Experts At The Table: The Business Of IP
By Ed Sperling
Low-Power/High-Performance Engineering sat down to discuss IP supply chain issues with Jim Hogan, an independent VC; Jack Brown, senior vice president at Sonics; Mike Gianfagna, vice president of marketing at Atrenta; Paul Hollingworth, vice president of strategic accounts at eSilicon, and Warren Savage, CEO of IPextreme. What follows are excerpts of that conversation.
LPHP: There are fewer IP companies these days providing more IP for fewer design starts. Will this trend continue?
Savage: There has been a big evolution to companies like Synopsys and ARM, and there’s going to be a continuous evolution of big companies acquiring smaller companies and then shedding IP people who innovate and do new things. There will be a cycle of innovation, consolidation and disaggregation that will occur over and over again.
Hollingworth: If you look at what’s happening in the industry, a big source of innovation has dried up. The VCs are not funding semiconductor startups anymore. There is a need for some mechanism to restart that cycle of innovation. We’re an ASIC integrator that takes IP from multiple different sources and stitches it together, and what we’ve seen is that the number of design starts has fallen. Now it’s more important that you win these really important designs that go into high volume. As a result, you’ll see an emergence of new business models. The IP guys will place more focus on the royalty aspect of IP rather than just the up-front licensing cost. And second, I’m a big believer in 2.5D and 3D packaging. A big benefit there is the ability to take heterogeneous arrays of chips, which might be in completely different process technologies—a 130nm PHY next to a 28nm microprocessor subsystem, for example. That’s an easy way for IP companies to become chip or tile vendors.
Gianfagna: Third-party IP has to survive and to grow. Nobody is going to design everything from scratch anymore. You need pre-designed blocks. The question is where does it come from and who’s going to source it. Warren brought up a really important issue involving small versus large companies. If everyone only trusts the big guys, because they can stand behind it better and because they have better methodology, that will slow down the industry. The innovative guys won’t be able to get traction. Somewhere along the way there has to be a common methodology and language we can use to describe the deliverables. Right now, all you have is the marketing machines of Synopsys and ARM. Little guys have difficulty there. We need some kind of industry analysis of what IP is and what the deliverables are and how good there are, there will be a bigger opportunity for smaller companies to participate. That will make it a more vibrant industry. There’s a lot of work going on with the GSA, Si2 and by companies like us and TSMC. We hope somebody succeeds with this.
LPHP: So is money going into this industry at the rate it used to? Will there be more startups?
Hogan: You need to separate the hardware from the software. Let’s use the iPhone as an example. If you look at the value in that value stream, the one taking the value out of that stream is Apple. The BOM is $200. They sell it for $500. That’s $300 in profit per phone. The other guys making all the money are the guys making wafers. If you look at the value for what they do, the manufacturing guys generally don’t see those kinds of margins. They’re capturing a lot of value. The guy in between is not being recognized. You can put $120 million in an IP company, but the return on investment isn’t what it’s going to be for a software company developing an app. Three kids can build an app and make $6 million. That’s increasingly difficult for institutional investors to do. So here’s what will happen, and it’s a lot like the music business. If you go back 10 years, there was no iTunes. There were still big music companies like Atlantic Records being distributed through Tower Records. Today none of that exists. There is more music and more artists and more new songs, and the consumer benefits from that. We need to figure out a way to integrate easily on a silicon substrate for an SoC that’s focused on a device running apps. That’s where the value will be captured. We can disrupt what’s there today if we’re clever. There will have to be major innovation in the business model at that level. It’s cracking that value stream and then distributing it differently.
Browne: It’s Darwinian. Companies grow, they shrink, and they go on. The complexity we’re dealing with is increasing. We’re talking about a handful of transistors which, if they don’t work right, leaves you with a useless brick. What the IP companies need to do is become better at marketing so it’s easier to understand the value. And it has to be powerful enough so that a user who doesn’t understand a particular subject can still use the IP and get world-class results. We have to do a lot more work. The products that are in the industry today require you to be an expert. How many ARM cores do you need for a better iPhone experience? How do you get two or three guys in a room and figure that out instead of 200 people? Speed of execution is more important. Guys don’t have time to do IP discovery. By the time they finish taping out one project they’re already working on the architecture for the next project. You can’t have a team of 300 people in India sitting around for a week while you figure out how you’re going to win the browser benchmark. There’s a lot more intelligence needed, a lot more expert support help, and not all of us are up to it. The guys who do it first will have an advantage.
LPHP: As the industry consolidates, do we lose some innovation? Do we get vanilla solutions?
Hogan: This is a conversation I had with a design manager of a large cell phone company. He said he can buy the same IP that Apple can, use the tool chain of IP suppliers, and if his team does a really good job they’ll be just as good as Apple. But they have to find a way to differentiate. So where is that differentiation going to occur? In the hardware, it has to occur in the IP. You can do some software things, but hardware is what’s going to make it run fast using less power and cost less. If you need a graphics engine and go to Imagination then you’re no better than anyone else. But is there a five-person shop in Brazil that can build a better graphics engine for less money and license it to me for half the price? That’s differentiation and there’s always room for that.
Hollingworth: According to the Linley Group, Apple has spent $500 million on the A6 chip as a result of buying Intrinsity, PA-Semi, plus the $100 million it took to get the chip itself. In round numbers they believe that’s about $500 million. You look at that and say, ‘How can they ever recoup that investment?’ But if it allowed Apple to come out with a chip three months earlier than Samsung, that was worth $25 billion. Still, that’s not really representative of the industry as a whole. There are between 2,000 and 2,500 new ASICs designed each year and one iPhone chip. I fear for any business model that tries to emulate that.
Savage: I disagree. That’s exactly the paradigm for the future. Differentiated semiconductor suppliers that can be the companies that will invest in that level of differentiation will drive the rest of the market into an ‘almost-as-good’ solution using commodity products. There’s going to be leapfrogging of companies that will perform that level of investment to achieve that No. 1 slot in the market, where all the margin is, and No. 2 and No. 3 will be battling with cheaper technology but also with lower margins. Apple could be a glimpse of the future with differentiated technology investment that won’t take just three months to catch up with, but three to five years.
Hollingworth: But if you have just one guy doing the differentiation and everyone else copying, that’s a nightmarish scenario. I would hope that would make things ripe for someone coming along with a different way of getting a cell phone out six months earlier of that has features that become a killer app, or whether it’s a completely different part of the market.
Gianfagna: But that happens. How many cell phones have touch screens? Apple was first, and everyone followed like lemmings. There are innovations that occur. Can they be more frequent? That’s an interesting discussion point. That would drive higher growth. If IP is going to be $1.5 billion to $3 billion, how do we get that up to $10 billion?
Browne: In the IP business, we’re doing a technology transfer to our customers. We’re saying we have domain knowledge in this space that we’re going to move to our customers. If we end up in second place, then fighting on price becomes tough. But if you say your role is to bring innovation to your customers, what domain are you addressing, what are the adjacent domains, and how can you scale that? We recognize that companies get IP from lots of places. Our goal is to make it possible to use IP from anywhere at any time and make it work. I’ve been asked how an engineer is going to use a bus interface he’s never seen before. With OCP he needs 5 gigabits, and he needs 10 gigabits to get to AXI. You get one benefit with OCP and another with AXI. But you also have to think ahead and ask, ‘Will you buy more IP from the same company next time?’ And the other side is how you tie to the software. More and more of the investment cost is software. How do you make users 100% successful and make the product bulletproof? How do you transfer domain knowledge and how do you scale? It has to be easy to use so you can go from 1 to 100 customers.
Tags: Apple, Atrenta, eSilicon, IP, IPExtreme, iPhone 5, Sonics, supply chain, Tela Ventures, venture capital