Taking advantage of new markets requires some fundamental changes throughout the EDA industry.
Semiconductor companies have been knocking on doors outside of the computer industry for the better part of two decades, pitching the value of digital and mixed-signal technology for improving efficiency in many market sectors.
For most of that time, they received polite nods, occasional inquiries for more information and not much else. But over the past several years, those doors have opened up wide. The benefits of semiconductors coupled with the open connectivity of the Internet have created a business case that is so overwhelming it’s impossible to ignore. Business as usual is, to put it bluntly, uncompetitive. There are tangible and measurable benefits to getting early warning that a valve isn’t working properly, an assembly line is about to go down, a life-saving medical device will fail in the next 24 hours, or that a car is about to swerve out of its lane because the driver is distracted or dozing off.
This is a big opportunity for SoC makers, but in relative terms it’s a much bigger opportunity for EDA companies. EDA has been stuck at an almost constant rate of 2% of semiconductor revenues. By broadening that out to systems engineering—whether that’s a chip, a package or a board—that percentage can rise significantly. In his Kaufman Award acceptance speech in November, Mentor Graphics chairman and CEO Wally Rhines observed that shifting to system design could increase EDA revenues from the current $7.5 billion to about $38 billion.
That would turn EDA into a much different kind of market. It would make it much more attractive to investors, to venture capital, and to engineering students. A 407% growth rate would put EDA squarely back in the spotlight for hot markets to watch, a distinction this sector has not seen since the 1990s.
This potential for growth carries new responsibilities, of course. Systems need to be as reliable, or more reliable, than systems that exist today. Methodologies need to be improved or changed out entirely, tools need to be upgraded even further, and there need to be open standards in place that allow multiple tools from multiple vendors to work together more seamlessly—and for new companies to jump in with new ideas about how to do things better.
Building SoCs is hard work. Building tools to make those SoCs arguably is even harder. But the hardest piece of all may be getting different companies to collaborate on how to improve tool flows, share concerns about what needs to be changed and where the weak points are, and consider new methodologies and approaches that are more efficient, less expensive, and faster to implement.
The EDA world has been talking about these concepts for some time, and it has done a reasonable job in keeping pace with rising complexity. As investor Lucio Lanza has pointed out, without EDA Moore’s Law would not have been possible. More recently, he has said that EDA is at the center of the IoT revolution. He’s correct on both counts.
Still, for EDA to live up to its potential, it will have to continue doing what it does well, only it will have to do it much better and on a much bigger scale. When EDA took on the challenge of making Moore’s Law possible, it was a much younger industry with nothing to lose. What’s worth watching is how much of the adrenaline and experimentation of a new industry it can recoup to take advantage of a new opportunity, and how willing it is to remove some of the silo behavior for which it has criticized its customers and its customers’ customers.