Executive Briefing: Lip-Bu Tan

By Ed Sperling
LPHP: From a high level, what are your customers doing differently these days?
Tan: What system companies are looking for is time-to-market and differentiation. They want to differentiate on specific functions. IT has become very important in this process. And in terms of tools, they are looking for end-to-end solutions. Besides the advanced nodes and IP blocks, they are starting to move into software stacks and all the way to the package. This is why packaging is becoming very important. Last July we acquired Sigrity to address the power signal integrity for the packaging. If you wait until you design it to know what it looks like, then you have to redesign it all over again. So if you look at trends, system companies are dealing with much more complexity. On top of that they will have new challenges when they move down to 10nm and 8nm because EUV is still not there.

LPHP: Why is Cadence suddenly so focused on IP?
Tan: Just like with EDA 20 years ago, big customers are now willing to outsource if you have good quality and proven IP. In time, they’ll be willing to outsource software, too, because it will become more and more complex and non-differentiating for them. EDA is becoming more critical. We are helping our customers solve problems, not just selling them EDA tools. With IP you have to qualify it, optimize it and then you have to integrate it. Hardware-software co-design and co-verification are becoming more critical. It’s very much application-driven.

LPHP: How does fragmentation of the industry, and the resulting need for vertical market approaches, affect EDA and IP?
Tan: It’s not just pure sales anymore. You need to focus on the customer’s customer and the trends there, what’s needed to differentiate products, and then the software that’s required. It’s a lot more complex than just selling a tool. We need to know what the customer’s challenges and needs are, and then tailor differentiation. More and more it’s providing a solution. That’s where we see IP playing. Low power is becoming more critical, too. We are working with customers to fine-tune the power. That’s where the Big Three can provide optimal solutions to provide differentiation.

LPHP: In the past, it was very difficult to modify ASICs for vertical markets. Are we moving to an era of platforms?
Tan: The key thing that attracted us to Tensilica is the ability to customize ICs for a specific function. They’re also addressing some of the performance requirements and they can drive differentiation for customers. They even have an off-the-shelf generator for customization. That’s why 7 of the top 10 semiconductor companies are using Tensilica. By integrating with our tools, we can really optimize performance and differentiation that our customers want to provide.

LPHP: Cadence has staked a claim in IP markets and, like Synopsys, it has tools to develop new IP. How will this play out in the market with partners like ARM?
Tan: The good news is this also is a growing market. We don’t have to eat each other’s lunch. We can go together to customers with other IP companies to see what they need. We are trying to avoid competition, and ultimately it will be a customer win for both of us. With anything we do there will be some conflict and overlap, but the key is to be up front with what we’re doing and communicate.
products.

LPHP: So how big can the EDA pie become?
Tan: We shouldn’t let people think EDA is only $5 billion or $6 billion. The semiconductor industry is $300 billion, and if EDA is so critical that they depend on our tools, that number seems low to me. First, we really need some discipline on our value proposition. And second, if you look at all the software companies, how many of them have generated more cash than Synopsys and Cadence? And we’re trading at 2.5 times. So the question is how can we grow the pie from EDA to IP to software? There has to be a bigger pie for the Big Three to share. But we also have to do a little better job in providing value to our customers. If you’re not growing, people are asking for dividends. There has to be a return one way or the other. But if you can continue to show growth—and I believe we can—if you look at the multiple of 2.4 or 2.5, that’s really undervalued. We need to grow the semiconductor industry from $300 billion to $500 billion over the next 8 to 10 years, and grow the EDA pie from services, outsourcing, IP, software and beyond that to optimized solutions for the customers.

LPHP: Where do you see the big opportunities for EDA?
Tan: If you look at the trends—big data, cloud, mobile—low power is critical. So are massive parallelism and high-speed connectivity with networks that address latency and scalability. With cloud and big data, connectivity is not stopping. The market will be at least 5 to 10 times more than it is today. There is huge growth potential.

LPHP: EDA has always been good at reducing pain points. Where are the bottlenecks for the future?
Tan: Memory is one bottleneck. So is verification. Chips are so complex these days that it’s difficult to verify them. That’s why Palladium is doing so well. There is also pain with power and signal integrity. Right now a lot of SoCs combine analog with digital. Anything we can do to make that easier for the customer will help—and if we’re successful they’re going to buy more from us.