Executive Insight: Grant Pierce

Sonics’ president and CEO sounds off about startup exit strategies, the Internet of Things, competing for highly trained systems engineers, and future opportunities.


SE: What worries you most?

Pierce: At the very top of my list is continuing to attract the talent into the company that we’ve managed to build up over time. Inside of Sonics, it’s a unique situation. We’re at the boundary between system-level architecture and chip-level micro-architecture for having what the system guys are looking for in the silicon. We need expertise in any IP that touches an on-chip network. Our people need an expertise and understanding not just for how any one IP core works, but how they interoperate through our technology. And then they need a system-level expertise to develop our products and within the ranks of those who are supporting our customers. It’s like searching for unicorns. They’re almost mythical. There aren’t very many of them, and increasingly we’re competing with a lot of semiconductor companies for those people. They need experience not only on the logic side, but also through the back end. These are very good people, and they command premiums of attention and compensation.

SE: So how do you compete for them?

Pierce: We have a really interesting position in the supply chain. Unlike at the systems companies, where those people work on one design, they can work on multiple designs here. That’s much more interesting.

SE: Is the solution to this problem training more people? Is it competing with the social media companies for the best and brightest? How do you get more of those kinds of people?

Pierce: It’s very difficult, and right now there aren’t enough of them. We try to use the culture of a small company so they can come here and have an impact immediately. That’s probably the biggest attraction. If they were to join a systems company they might have to go through some internal revolutions because a lot of system companies are having to reinvent themselves as chip designers. It’s a difficult task, and it’s a mindset where chip guys who go into these large companies are really in an alien landscape. We work with the best SoC designers on the planet, so we can take that talent and help architects design the leading SoCs. That’s pretty attractive.

SE: If you look out at the market, though, midsize companies are getting squeezed. What’s next for Sonics?

Pierce: To be the CEO of a company that has raised private equity, an exit strategy is required. Investors demand that. At the same time, I don’t believe there’s a market to go public right now. We filed for an IPO in 2007 and received approval from the SoC. We were going to be the first IPO in the semiconductor space in 2008. When the market turned down, we were at the smaller end of eligible companies to go public. And since then, going public has become much more difficult. We haven’t seen one in EDA and IP for a long time. And the size requirement has grown since then.

SE: What’s the minimum size now?

Pierce: More than $100 million. That’s where your revenues need to be.

SE: Six years ago, when you were looking to go public, it was $20 million to $40 million, right?

Pierce: Yes. Now you need $2 million a year just to be able to properly participate in that public market for your shares. Maybe the market will change, but going public is not the best thing for our shareholders.

SE: What do you consider your core value?

Pierce: The critical need is more than just the technology, the patents and the products. It extends to our methodology, our know-how for how to leverage it. It’s the team.

SE: What makes your products and your know-how really valuable?

Pierce: It’s the risk and the uncertainty of participating in what’s being identified as the largest market opportunity in the history of semiconductors—the IoT. It’s a market that demands an SoC level of integration far earlier than the companies that are creating the products have the knowledge to be sure they’re integrating the right thing. In the old days we used to build breadboards of systems and figure out what critical pieces needed to be integrated together. Maybe in the breadboard process you would try one thing, throw it out, try another, and then you’d figure out what you needed and create one product and go after the market opportunity. If you look at the early conceptual pictures for Google Glass, it was a guy wearing a belt and a backpack—his battery system and interconnect—with an army-style helmet, goggles, a cell phone duct-taped to his ear, a consumer camera system mounted on the helmet. It was a Rube Goldberg solution.

SE: It sounds like the old video game characters.

Pierce: Yes, and that could not get to a practical form factor without adopting an advanced chip that had all the different parts needed to develop a compelling prototype. But that chip originally had capabilities built in to do things like drive off-chip HD video, so if you captured video and went to your hotel room and you wanted to watch that video content, you could drive it out. You don’t need that kind of display resolution for Google Glass. It also had numerous different interfaces for going off to different networks. There was too much stuff and it burned up the battery too quickly. So what is it you really want to do? You want a means of going through this very early integration, and you need to integrate at a reasonable cost and iterate at an even better cost. But if you can’t start at a sufficient level of integration, you can’t even begin to prototype. The market opportunity is there. It’s staring us in the face. But how to get to an integrated and compelling prototype quickly, and have that prototype feed directly into a production part, is something we think about a lot.

SE: Where do you see Google Glass types of applications playing in the IoT?

Pierce: A lot of what Google Glass is good for is providing information so you don’t have to take a cell phone out of your pocket and fire up a huge display. If you don’t have to turn on that big screen or fire up an apps processor, but use it as a link service to the broader data network, there’s a lot of opportunity. How many times a day do you look at your smartphone? The typical user does that 150 times a day. If you could reduce that number, there’s a lot of extended battery life.

SE: The IoT and all its iterations play heavily into your core market, right?

Pierce: Yes, absolutely. And the future for Sonics is to look at those market opportunities and make adjustments for the needs of SoC designers who have an intellectually challenging design to do—multiprocessors, multi-heterogeneous processors, memory subsystem requirements. Most of it needs to be purpose-built. We can’t tolerate the inefficiencies for power or bandwidth that you get with one humongous processor that takes care of all the processing and bandwidth needs of the IoT. There will not be one dominant architecture out there. We won’t repeat the computer history of mainframes to PCs to tablets, or even into smartphones, because the specialization of IoT to get it to be optimal for function and user interface demand very different chips for each way we might use that device. That’s especially true when you extend the IoT into wearables. Wearables starts to introduce fashion differences. The chip itself has to be able to support things about that user interface and the way the customer interfaces to your product. All of these other things are going to come into play. Form factor is huge.

SE: So what will be the defining attributes?

Pierce: It won’t be about power, performance and area/cost alone. It will include performance per watt, but it will have to be the right performance. It can’t be wasted performance. And it won’t be the highest performance for every application. It will be the right performance given the battery you’ve got and the application you’re trying to run. The thinking is very different. At the beginning of Sonics, we were driven by the emergence of digital television driving a new type of architecture. We had to worry about digital processing rates and how you get the right bandwidth from memory to keep up with frame rates and resolution. As we move into the wearable side, you want as little chip as possible. You will have multiprocessors, but if you’re going to run it at very high performance that’s going to be a very short use time for that chip. And you’ve got to be able to dynamically fit that chip to the application’s purpose. So how you are you able to conserve power, but also take a system-level view of power consumption and safely implement that power policy as dictated by the system developer?

SE: You’re looking at this as a gatekeeper for power, right? And maybe even security.

Pierce: Yes, it will very much be about power and security. If you’re wearing a sensor that can predict you’re about to have a heart attack, who do you want to know about that? And who do you want to make sure gets that message securely with the guarantee that they got it? Those are hugely important factors, and they have nothing to do with the old metrics by which lots of companies interacted with us in the past. That know-how for applying our technology is where our future customers begin to put new kinds of demands on us. We’re not going to be in a market we we build the interconnect and they will come. We need to help them use it.

Leave a Reply

(Note: This name will be displayed publicly)