As a leading venture capitalist in the electronics technology, as well as CEO of Cadence, Lip-Bu Tan has unique insights into ongoing changes that will impact EDA providers and users. Tan shared some of those insights in a “fireside chat” with Ed Sperling, editor in chief of Semiconductor Engineering, at the Design Automation Conference (DAC 2015) on June 9.
Topics of this discussion included industry consolidation, the need for more talent and more startups, Internet of Things (IoT) opportunities and challenges, the shift from ICs to full product development, and the challenges of advanced nodes. Following are some excerpts from this conversation, held at the DAC Pavilion theater on the exhibit floor.
Ed Sperling (left) and Lip-Bu Tan (right) discuss trends in semiconductors and EDA
Q: As you look out over the semiconductor and EDA industries these days, what worries you most?
Tan: At the top of my list is all the consolidation that is going on. Secondly, chip design complexity is increasing substantially. Time-to-market pressure is growing and advanced nodes have challenges.
The other thing I worry about is that we need to have more startups. There’s a lot of innovation that needs to happen. And this industry needs more top talent. At Cadence, we have a program to recruit over 10% of new hires every year from college graduates. We need new blood and new ideas.
Q: EDA vendors were acquiring companies for many years, but now the startups are pretty much gone. Where does the next wave of innovation come from?
Tan: I’ve been an EDA CEO for the last seven years and I really enjoy it because so much innovation is needed. System providers have very big challenges and very different needs. You have to find the opportunities and go out and provide the solutions.
The opportunities are not just in basic tools. Massive parallelism is critical, and the power challenge is huge. Time to market is critical, and for the IoT companies, cost is going to be critical. If you want to take on some good engineering challenges, this is the most exciting time.
Q: You live two lives—you’re a CEO but you’re also an investor. Where are the investments going these days and where are we likely to see new startups?
Tan: Clearly everybody is chasing the IoT. There is a lot of opportunity in the cloud, in the data center. Also, I’m a big believer in video, so I back companies that are video related. I’ve invested in DJI, the largest drone company, because there are so many applications for drones. Another big area is automotive. ADAS [Advanced Driver Assistance Systems] is a tremendous opportunity.
These companies can help us understand how the industry is transforming, and then we can provide solutions, either in terms of IP, tools, or the PCB. Then we need to connect from the system level down to semiconductors. I think it’s a different way to design.
Q: What happens as we start moving from companies looking to design a semiconductor to system companies who are doing things from the perspective that we have this purpose for our software?
Tan: We are extending from EDA to what we call system design enablement, and we are becoming more application driven. The application at the system level will drive the silicon design. We need to help companies look at the whole system including the power envelope and signal integrity. You don’t want to be in a position where you design a chip all the way to fabrication and then find the power is too high.
We help the customers with hardware/software co-design and co-verification. We have a design suite and a verification suite that can provide customers with high-level abstractions, as well as verify IP blocks at the system level. Then we can break things down to the component level with system constraints in mind, and drive power-aware, system-aware design.
We are starting to move into vertical markets. For example, medical is a tremendous opportunity.
Q: How does this approach change what you provide to customers?
Tan: Every year I spend time meeting with customers. I think it is very important to understand what they are trying to design, and it is also important to know the customer’s customer requirements. We might say, “Wait a minute, for this design you may want to think about power or the library you’re using.” We help them understand what foundry they should use and what process they should use. They don’t view me as a vendor—they view me as a partner.
We also work very closely with our IP and foundry partners. We work as one team—the ultimate goal is customer success.
Q: Is everybody going to say, FinFETs are beautiful, we’re going to go down to 10nm or 7nm—or is it a smaller number of companies who will continue down that path?
Tan: Some of the analog/mixed-signal companies don’t need to go that far. We love those customers—we have close to 50% of that business. But we also have customers in the graphics or processor area who are really pushing the envelope, and need to be in 16nm, 14nm, or 10nm. We work very closely with those guys to make sure they can go into FinFETs.
We always want to work with the customer to make sure they have a first-time silicon success. If you have to do a re-spin, you miss the opportunity and it’s very costly.
Q: There’s a new market that is starting to explode—IoT. How real is that world to you? Everyone talks about large numbers, but is it showing up in terms of tools?
Tan: Everybody is talking about huge profits, but a lot of the time I think it is just connecting old devices that you have. Billions of units, absolutely yes, but if you look close enough the silicon percentage of that revenue is very tiny. A lot of the profit is on the service side. So you really need to look at the service killer app you are trying to provide.
What’s most important to us in the IoT market is the IP business. That’s why we bought Tensilica—it’s programmable, so you can find the killer app more quickly. The other challenges are time to market, low power, and low cost.
Q: Where is system design enablement going? Does it expand outside the traditional market for EDA?
Tan: It’s not just about tools. IP is now 11% of our revenue. At the PCB level, we acquired a company called Sigrity, and through that we are able to drive system analysis for power, signal integrity, and thermal. And then we look at some of the verticals and provide modeling all the way from the system level to the component level. We make sure that we provide a solution to the end customer, rather than something piecemeal.
Q: What do you think DAC will look like in five years?
Tan: It’s getting smaller. We need to see more startups and innovative IP solutions. I saw a few here this year, and that’s good. We need to encourage small startups.
Q: Where do we get the people to pull this off? I don’t see too many people coming into EDA.
Tan: I talk to a lot of university students, and I tell them that this small industry is a gold mine. A lot of innovation is needed. We need them to come in [to EDA] rather than join Google or Facebook. Those are great companies, but there is a lot of fundamental physical innovation we need.
Richard Goering
Related Blog Posts
- Gary Smith at DAC 2015: How EDA Can Expand Into New Directions
- DAC 2015: Google Smart Contact Lens Project Stretches Limits of IC Design
- Q&A with Nimish Modi: Going Beyond Traditional EDA