Startup Challenges In A Changing EDA World

The Electronic Design Automation (EDA) industry is a mature industry, but it’s also one that is constantly changing. Each process node and packaging technology advancement places new demands and constraints on existing tools. In addition, changing design problems and paradigms transform how design teams operate, and the goals they target.

For a relatively small industry, EDA requires a disproportionally large amount of investment in research and development. In the industry’s early days, much of that was done within universities, or startup companies. But during the lean years of the 1990s and 2000s, when design starts were falling rapidly, the industry collapsed down to a few large players. Almost all startups either perished or were acquired. The innovation ecosystem disappeared.

Also, during that period, the problem statement started to evolve. “EDA was a mature business 20 years ago when it was 40 years old, says Steve Roddy, chief marketing officer at Quadric. “Now it gets the senior discount at most restaurants. The difference between 20 years ago and now is a reflection of the complexity and interconnectedness of the EDA tools themselves. Two decades ago, users could pick and choose best of breed ‘point tools’ for each major step of the design process — at least in the complex digital SoC domain.”

As the semiconductor industry again changed, it now requires more specialization, and that has resulted in a turnaround for design starts. The most complex designs also demand more circuitry than can fit on a single die, resulting in a major change to the manufacturing process. As the problem space shifted, the industry found itself unable to adequately respond. That, in turn, spurred a lot more internal innovation, which still has not filled the void completely.

In other industries, this would have been solved by new startups emerging. But those industries do not have the same cost of failure. If a chip fails, the financial and opportunity costs can be huge. “The semiconductor industry is conservative because of the large amount of capital spending that is required to bring a new chip or product to market,” says Ninad Huilgul, founder and CEO of Innergy Systems. “Penalties for making mistakes are high. This leads to cautious decision-making by stakeholders, and that can mean longer sales cycles for new tools. This means that products are sticky, and it’s much more difficult for alternative solutions to be considered.”

This is a problem for both startups and established players. Established players are more inclined to focus on incremental innovation that is easier to deploy. They invest in areas with the biggest total addressable market and quickest time to profit. “Core EDA technologies receive massive amounts of development focus from the major EDA companies, and this makes them hard to address at the startup level,” says Paul Brownlee, vice president of technology solution sales at Siemens EDA. “Startups traditionally have focused on niche and new development applications that are complementary to these core EDA technologies.”

But that doubles the problem for startups. “Startups must navigate not only the technical challenges, but also the need to convince conservative industry players to adopt new solutions, says Badru Garwala, co-founder and CEO for Rise Design Automation. “This can be a slow and arduous process.”

The odds for independent success are stacked against startups. “Most IC and systems companies limit how invested and dependent they become on EDA startups,” says Siemens’ Brownlee. “This means that many startups are only viable to scale after being acquired by a large player.”

But that doesn’t diminish their importance. “If you look at the whole ecosystem — academia, startup, midsize company, large EDA company and, semiconductor company — we each play a different role on the team, and each role is critical,” says Rob Knoth, group director for strategy and new ventures at Cadence. “Startups are really critical because they traditionally take different risks, with a seriousness that none of those other players on that team can.”

That impacts everybody. “Fewer successes in EDA startups equals less involvement from would-be entrepreneurs and less innovation,” says Innergy’s Huilgul. “Large EDA companies get impacted by having to rely much more on organically grown solutions. This affects their valuation, as well, since Wall Street is looking for faster revenue growth, driven by new, innovative solutions.”

Today, there are several changes occurring. The first is a new focus on the system instead of the silicon. The systems market never had a closed tools market. Second, more specialized designs likely will result in the tools market becoming more fragmented, giving startups an edge over the larger EDA companies that need to see a larger market opportunity before they will invest.

The innovation chain
University research, once a fertile ground for EDA innovation, also has seen a decline in its influence on the industry. “This shift is partly due to reduced funding and the increased complexity and time of translating academic research into viable commercial products,” says Rise’s Garwala. “The diminishing connection between academic research and startups may be contributing to the current innovation gap. In addition, we are seeing that the new talent pool coming out of academia is going into software, which leverages open source to innovate, and it has a larger potential market where acquisition targets/valuations are very high.”

To go from university to startup takes money. “Consider the ever-larger upfront technical investment required to break into the EDA business against a backdrop of the overall size of the EDA business, which has been growing nicely, but still is only single-digit billions in total revenue,” says Quadric’s Roddy. “How many venture capitalists are willing to fund an entrepreneur with hundreds of millions of $ of startup capital only to potentially win a small slice of sub-$10 billion market. Plus, the exit path only consists of selling to one of three big industry titans. Alternatively, there is an endless number of SaaS and AI companies asking for the same VC dollars, pitching fever dreams of rapid ramps to billion-dollar-plus revenue potential in markets worth $100 billion or more. Starting an EDA company in 2024 is indeed a tough hill to climb.”

Innergy’s Huilgul concurs. “Investors are very wary about investing in EDA, even today, after hearing for the past few years about how semiconductors are cool again. The reasons cited are that EDA is a small market, the sales cycles are long, and it is not thought of as sexy. Thus, raising money is difficult in the EDA space, which means startups really struggle to take off. While market validation is needed in any industry, in EDA it is especially important to get customer traction.”

Startups coming out of universities face another huge hurdle. “In terms of sales and marketing, industry veterans have been talking with leaders for decades,” says Allan Klinck, co-founder of Rise. “We know the problems, we know the people who have the problem, we know the solution, we know the changes in the industry. If you do not have an existing network, that’s going to be very difficult. Even if you have a good idea, getting that deployed into big corporations is very difficult. It must be close to impossible.”

But there are enough changes underway in semiconductor designs to open up some new opportunities. “The market has shifted with 3D-IC,” says Marc Swinnen, director of product marketing at Ansys. “There is a move into the systems area, creating a lot of ferment and new capabilities. 3D-IC has opened up a lot more areas that were traditionally outside of EDA, such as a Maxwell’s equation solver. Once only used for antenna and radar designs, it is now being used for analyzing chips. It is being used with interposers and 3D-IC, doing the electromagnetic analysis of large interposer structures. This has opened up new areas of research at universities.”

It also may be possible to provide the universities with frameworks into which they can innovate. “We are fostering a collaborative ecosystem through our high-level synthesis and open Intermediate Representation (IR),” says Garwala. “This allows users and partners from various sectors — including academia, startups, and established companies — to experiment and validate their ideas in real-world scenarios. They can explore adding design features, developing new domain-specific languages, or even generating IR from their own tools to fully realize the benefits of hardware implementation. This approach enables them to focus on their specific areas of expertise while accelerating the innovation process.”

Still, while this may enable more startups, it only fixes part of the problem. They still need to find a way to grow to a sufficient size to have a respectable value. “The size of the top three EDA companies certainly has made it difficult for startups to generate enough revenue to be meaningful in terms of the larger picture,” says Siemens’ Brownlee. “Startups need to focus on adjacencies and compelling gaps, rather than building a better mouse trap. Large EDA companies still do a make-versus-buy decision when it comes to innovation, and decisions to buy have tended to involve proven startups in areas outside the established EDA company’s field of core competence.”

New problems, new opportunities
Nonetheless, change creates opportunities. “You need to solve a problem that the big players have not solved yet or have a unique technology that they do not have,” says Garwala. “You also need to be able to bring it to a customer that has established flows and processes, which your technology must both fit into and improve. Ideally, it should be something that needs to be solved to drive growth within the semiconductor industry.”

Invariably, that means identifying a weakness. “Startups only make sense if they do something different or better than existing solutions,” says Michiel Ligthart, COO at Verific Design Automation. “They need to be disruptive. Recently, international trade relations were a driver, and we have seen a sudden increase in EDA startups in China over the past five years. Competition also spurs innovation.”

Innovation is happening in places different than the past. “The AI revolution is re-injecting some of that magic that we’ve lost, and creating the ground for new startups,” says Cadence’s Knoth. “I see this happening at some of our large partners, that are really pushing the edge of what AI can do in semiconductors. Some of these commercial companies are bringing on researchers, bringing on more startup-esk ideas. It doesn’t remove the need for a startup. There’s still an important cauldron that a startup forges in an engineer. Just embedding them inside an EDA company or a semiconductor company won’t replicate that, but it does help with that issue.”

Startups tend to be more flexible. “A startup can be extremely nimble and can develop and incorporate new technologies, creating new architectures much faster,” says Garwala. “This ability to be nimble, fast, and solve complex challenges is unique, and often requires the environment of a startup.”

However, expectations may need to be reset. “With digital design, everybody agrees on the same way of doing things, but for analog there’s a huge range of things that need to be done,” says Ansys’ Swinnen. “There is such a broad range, it’s impossible for one company to have all the bits. There is this whole ecosystem, little mom-and-pop stores that develop small utilities for analyzing Smith charts or something. It takes one or two developers to develop, and they can sell that to all the RF guys, and they can make a nice little business. I assume Cadence is not interested in getting into that tiny market, but they can service their customers through (Cadence’s) SKILL, where you can get these utilities and you can integrate them into the tool. That shows that such a system can and does exist in EDA, made possible through SKILL.”

Interfaces make a lot of things possible. “Interfaces are important for innovation, but interfaces slow things down,” says Knoth. “Interfaces are lossy. There is an accuracy penalty, a speed penalty, all sorts of things, but they are still critical. But we shouldn’t wish for the past to return. We should look where we are and look ahead, use the past as a guide.”

We rely on universities for innovation. “We need to reignite this effort of cooperation and growth in the universities to help with the serious shortage of new talent we are seeing,” says Garwala. “The industry must explore new ways to foster innovation that includes both universities and startups. Building partnerships between large EDA companies, startups, and academia could reignite the innovation engine. By fostering a collaborative ecosystem, instead of isolated, proprietary ones, startups can gain the resources and expertise needed to overcome entry barriers, while established companies can access novel technologies.”

The semiconductor industry also relies on startups for talent. “A vibrant startup ecosystem is critical for producing the right humans to change the world,” says Knoth. “I’ve been fortunate enough to work with some really inspirational, talented humans, and they wouldn’t have been the same humans without that startup experience. It’s incredibly important for our health as an industry, not just EDA specifically, but semiconductors and systems design broadly. Without a vibrant startup ecosystem, we won’t have the quality of humans to go forth into the world. It is really important to nurture that. It doesn’t mean the startup ecosystem has to look like it did in the ’90s, where you had little place-and-route startups. It doesn’t have to be that, but it has to be some new form of a vibrant startup ecosystem. Our goal should be to ensure there are enough startups doing interesting work, producing interesting people, producing useful products that can help us solve new problems.”