A New Era For EDA

Is EDA in decline? Will its fortunes turn around with the end of Moore’s Law?

popularity

For almost the entirety of my time in EDA, which is approaching 40 years, the total number of design starts has been declining. A lot of this is to be expected. Semiconductors was the new and exciting world where startups popped up hoping to make it big in Silicon Valley, or indeed to become the next Silicon Valley, Forest, Swamp or Desert. Many of them didn’t manage to get beyond a neat concept or, when they did, they turned out to be one-hit wonders. The introduction of processors changed things. No longer did everything have to be implemented in custom hardware and so more generic devices started appearing.

Eventually, costs started to escalate, and the number of startups dropped. While new segments emerged, few venture capitalists were willing to invest when the outcome was uncertain. At one point, in a slightly comedic keynote some years back, Wally Rhines, president and CEO of Mentor, a Siemens Business, predicted the date at which the number of design starts would not only get to zero but go negative.

But the industry has changed over the past few years. Design starts have started to climb and money is moving back into the industry. Startups are appearing. The irony is that part of the reason for this is that we reached the end of the road – the end of the road for cost effective scaling, the end of the road for processor performance, the end of the road for chip complexity.

Some of this was to be expected as part of Makimoto’s Wave, where the industry swings between more programmable devices and more specific devices – but this is slightly different. The money in the semiconductor industry was coming from a smaller and smaller set of products until we got to the point that cell phones became the dominant driver for the industry. General computing never went away, but they weren’t really trying to change the game, they just wanted more of what they had in the past.

But today, it is not one new industry. It is many that are pushing forward and they all have new requirements. Some of them, such as machine learning, are enabling technologies. Machine learning is using massive amounts of compute today, but that is just because the industry is using the technologies that are available rather than developing optimized processors for that task. That is starting to change and we are seeing many design starts in this area.

Machine learning is revitalizing other industries, such as automotive. It has enabled the vision (pun intended) for cars of the future and that has created a stampede of development, even though most of it is unlikely to succeed. Machine learning has brought about a realization of the value of data and the ways in which it can be used. This in turn means that everyone wants data and we are seeing huge amounts of development in IoT devices. No longer are these the simple sensor devices that people first imagined, but increasingly intelligent devices that not only sense but have significant amounts of local processing, and before long will probably also contain significant machine learning capabilities themselves.

Gateways are becoming smart, and many are now talking about this being the development of the “Fog” – the Cloud just a little closer to the ground. We have moved into a phase of massive proliferation of devices and technology.

So, is this a good time to be an EDA company? Possibly, but it won’t be easy. EDA companies still have to keep developing the technologies that enable the latest geometries to be implementable – and those are not small investments. They also have to look at the new demands coming from each of the new industries. Automotive is pulling them in a very different direction than machine learning, which is again very different from IoT or Cloud.

EDA is not as sexy as it used to be. There was a time when it was one of the top fields attracting the best talent, and the wages reflected that. But the best and brightest grads no longer want to be in EDA or even semiconductor – they want to be with Google and Facebook and those creating the application layers. The EDA industry is also getting old. Many of the senior folks in the industry joined at about the same time that I did, and they are still there. They are all getting close to retirement and with that a large amount of talent and experience will disappear.

Perhaps that is good news. Perhaps new blood will reinvigorate the industry. If EDA is pulled in many directions at the same time, that could create more opportunities for startups and that may make investment in the industry more attractive again. What is certain is that the industry is about to go through a period of change. It will emerge from it. It has to, because the semiconductor industry has no alternative.

I think it will emerge stronger. What do you think?



Leave a Reply