Bigger Systems, Bigger Profits

After years of talk about expanding the definition of the system, it’s finally happening.

popularity

Markets work in very mysterious ways. Technology that should be a slam dunk—think 2.5D with its promise of re-usable analog IP and faster performance, for example—are still hobbling along because no one wants to deal with the risk of a new architectural and manufacturing approach. They haven’t even shown up yet in servers, where price is almost irrelevant. At the same time (no pun intended), the Apple Watch, with a projected battery life that can be measured in hours rather than days, seems destined for greatness.

EDA fits nicely into this picture, too. Growth in this market has been flat in recent years, which explains why most of the available startups have been consumed at bargain prices and VCs are reluctant to invest in new ones. While EDA Consortium stats show steady growth, the EDA portion has been masked by the uptick in related markets such as IP, software, and sales of tools in adjacent markets such as optics and LED design.

It’s long been known that tools developed in EDA are applicable to a broad range of much bigger problems. If you start with the most complex electronic real estate on the planet, which is an SoC, then everything else is manageable. You can even use the same tools, methodologies and flows, because this is an extreme environment where space is measured in nanometers and atoms, and physical effects such as electromigration, electrostatic discharge, thermal runaways and noise are well understood. Everything else seems tame in comparison.

This is just starting to sink in, too, which explains why everyone has begun talking about “Shift Left” on the verification side, and why everything from layout to verification to overall flows are being adopted upstream. Sales of tools are increasing well outside of classic semiconductor installations for development of everything from software to managing the thermal flows inside of data centers.

It’s hard to say if EDA is the right terminology anymore. As Brian Bailey points out in a survey of the industry, it’s clear that electronic system-level design is falling out of vogue. But the bigger picture is that EDA, or whatever it should be called, has a much bigger opportunity for growth as everything becomes connected. Even the Internet of Things may not be the right term, but pervasive connectivity is definitely happening, even in places where it may not make sense given the infrastructure and lack of security.

All of this leads to a really big question: If classic EDA tools are used for more than just chips, what else can they do? The next phase of R&D in EDA may be less about solving challenges at 7nm, although that will certainly continue to happen, and more about whether they can be used to improve software or identify security holes across complex networks that mimic all the interconnections on a piece of densely packed silicon.

We are entering a new phase in electronics, and EDA has a major opportunity to stake out a growth path unlike anything it has ever done in the past. Markets move in strange and unpredictable ways, and this time the path leads to the feet of an established industry with the expertise and experience in solving very similar and very difficult problems.