Executive Insight: Kathryn Kranen

Semiconductor Engineering sat down with Kathryn Kranen, president and CEO of Jasper Design Automation, to discuss what’s changing in the semiconductor industry, why that’s happening, and what to watch out for. The interview is part of an ongoing series of in-depth interviews with top executives from all segments of the industry.

SE: What keeps you up at night?

Kranen: Figuring out ways to continue our aggressive growth. We cannot stop growing. We analyze growth by product, by customer, by sales channel. Growth is situational, too. You have to be close to the opportunity and not miss the tiniest detail, which might be in your sales approach or your feature set. The ways to grow are to add more customers, more sales channels, or more users or project groups within those customers. Revenue growth can come from multiple groups starting to use your products. Or it can be more users, or more usage per user. We talk about a revenue growth stack to illustrate all the sources of revenue. It can come from a brand new product, a brand new account, and all the hierarchies within that account and product. And if you shrink one place, it’s hard to make it up somewhere else!

SE: How does that tie back into the overall industry?

Kranen: We think about this in the context of our competitors’ product packaging and business models. Each competitor, large or small, is going to use its assets. What are the assets of the big companies that may get bundled together? We bundle our own assets together sometimes. You have to think about your products in comparison to smaller companies, which may have a sub-solution that may be sufficient for some customers, and you have to think about it in the larger scheme where the big companies might be rolling things together to do some market-share grabbing. One thing that worries me is there is too much focus on stealing market share instead of growing the pie. That’s uniquely endemic to EDA.

SE: So how do you grow the pie?

Kranen: You have to solve new problems. That’s the way EDA revenue grows. Otherwise products get commoditized and thrown into the mix. The rate of innovation in the EDA industry seems to be disappointing to the semiconductor industry, and therefore people have dumbed down their hardware plans as a result. Customers have told me they have stopped expecting too much change from EDA. The effect is platforms that are simply software variations. It’s too expensive to build hardware, so you have mega companies that do high-volume things and not enough other things. Part of that is due to physical design cost, but part of it is due to EDA’s failure to keep up with complexity, and the increasing costs of development.

SE: Isn’t that partly because the cost of developing EDA tools is so high and there are limited dollars for multiple possibilities?

Kranen: The pendulum swings a lot. It swings from customers wanting to push the envelope on their own innovations to doing the tried and true for a lower cost. There’s one large OEM you probably know that has been a big gift to the EDA industry lately. They were the strongest growing account last year. They’re doing more hardware, more ambitious hardware, and they’re considering how to do it for less engineering cost, which means more EDA. Some companies have an EDA tools budget and a separate personnel budget. When companies save on EDA tools, they hire another person; but when they save on a person they don’t buy more EDA tools. Yet the whole purpose of automation is to reduce labor costs, right? Organizationally, there’s a separation of methodology from purchasing these days. So the majority of most EDA groups’ effort is about reducing price. You don’t see as much about analysis of benefit and methodology reviews. We used to see that a lot—figuring out ways to save 20% in engineering costs through better use of EDA. Today it’s much more detached.

SE: Part of this equation involves when customers are going to move to the next process node, the next tool set. How does that affect what you do?

Kranen: All of the companies I’ve worked for have been geared toward exciting new domains in the front-end, so not so much about changing process nodes. We’ve never been selling a commodity product. When they’re new, customers don’t even know what a product can save them. The big EDA companies don’t seem to compete on products. Their solutions are all balled together into a transactional sale. There are not many specific separate tool decisions being made anymore. With $45 million paid to ‘vendor A’ and $60 million paid to ‘vendor B,’ customers don’t necessarily know which products carried the load. It’s difficult to compete on a single slice. But this complexity of understanding what are you selling to whom, in what context, against which competitor—it’s important to do all of that.

SE: How will Jasper position itself going forward? EDA company? IP vendor? Verification company?

Kranen: We’re an EDA company. We’ve always been an EDA company. Our current solutions are almost all for functional verification. But we do have an architectural side. You will see more solutions that are different. Security path verification is not functional verification per se, but that’s one of our solutions this year. There are groups that do nothing but try to break a design. In the past they didn’t have tools to do this. It’s very likely we will have other front-end tools in the future, besides verification. I don’t see a limitation on growing into adjacent domains.

SE: If you had one decision you could re-do, what would it be?

Kranen: I have never had a five-year plan. I’m a change agent. I will change things until I’m satisfied. And I love challenges, even self-imposed challenges where we decide to go for something new. I can think about horrible times in my career, but I’m not sure I would change them. I’ve learned from those times. If there are more challenges at work, I do more outside of work. They both go up together and that tends to keep me sane. It’s so easy to have new challenges in this industry. When there are no challenges coming from the outside, I’ll create them inside.

SE: Any worries that Jasper can remain independent?

Kranen: I would hate the thought that Jasper gets acquired and the product dies or isn’t successful or the employees are not taken care of. But if and when an acquisition ever happens, we’re way past that point. The same was true for the past two companies I worked for, Verisity and Quickturn. Although I wasn’t there by the time those companies got acquired, their legacies were well established. Verisity created the market for constrained random simulation, a major pillar of modern verification. Quickturn pioneered emulation, another pillar of verification to this day. Jasper has already evolved formal verification to that same point, and I believe formal will dominate verification in the future. But in answer to your question, if you want to remain independent the answer is to grow, grow, grow—and that’s what we’re doing. If you don’t have a ‘tired’ investor that forces a sale, as long as you keep growing investors will stay with you. There are a lot of distress sales in EDA where executives are out pitching the company. We’re fortunate to not be in that position. But every quarter there are conversations about us acquiring some other company or being acquired. It’s not something we have to do, and that’s what’s important. No one is bored here. We’re doing so many things and it changes so fast that I can’t think of anything I’d be doing that would be more fun.

SE: One of the industry’s big challenges is complexity. It’s hard to explain to other people. How do you change that?

Kranen: You narrow it down to something that’s important to the audience and explain it that way. If you consider the incredible rate of change for all these electronics, we make the software that helps the engineers who design those products get their job done. It’s the same as explaining what engineers do.

SE: How about to Wall Street?

Kranen: EDA is a great place to be in uncertain times. It’s far more stable than other tech sectors. We don’t have the horrible dips of the semiconductor industry. When there’s a dip, we are usually flat or might even grow a little. We enjoy a higher percentage of the upside than the downside, if you’re looking for something with a better degree of safety. But the industry problems are investments. It’s really, really hard to get investments in EDA startups anymore. Typical VCs have turned away from EDA because the outcomes aren’t big enough and it takes too many years. I see angel investors, but they can only do so much, and most of them are getting older.

SE: Do you see that changing?

Kranen: If you look at customer behavior, EDA behavior and investor behavior, we have a really bad combination right now. The EDA customers have gotten large and very experienced in their EDA cost optimizations. Everything is about cost optimization. In reality, we in EDA sell net cost savings, and customers wouldn’t buy what we’re selling otherwise. But some customers are doing vendor switching to reduce the balance of power and cut their costs as much as they can. They complain there isn’t enough innovation, yet the system is not set up to reward innovation. They’re not paying for innovation; they’re paying for cheaper EDA tools. And the dominant amount of the budget goes to the big companies. So who is going to fund the innovation? It’s not the universities. It’s not internal. So how do you take a nascent technology and make it mainstream? There are government grants in the nascent technologies, but that’s not what it takes to make something usable. How do you get people to switch to technologies that you can’t guarantee are going to work yet? There are a lot of steps of incremental technology and market psychology innovation that have to happen, and you have to do them in a spiral. There’s not much going on in terms of how to fund innovation to get nascent technologies to mainstream use. That’s what EDA startups historically have done. If we’re not funding EDA startups, how is that going to happen?

SE: Isn’t there also a question about what goes mainstream going forward? Some companies may be at the end of their process scaling.

Kranen: Moore’s Law is an interesting metric after you’ve developed something. Nobody set out to design a device that has twice as many transistors. They said they wanted to design new functionality. To do that you need more complexity. It’s an after-the-fact metric. You do it in the most economical way. It’s pragmatic, and this metric of Moore’s Law is about a monolithic device. The fact that technology is compartmentalized into IP blocks and the business models are different doesn’t limit the growth in complexity. Maybe people look at it too much on the transistor size. It’s not that the cost is going up because of finFETs and power.

SE: Will the lack of EDA startups hurt the industry going forward?

Kranen: It will certainly hurt the top-line growth numbers of big EDA companies if there are no available sources of inorganic revenue from acquisition, and that might hurt EDA stocks. Lack of startups would create a drought in certain types of EDA innovation. But I suspect something will change. Either companies individually or as a coalition will fund technology to solve problems, or big companies are going to see economic benefits of building incubators inside their own companies—or customers may start doing it in-house again.

SE: What are your big challenges going forward?

Kranen: Sorting the many opportunities and deciding which ones to put our resources into. In the last few years, I am much more top-down focused than I have been in my career. The alternative is pushing the rock up the hill. Of all the things we can imagine doing, which ones do we want to make a serious attempt at? You have to court and nurture these opportunities.

SE: How do you choose?

Kranen: The best way is to validate these with money. Which opportunity is going to pay us how much, by when? And is that sufficient? Can we deliver all of the pieces that would be required for customers to buy? If they’re not going to buy it, or if only 8% are going to buy it, it’s not right. You need to analyze where all these opportunities are going. You have to have the experience and the discipline to look at opportunities in terms of which are real, concrete opportunities. That’s why I rarely look out more than a year. We’re doing a few things that won’t manifest themselves until beyond that point, but for those you need a strong customer commitment. You need validation mechanisms for these opportunities that are painful enough to show people will fight for them. If the customer has to pay a significant amount of money, you know for sure they really do want it and that they perceive high ROI. Talk is cheap. Too many customers say build it and we will buy it. Prove it. You can’t make a plan on “hope”. You need to test, validate, and deliver things incrementally.

SE: Who’s going to lead those changes?

Kranen: There are some companies that are willing to do things that are forward-looking, and there are others that wouldn’t even think about it. The industry goes through these cycles of high innovation followed by lulls. On the front end, it looks like we’re coming out of one of those lulls, because the pain is incredibly high. What can you do to offload the simulator? It’s a constant conversation between tool vendors and their customers. I hope we’re going to return to the days of very tight vision-led partnerships between customers and individual suppliers, or collections of suppliers, where we both put skin in the game and go make good things happen. Customers have to be willing to invest in innovation. Of course, you have to be a really reliable partner so you don’t compromise their go-to-market plan, but we used to see a lot of that back in the 1990s and 1980s. Maybe we will see a return to successful technology partnerships fueling EDA innovation.