IP Market Shifts Direction

Semiconductor Engineering sat down to discuss intellectual property changes and challenges with Patrick Soheili, vice president of product management and corporate development at eSilicon; Navraj Nandra, senior director of marketing for DesignWare analog and MSIP at Synopsys; Kurt Shuler, vice president of marketing at Arteris; and Charlie Cheng, CEO of Kilopass. What follows are excerpts of that discussion. To view part one, click here.

SE: Is there a difference with developing IP at 28nm vs. 16/14nm?

Nandra: There are hidden challenges at 16/14nm. Because they’re newer nodes, they’re more complex. But even at 90, 40 and 28nm, these nodes have been refreshed for wearable markets. So older technology nodes are getting quite sophisticated, as well. They’re adding their own differentiation.

SE: What does this do for the cost of developing IP?

Cheng: By definition the market is going to pay it back. There aren’t that many choices of vendors. We are all working very hard for the road map, but it’s not like it’s a hyper-competitive, irrational market anymore. It’s not as if you have mom-and-pop shops coming in offering cut-rate pricing and customers will take a risk for now reason.

Soheili: But it is a negative growth industry. You see more of us not making money than those of us who are making money. It’s getting tougher and tougher. The investment capital required is getting higher, and the return on invested capital is getting hard to see and realize. It’s getting more difficult and more expensive. Standardization is going out the window. You can customize here and make something a little more differentiated, but it’s hard to sell the same thing to three guys competing for one market. The degrees of freedom you used to have in the market five years ago are absolutely gone.

Shuler: One thing all of us are seeing is the consolidation of the customers. There are probably going to be a few large fabless semiconductor customers left in the world several years from now. Everything is consolidating. I don’t know how many times we’ve had an existing customer get acquired by another customer. There is good and bad with that. You can proliferate inside parts of the customer you hadn’t reached before. But a lot of times one of those two design teams will get laid off or put on something else.

SE: For a long time we heard there would be a lot more IP players coming out of places like China. Has that happened?

Nandra: There is motivation coming from the Chinese government to evolve everything in-house. That makes for interesting discussion outside of China.

Cheng: For which space?

Nandra: We work in interfaces.

Soheili: We see the same. The big question mark is timing. Twenty years ago you would probably never buy a Korean TV. Five years ago you probably wouldn’t buy a Chinese TV. Today you’d buy either. The quality today is missing from the IP suppliers. There is a big difference between an IP factory and an IP design house. An IP factory has all the tools and equipment and the history and tenure to produce something that is ready for production. A small design shop cannot do the same. They have all the motivations and intentions, and one day they may become an IP factory, but they’re not that big. That’s where China Inc. is today. If you’re inside the three or four companies inside of China, you may use their IP. But if you’re outside of China, it’s hard to take that risk today. Five years from now it may be a different story.

SE: Are there enough new companies coming into the market to change the market dynamics—particularly the IoT?

Soheili: I don’t see it. There may be five data center chip companies funded, all going after the same market. They all have the same market and attributes. They’re trying to accomplish the same power/performance tradeoffs. Not all of them will succeed. There is money going into them, but not all of them will come out the other side.

Nandra: I’ve seen some activity on the mobile side. Looking at the wearable side, that comes down to wireless technology. The difference there versus a few years ago is that this wireless is targeted at very low power. There is a lot of innovation on the older technology nodes. We’re seeing a lot of innovation at 90nm. These technologies are designed for very low standby power so the batteries can survive longer. The startups we’re seeing are focused on wireless charging. There are some innovations happening there. There isn’t $50 million being invested, but there is some funding behind new ideas.

Shuler: If you look at ADAS (Advanced Driver Assistance Systems) for automotive, there are companies that began attacking the traditional automotive supply chain from the outside. At least one company has been very successful, but there are other established going after this market, too. So if you’re a startup, by definition you’re in a niche. There’s another side to this question, though. There are fabless companies that are no longer buying everything from IP vendors. We’re seeing a lot of that.

SE: But in the automotive and industrial markets, isn’t there a new opportunity to sell to companies that never bought this kind of IP in the past?

Soheili: Yes, and if you look at the home market there are some pretty sophisticated chips going into that market. There are large companies spending a ton of money on that market, and other big names are right behind.

Nandra: We talked about consolidation, but there are search engine companies giving out RFQs. The chip opportunities are moving around to non-traditional semiconductor companies.

Cheng: We are selling into different companies, but for us it’s either very early or very late in the cycle. It changes how we market and communicate. At the very end of the cycle, one of the large systems companies may go to a fab and say, ‘We need to do this function,’ and Kilopass might be the only vendor. Very early, we try to persuade them to create a socket for us, and that’s the challenge. There are a lot of non-obvious systems companies, and some of them move like molasses in a very fast market. Those are the hard sockets. It’s not just the Silicon Valley companies. It’s all the systems companies trying to put in silicon to be smarter and more aware for the Internet of Things. Some of them run on very slow cycles. But that’s where the big money is.

Shuler: The easy way to find those guys is on social media sites. All the people you work with over the years end up somewhere else. You may not know a company is creating a chip, but the relationships you have with these people provide you with a way to find out.

Soheili: There are lots of systems houses everywhere.

Cheng: Eventually the semiconductor content for all electronics systems will break the 30% barrier in the next five years. It used to be a couple percent.

SE: Hasn’t it already?

Cheng: No, 30% is a lot. You wouldn’t spend that in a car today

Soheili: In an electric car, it’s more than 100 microcontrollers. But it’s still not quite 30%.

Cheng: Right now it’s about $500 per car. The car costs more than $1,500.