-----------------------------------------
Interview between The Register's Thomas Claburn and DARPA's Dr Dev Palmer
Transcription generated by machine and lightly edited by staff for clarity. This is provided as-is, typos and all. Please direct complaints to /dev/null.

Date: Monday, January 23, 2023
-----------------------------------------

The Register 0:08

Welcome to Government Tech Week with The Register. I'm Tom Claburn. I'm a reporter here at The Register. I'm speaking with Dr. Dev Palmer, Deputy Director of the Microsystems Technology Office at the Defense Advanced Research Projects Agency or DARPA. Dr. Palmer, thank you so much for making the time. I appreciate it.

Dr. Dev Palmer 0:25

Thank you, Tom. Glad to be here.

The Register 0:28

So I was hoping we could start by just getting a sense of what the JUMP 2.0 initiative is and what DARPA is interested in it is,

Dr. Dev Palmer 0:39

JUMP 2.0 is a government industry partnership, managed by a not for profit that sponsors microelectronics-related research at universities across the US. The public-private partnership is a great benefit to DARPA because we not only leverage each other's investments, but we can also make sure that we're investing in research that will benefit not just the military, but also the commercial semiconductor world as well. It's a little bit different from the public-private partnerships in microelectronics that DARPA has been funding for the last 25 years in that the environment for the semiconductor industry and microelectronics has shifted a little bit. And we're getting towards the end of Moore's Law. I'm sure everybody's heard of Moore's law. And so where do we go from there? The next level of consideration is more of the system level. For information and communication technologies. And that's right where JUMP is working right now. And we're aiming at a long time horizon for the research in JUMP. Five to 10 years that commercial applications will typically develop before the military. But we're looking out that far in order to make sure that we're working on research that's going to maintain our advantage in microelectronics and microsystems.

The Register 2:17

A number of the systems outlined in JUMP 2.0 talk about intelligent sensing and embedded intelligence, and things like that. How does DARPA define that for these contexts? And are you looking to develop autonomous systems? What's the goal there?

Dr. Dev Palmer 2:36

Well, the microsystems technology office, despite the fact that there's "systems" in our name, we're focused mostly on the enabling technologies and components, and that could be electronics, photonics, MEMS, some quantum sensors and also the software that helps run those systems. The line between hardware and software is getting more and more blurry as time goes on. We don't really build systems. We're trying to expand the design environment, the design space for system designers to give them more capabilities. So when we're talking about edge intelligence, a lot of people popularly know about the artificial intelligence that could be human players in Dota 2, that live in that role-playing game. What they don't really think about, if you're looking at that AI engaged in gameplay, it looks like somebody else just sitting in a laptop, but behind that is 1000s of GPUs 1000s of cycles per second, many gigabits of training data and gameplay data going back and forth. And there's no way you can pack that into a small mobile platform. So how do you get there from here? It's kind of a three-pronged problem. I believe. First, you have to start with the math. How are we defining the algorithms that mimic thought, if you will, and then how do we translate those algorithms into something that a computer can understand? And then the most important part for MTO is how do you build hardware that runs those programs optimally? So that you can maybe, you know, someday in the future, long after they've kicked me out of here, we'll have that same sort of AI capability and something that's, you know, this something that can fit into a small mobile platform within the volume and power constraints.

The Register 4:48

Among the technologies that you're looking at, what are some of the ones that you think are most promising in the next decade for advancing DARPA's mission?

Dr. Dev Palmer 4:56

That's a great segue from the previous question. Actually, if you look back, I think you could make an argument that the 1800s were the century of chemistry. The 1900s were the century of physics, that's the branch that produced the most interesting advances. The 2000s look like they're going to be the century of biology. In the same way that online search engines give you access to the entire body of knowledge of the human race in sort of a sorted and unverified fashion. That's not really an implant. That's a technology that enhances our cognitive capabilities. I think there's a possibility that we'll see that develop in terms of interface between biological systems and electronic systems over this year. It'll become as commonplace as carrying around a cell phone. Now interestingly, MTO is not currently working on any electronics that interface directly with the body. But in building the systems that we build, the microsystems that we build in MTO, we can draw a lot of inspiration from how the human brain and human sensory system processes information. For example, you take in much more visual information than your brain can process in real time. A lot of the processing happens actually in your eyeball, the retina and the optic nerve. And there's almost more information traveling from the brain, to the eyeball to control where you're focused and how you're interpreting the images, as there is going back to your brain. How can we learn to use that in an electronic system?

The Register 6:56

Given the focus now on returning to the moon and to space, how is DARPA thinking about these sorts of parallel developments in terms of making its systems more robust for extreme environments? I know that you know, of course, with the military, that's a big focus. But now it seems like the challenges are going to be even greater where you have radiation and lack of pressure to consider.

Dr. Dev Palmer 7:22

Yes, we aren't NASA and MTO doesn't work on operational systems, but we did recently introduce a new focus area in extreme environment electronics. And space is certainly an example of where you have pretty much all of the bad things, right – high temperatures, low temperatures, radiation exposure – but there are a lot of other places where high temperature is an issue. And radiation tolerance, with the proliferation of low Earth orbit satellites, I think is going to be much more of a commercial issue. The DoD requirements are a little bit different. Not going to get into that but we're still ... as part of our advanced manufacturing initiative, we're also including how you manufacture systems that can survive and all the screen extreme environments. Most people think about the extreme environments as being external to the electronics, but we're also looking at operating conditions that can produce stresses inside the electronics like high voltage, high power, those kinds of things are going to become more and more important in power distribution and electric vehicles, and other technologies of that type.

The Register 8:56

Are there common issues or challenges you hear from the you know, either members in the military or from your partners about the technology that's been worked on in terms of what they want? Are there common themes, like the batteries don't last long enough or the interfaces are not well thought out? What do you hear as feedback?

Dr. Dev Palmer 9:16

Well, as I mentioned earlier, we don't at MTO in particular don't typically build systems that go out in the field. But we do rely very heavily on our relationships with both our military customers and our commercial partners. For example, we are constantly struggling with trying to bridge the gap between not restricting fundamental research – we don't want to wreck the creativity that researchers and possibly miss a real breakthrough idea – by focusing too much on what the application is getting. But we do involve our partners on a regular basis from the beginning, from program conception through program operations. They're not going to tell us specifically what to research but they provide invaluable advice on how to vector the research to maximize the probability that it will transition into a new capability for the military.

The Register 10:30

How do you end up sort of navigating your relationship with the public and private sector and your various partners? Is your job a lot of just sort of meeting with people and hearing their concerns, things like that? What's your approach to that?

Dr. Dev Palmer 10:52

It's definitely a positive symbiotic relationship. If you look at the global market for electronics, I think I heard somewhere that it's actually surpassed $500 billion a year in terms of sales, manufacturing and sales. And the DoD is a fraction of a percent of that, I mean, will never produce electronics in the same volume as even the most common commercial product like the cell phone. So we cannot mandate particular research in the commercial sector that makes those partnerships critical. And one of the great things about JUMP 2.0 Is that we had a lot of time during the program's development to determine what the mutually beneficial research topics were, and making sure that those got through the guide into the solicitation for the program, and that we selected proposals that that had a high probability of success in those areas.

The Register 12:08

We're approaching the end of our time. I was just wondering if there's anything that we haven't touched on that you think people particularly ought to know.

Dr. Dev Palmer 12:17

Yeah, JUMP 2.0 is a great program. It covers the nation. We're working with the best and the brightest in both academia and industry. It's a part of the Electronics Resurgence Initiative 2.0. The JUMP time horizon, as I mentioned earlier, is that pretty far out there. ERI 2.0 is continuing to research on various military and commercial topics of interest from ERI 1.0 out but we've also added the extreme environment electronics and advanced microsystem manufacturing, because one of the ways that you can implement microsystems to directly contribute to information and communication technology issues, is by going into the third dimension since we're running out of space and in 2D chips. There's a lot of that already out in the commercial world, but it tends to be silicon on silicon, adding in three fives and other technologies like photonics and possibly MEMS, really expand the capabilities that you can pack in with small volume. The problem is, it's really hard to build those right now. So the big investment in ERI 2.0 is to create the capability for anybody in the academic, commercial or government research communities to prototype these 3D heterogeneously integrated systems in, and actually let, you know, let the rubber hit the road. Test whether we actually have achieved the capabilities we're looking for. In addition to the actual prototyping services, we're also spending a fair amount of time thinking about what other advanced research topics we can throw out there and that will create the capability to scale up this prototyping ripple domestic manufacturing capability for 3D systems.

The Register 14:33

I think that brings us to the end. Thank you so much for making the time, Dr. Palmer, I enjoyed speaking with you. And that's it for this segment of Government Tech Week.

Ends.