The subject of how to grow the EDA industry has been around for a few years, and is now even more pertinent since the consensus is that we have covered all aspects of electronic design, from architectural investigation to mask making and yield analysis.
The present economic situation is a powerful stimulus to transform the issue from a subject of academic speculation and after hours banter, to a topic for analysis by CEOs and executive managers, not to speak of the venture capitalists looking for a way to gainfully invest the few resources still available.
The financial picture
A month ago or so, Magma announced that it was restructuring its operations and reducing its workforce, by the now traditional 10%, give or take a couple percentage points. The news is a clear indication that even a company with good technology and access to the public financial market cannot, in our present industry climate, gain significant market share when competing broadly with the three other companies each of which have annual revenue at least four times larger than its own.
What the established companies have in their favor is the length of acceptance and use of a new technology by our customers. Even a breakthrough tool will take about two years to generate a significant revenue stream, since it will take that long for a semiconductor or system company to evaluate, accept, and use a new tool. The cost of training its engineers, integrate the new tool, and use it in pilot production mode to minimize losses due to unexpected bugs, is just too large. It takes a major error on the part of the established vendor to induce users to consider a change.
Just over a week later, EDAC, which by the way has decided to save money by using either Synopsys or Mentor as their public relations agency, released its Market Statistics Service results for 2Q2008. It showed that revenue declined 3.7% compared with the same quarter a year earlier. The only sectors showing any growth were PCB and Services.
But an article in EE Times European edition reported less than a week later, that the German PCB industry sees order entry plummet.
And in the same week, to make matters worse, Reneas COO stated that the economy will have a severe impact on ICs sales and profits. This observation should not surprise anyone, since the semiconductor industry is now dependent on the appetite of consumers for new gadgets and high end products, such as automobiles, with a high electronics content. But, since consumers worldwide are now seeing their ability to purchase on credit significantly curtailed, sales will slow significantly. Will we really need a 22-nm process in 2011?
Of course the most glaring sign that all is not well in the industry came with the Cadence's announcement that Mike Fister and four of his executive vice-presidents had resigned. That news, significant in itself, was then followed in short order by the admission that Cadence was delaying its third quarter financial report pending review related to customer contracts signed during the first quarter of this year.
And finally just a couple of days ago Cadence described what may be the first of a round of layoffs.
The search for growth
Since the customer base shows only weak signs, at best, of expanding, growth must come from new sectors. As semiconductors fabrication becomes more complex, there is the obvious opportunity to invent additional design algorithms to take care of parasitic effects, more complex verification tools, better system to silicon flows, and so on. But although doing this will represent additional sources of revenue, the number of customers ready to purchase the new products will be both small and slow to embrace them.
The alternative is to enter new application areas. The blessing of EDA, to be guided by leaders that truly understand the semiconductors technology is also its major curse. In my opinion it is time to loose the "E" and realize that what we need is Design Automation.
Electronic components are now so pervasive in most products, that limiting ourselves to just solving the problems associated with the design and development of the electronic portion is simply shortsighted when one considers that the most challenging problem is the system design and system integration portion of the development.
In a very recent opinion piece Paul McLellan offers his opinion on how EDA can regain some of its vitality. I agree with much of what he says, although I do not share the view that returning to an industry fueled by small startup is necessary, but I think that this is a time to be bold, to realize the inherent limitations in serving just one major category of customers, developers and builders of integrated circuits, and truly embrace the opportunities that a focus on system design offers.
The EDA industry should look closely at what Mentor Graphics is doing. In addition to being active in the embedded software market, Mentor is now expanding into the Design Automation market. With its acquisition of Flomerics, Mentor did not just retain the part of the company that directly applies to electronic design. In fact it entered new markets. Although fluid dynamics is a discipline germane to temperature analysis in board level design, it has many other applications, in areas such as automobile, aerospace, and hydraulics, among others. Mentor is retaining that group and is purposely continuing to grow its presence in those markets. This is a powerful answer to those who think that EDA can only focus on what it knows best: silicon.
My biggest mistake
Having reached an age that allows introspection without suicidal risks, and having changed career (although not by design), I have enjoyed the luxury of analyzing my past actions in EDA. I did a few good things for which I am proud, a few mistakes, and one thing that now stands out as a real failure. I missed the opportunity to expand the role of EDA in the late 1990s.
In 1992 I joined Intergraph, a company that was a leader in Design Automation and had just acquired the remnants of Daisy to enter the EDA market. Until the end of 1999, when that portion of the company was sold to Mentor Graphics, Intergraph tried to grow a "typical" EDA company, which it named VeriBest, that would serve all of the traditional EDA markets, from design entry, to simulation, synthesis, analog design, board layout, even testing its opportunities in IC layout. At the company I played different roles: one official, as VP of engineering, and others less official mostly dealing with tasks that the CEO would assign to me when he lost faith or patience with a specific project or individual. In those roles I got to deal with Jim Meadlock, founder and CEO of Intergraph.
My big mistake was to never even address with Jim the fact that it was a mistake to locate VeriBest's headquarters away from the major Intergraph campus in Huntsville, Alabama, since doing so meant to make VeriBest a totally different entity with no relationship to the company major business. I should have pointed out how the other divisions would have benefited from using what VeriBest produced and could develop to enhance their own business and thus bring EDA in the MCAD, Enterprise, and even mapping markets.
It was a synergy I did not see and was never able to verbalize, even when I discussed with Jim various ways to improve VeriBest and its financial results. I hope that this lesson will not be lost and that someone will be able to see the real synergy existing among various aspect of Design Automation.
Much has been said about the opportunities of adapting EDA methods to biology. This is not due to the fact that electronics design and biology have much in common. It is due to the fact that biology has moved from being solely a science, to incorporate engineering activity. As such the field of biology offers opportunities to support with computer aided tools those development tasks that require an engineering approach instead of pure science. Would EDA companies be able to expand in the field of biology? Anything is possible given enough time and money: but I believe that Design Automation is a much more natural and easier expansion path.