different ways. Apple and Samsung are not 100% of the market.
Tan: Being a VC at heart, I would look at it as a net gain. It's good to have systems companies thinking about verticalization and application-driven design.
Q: What's most important in the electronics food chain?
(As ordered by EDAC members: IP companies, system companies, fabless semiconductor companies, software companies, IDMs, foundries)
Segars: IP has become part of the overall design chain because of the complex chips people do. If you can buy a standard component from somebody else, you really should do that instead of wasting your R&D on reinventing the wheel.
Q: What's the likelihood of any private EDA company going public in the next 12-18 months?
(EDAC members: 19% yes, 81% no)
Tan: EDA startups tend to use angel investment. I think there's a really good return if you pick the right ones. In most cases there will be M&A. To me an IPO is a process, it is not an end game. When you go IPO you can celebrate for a nanosecond - then you have to focus on how to keep that earnings momentum and growth and profitability.
Camposano: Funding is indeed difficult if you stick to the traditional model. You have to go a little outside EDA. We (Nimbus) focused on computing models, using the cloud and parallelism. We're also focusing on board design, where there are over 80,000 customers.
Q: How much coverage is EDA getting from Wall Street, relative to 5-10 years ago?
(EDAC members: 64% said less) What can we do to make ourselves more attractive to Wall Street?
Tan: We underestimate ourselves and how much value we provide. I think the responsibility falls on us to drive growth and drive value. If we can do that the market will love us. The worst thing we could have is stagnation. Then everybody eats each other's lunch and that would be a horrible industry to be in.
Segars: The challenge for the technology industry as a whole is explaining to people what we do. Unfortunately it's not well understood.
Audience question: Why did ARM endorse the Cadence purchase of Tensilica? (Apparent reference to quote in press release about announcement of acquisition agreement)
Segars: On-chip DSP is an area that needs continual innovation. Tensilica is a small company that's been around for a long time with great technology that's often found right next door to an ARM processor on an SoC, providing a complete solution. The fact that Cadence is acquiring [Tensilica] and will keep the roadmap going forward is a benefit to our mutual customers.
Audience question: Do you see the cloud being leveraged more in EDA?
De Geus - We put a substantial effort into that, and of all the things we've done in the last 25 years this is probably the single one where the result is essentially zero. I don't mean some money, I mean zero. The degree of uncertainty about protection of IP was enormously high.
Tan: We are keeping a very close eye. Today customers are not asking for it [cloud] except startups and early stage companies. We're ready, we just have to wait and get customer requests for it.
Next EDAC event: Joe Costello, former Cadence CEO, speaks on "Communicating a Compelling Company Story" at Cadence San Jose headquarters May 1, 2013. For information, click here.
(Photo by Ross Mehan)