Inside The SRC

SRC’s new CEO discusses chip R&D and the challenges ahead.

popularity

Semiconductor Engineering sat down to talk with Ken Hansen, the new president and chief executive of the Semiconductor Research Corp. (SRC), a U.S.-based technology research consortium. Prior to joining the SRC in May, Hansen was vice president and chief technology officer at Freescale. What follows are excerpts of that conversation.

SE: My impression is that the SRC allocates funding for various R&D projects. Then, the funding is directed towards companies and universities. In a nutshell, what is SRC’s charter?

Hansen: We are trying to capitalize on our true core competence, which is managing and directing research.

SE: Can you describe the SRC?

Hansen: SRC today really consists of four entities. There is GRC or the Global Research Collaboration. There is STARnet. There is the Nanoelectronics Research Initiative or NRI. And then, there is the Education Alliance or EA. STARnet is a full partnership with the government. In that case, the agency is DARPA. With NRI, it’s a full partnership with primarily NIST and a little bit with NSF. All of those agencies are restricted to research being done basically in the United States. There may be a rare exception. Meanwhile, the Global Research Collaboration piece is an entity that allows for research to be done around the globe. All of our companies are global. The GRC projects are individually chosen through a formal solicitation process. In the case of STARnet and NRI, both are run by center directors that are selected from universities. It’s a little bit more of a hands-off from the member companies. The Education Alliance is all about students who are engaged in research, and we are helping them with their career.

SE: What direction do you want to take the SRC?

Hansen: Our core competence is traditional semiconductor research. We have 18 members today. Despite all of the consolidation today, there’s many more semiconductor companies out there we can engage, and will engage in, to become members of SRC. In the past, it was a bit more of a build it and we will come model. Where are we taking it now? We’ve got to go out and market our goods to attract potential member companies and bring them in that way.

SE: Have you added any new members?

Hansen: We’ve added four companies in the last 12 months. Qualcomm came in October of 2014. Analog Devices came in March of 2015. Microsoft came in June of this year. And our most recent member that joined was ARM. They joined in late September.

SE: How does the SRC work?

Hansen: We are a member-driven company. Our members provide dues. They are providing dues so they can pool their resources. Despite generally being fierce competitors in the marketplace, none of them believe they can afford to do this basic, and further out, research alone. This is a way they can leverage their funds.

SE: What is the overriding goal with the SRC?

Hansen: I like to say we solve problems for our member companies, where there is a no known solution. We look to define a problem like that. Then, we send it out to universities. They send back proposals. We select project proposals based upon that. And that’s what we fund. That becomes the input research into our all of our members to take. They can then differentiate and customize it, and develop it into their own products.

SE: What is driving some of your R&D efforts at SRC?

Hansen: There is no question that Moore’s Law is slowing down. It’s well documented. It may reach an end. So, we’ve got to find another way to extend that cost reduction and integration capability that we’ve enjoyed for a long, long time. That suggests that there is no more important time in the history of the industry to start investing in basic research to solve that problem.

SE: In the U.S., I see some troubling signs on the R&D front. First, Sematech was recently disbanded. Second, basic research is down in the U.S. And then, China and other non-U.S. nations are investing heavily in R&D. Is there an R&D crisis in the U.S.?

Hansen: The U.S., for sure, is relatively under-invested compared to other regions in the rest of world. We are working with our partner, the SIA, to address that. We put out a recent white paper, entitled “Rebooting the IT Revolution: A Call to Action.” It highlights some areas that are important. The government has reacted to that and has come out with some grand challenges around computing that are supporting that basic premise. It’s moving in the right direction, but we are still way, way under-funded.

SE: What are the roadblocks?

Hansen: The government, in a way, views the industry as being very mature and thinks the industry should support itself. What they are missing is that the complexity and where we are now to getting beyond the end of so-called Moore’s Law is much, much greater than it’s ever been before.

SE: SRC funds a multitude of R&D projects. How do you make sure you are placing the funds in the right areas and making the right bets?

Hansen: There are hundreds of device options out there. Our STARnet and NRI programs are both exploring those options. Between the two of them, they established a method for benchmarking fundamental devices. It’s a program that’s been set up to run through Georgia Tech. It’s a benchmark for both logic and memory, and also analog devices. The intent is to narrow down the field to get to a select few to bet on and then go forward. Now, there is no guarantee that this is going to pick the right horse. But it’s a methodology and process that should eliminate a lot of things.

SE: Have you narrowed the field to a select few technology options?

Hansen: In fact, our programs today are doing the exact opposite. We’ve eliminated a lot of them. But there are more devices being created or concocted every day at this point. We feel this is a good thing, because we don’t feel that we have a horse that is ready to move forward. So let’s get many ideas on the table. We probably won’t be at a phase for easily a couple more years before it gets narrowed down. It could easily go beyond a couple more years to the next seven years or so of programs.

SE: At what point does the industry have to come up with a select few options?

Hansen: We as an industry see a way to 5nm. That’s a good 10-year run at least. So as it stands now, that’s that point on when something has got to come in. Now, even at the 5nm node, there are significant changes in the way you build devices and such.

SE: What about More than Moore?

Hansen: Where we are now is that Moore’s Law is slowing down. So it requires companies to get creative in different ways. Coming from a design background, I know there are architectural and circuit things that can be done to take advantage of any given node. One of those things is definitely packaging and systems-in-a-package kind of solutions. There’s multiple ways of doing it. They all have their place. It could be stacked or wirebonded. It could be side-by-side with some clever fan-out technology.