EDA Economics Changing

What defines an EDA company may not be the same as the number of startups continues to dwindle.


From most perspectives, there has never been a better time to be in the EDA business. Automation tools are in demand as complexity rises, and new companies jumping into the semiconductor business are starting out with commercially available tools rather than developing their own—and taking years, sometimes even decades, to replace them.

EDA’s slice of the semiconductor market consistently has been about 2%, which is a healthy market given the fact that semiconductors account for about $300 billion ($299.9 billion last year, according to Gartner’s projection), with huge new opportunities in the Internet of Things, automotive, medical and industrial markets. Moreover, EDA companies have taken on side jobs in other markets ranging from wiring harnesses, power testing, software, photonics and services to bolster their balance sheets.

Even the problems being addressed by EDA are interesting. They’re getting more interesting, too, as chipmakers push the boundaries of physics at 10nm and begin stacking die. For engineers, materials scientists and mathematicians, these are the most challenging problems on the planet, and the ability to create profitable companies out of solving these issues is fascinating.

What isn’t so positive is the lack of startups in EDA today. On one hand this makes sense, because the ability to reach critical mass and go public has all but evaporated from the EDA world. The last IPO in EDA was Magma Design Automation in 2001. Apache Design and Jasper Design almost went public before opting to join forces with Ansys and Cadence, respectively. The same was true for EVE, which was swallowed up by Synopsys as the emulation market began exploding. For investors, this isn’t exactly inspiring odds. It takes up to 10 years (sometimes more) to see a return on EDA startups, while it takes a fraction of that in the social media space. Moreover, in social media they can sell air—companies with lots of market attention but no profits—at a huge return on invested capital.

There is rampant speculation that social media, and possibly even cloud services, could resemble the dot-com implosion of 2000. But that still doesn’t mean EDA will attract more money. In fact, most of the EDA startup investment occurred as an adjunct to previous bubbles—being able to build powerful and inexpensive chips for network and application servers and personal computers. Whether it can garner the same attention for the Internet of Things or related vertical markets remains to be seen.

So what’s next? It’s not that there won’t be more EDA startups. There will always be a market for smart people to create new tools that solve specific problems, with a significant payback. The IP market is one of the newer wrinkles in that fold, and it has been profiting quite nicely on the acquisition circuit. Moreover, it’s a seller’s market, and all of the Big Three are on the prowl for new companies with good products in search of increased market reach.

But there aren’t enough of these startups left to continue with the economic model of buy versus build—as in buy companies rather than build the technology from scratch. Buying companies with existing products is a less-expensive alternative to building from scratch. The products are usually market-proven and tested, the infrastructure is in place to keep those products current, including top engineering talent, and EDA has a long and rather unique history of successfully absorbing and integrating companies.

Buying companies, moreover, increases top-line revenue while allowing public EDA companies to hold the line on R&D as a percentage of revenue. It’s a smart approach to corporate math because it adds to the lineup of good tools and quells complaints from stockholders about spending too much internally. Developing new tools raises those operational costs, while buying them falls into a different part of the budget and gets offset by new revenues.

It’s a nice way to do math for investors, but as the number of startups and small companies continues to dwindle, big EDA companies either will be forced to control their R&D costs or somehow effectively explain to stockholders why certain costs are rising well beyond their already high R&D ratios. Or they will need to boost their overall revenue by selling more tools and services—or selling them differently, possibly through rental models—or begin buying startups in adjacent or different markets to continue with this strategy.

It’s not that EDA fundamentally is in trouble. Far from it, given the need for new tools to deal with complexity and rising sales of existing tools such as emulation and various verification and prototyping approaches. But it is about to undergo a shift that could make it far harder to define exactly what an EDA company does, or where its revenue base will be in the future.