From the Crows' Nest

Ken Miller (00:10):
Welcome to From the Crows' Nest, a podcast on Electromagnetic Spectrum Operations or EMSO. I'm your host, Ken Miller, Director of Advocacy and Outreach for the Association of Old Crows. You can follow me and the show on Twitter @ftcnhost. Thanks for listening. In this episode of From the Crows' Nest, we return for part two of my conversation with the distinguished team from Draper on microelectronic supply chain. Part one of our conversation was released on January 18th. We focused a lot on some of the big picture concepts and important policy developments. And in this episode, we look more closely at technology innovation from a technical perspective.

(00:45):
I'm again joined by Jen Santos, Chief Corporate Strategic Initiatives Office at Draper, laboratory fellow Geremy Freifeld, and program manager David Hagerstrom. You can learn more about them by visiting our show notes, and you can learn more about Draper, an independent non-profit engineering innovation company at draper.com. Without further delay, let's return to my conversation with Jen, Geremy, and David from Draper. In this episode, we want to dive more into the solution. From a technology perspective, what's going on today? Where are we going in the evolution that is taking place here?

(01:21):
I wanted to start off, and maybe David might be the right person to start off with. For our listeners, this is obviously a very complex issue, could we provide a lay of the land of who are the major players and where technology is right now in terms of what the capacities of these players are producing?

David Hagerstrom (01:46):
Well, thanks, Ken. Let me start to answer the question a little bit tangentially. I think if you can visualize an electronics board, a circuit board, the green widget that's in just about anything in your pocket these days, you can think of it in terms of four kinds of work that goes into that. One is a very integrated set of process steps, usually thousands of process steps, manufacturing steps, that goes into the silicon itself. Most of the state-of-the-art silicon, the computer chips that we revolve our lives around, come from offshore nowadays. But that kind of manufacturing is very distinct. It's very highly integrated.

(02:39):
It's very expensive. And then what happens with that wafer, that silicon wafer, it gets cut up sometimes with a laser, sometimes with a diamond saw, it gets cut up into individual chips, and those individual chips get assembled in some combination to make what I would call a module. That's called packaging. You'll hear a lot about heterogeneous integration or advanced packaging. That means taking those individual silicon chips and assembling them. Now, there's a couple of things that go into that as well besides just assembly machines. The substrate that they're all put together on is in itself a complex integrated product.

(03:26):
The substrate and then the circuit board that ultimately brings the system together, those are both very integrated processes. The United States used to dominate those fields, and we've lost about 80% of the market share, probably more to Asian manufacturers. And then the final assembly of the individual modules onto the ultimate host circuit board is kind of an assembly step. I think in both of those assembly areas we are in pretty good standing, or at least we have the potential to be very competitive through I would say modest types of investment in pieces of machinery, in plants.

(04:10):
But the two very integrated processes, the silicon manufacturer and the circuit board and substrate manufacturer, we have lost a great deal of leadership in those, and it's hard to get back. It's a substantial investment to get those back. I think those are areas that need particular focus as we go forward.

Geremy Freifeld (04:33):
And as far as the silicon itself, I mean, there's very few enterprises that can produce the state-of-the-art . When I say state-of-the-art, I'm talking anything I guess below five, seven nanometer these days. There's really only two. There's TSMC, which is in Taiwan, and there's Samsung in South Korea. Intel is currently trying to play there onshore or offshore in Ireland, in Israel, but they're not quite there yet. They don't have the same advanced technologies that the other companies do. Other than that, domestically, we also have GlobalFoundries, which as of last year is now a publicly traded company.

(05:16):
Most of their manufacturing facilities are in Upstate New York. Those manufacturing facilities were actually built by the government. GlobalFoundries has a long history back to AMD and other things, but they're still at around the 12 nanometer type node at the lowest. They're several generations behind state-of-the-art. I would say Intel right now is two to three generations behind, and GlobalFoundries is even more like four or five generations behind state-of-the-art offshore manufacturing.

Ken Miller (05:50):
When we talk about the supply chain, there are a number of companies globally and a few of them do certain things really well, and then others do other things really well. You mentioned the difference between component manufacturing and assembly and so forth. If you're trying to strengthen your supply chain, you can't do it in a vacuum. If you're trying to do something, obviously you have competitors globally that are trying to also keep their share of the market. From where we're at today, what would you say US does the best in terms of our position in the market?

Geremy Freifeld (06:35):
The two things the US is really great at still today, one is design and electronic design automation tools, which are the software that we use to do the large chip designs. I don't remember the exact statistics, but approximately 80% of electronic design is still done onshore. And that's largely to do to the fact that we have 90% of the electronic design automation market. The US is really one of the only countries that can integrate one of these 30 billion transistor chips into a cohesive component from a design perspective. And then second is industrial tooling. We share a very strong industrial tooling capability with Japan and The Netherlands.

Ken Miller (07:22):
What is industrial tooling?

Geremy Freifeld (07:24):
There's end item manufacturers, like I mentioned, TSMC and Samsung. They produce the silicon chips, but the machines which they use to produce them are not sourced locally to South Korea or Taiwan at all. They have very little industrial tooling capability. The tooling that they use is purchased from the US, Japan, and The Netherlands. We're very strong in that market still.

Jen Santos (07:49):
Let me just step back just for a second. When you think of supply chain, you often think of someone that builds a car and then it's got all the suppliers, or builds a truck and builds all the suppliers. When you think about microelectronic, it is from design through multiple steps until you have a product. If you think of that as the supply chain, I think that will... If you can visualize what I mean by that, right? We have a really great slide that describes the steps actually to the end product. But you think of each one of those steps has industry partners that do those phases of creation.

(08:26):
The design side of it, Draper designs... I mean, you guys can talk about all the different companies that design our exquisite designers of these chips. But there's a whole bunch of them. Now you have a strong industrial base. It's not just a single supplier that does design. And then you go to the next step and you look at the industry partners. When you get to the middle part of it, which is the foundries and the manufacturing, Geremy, you want to talk about that? Now you can see how it maps.

Geremy Freifeld (08:54):
That's where you're constrained to those couple of manufacturers today at the state-of-the-art nodes. The US has this very good capability for design, a lot of designers. Nvidia, AMD, most of those chips are designed in the US. And then to be manufactured, they're sent over to foreign companies for manufacturing. Packaging, they're sent over to even different countries like Singapore and Malaysia for advanced packaging, although Taiwan and South Korea are investing in that area as well. And then they're sent primarily to China for circuit board assembly, and then they come back to the US for integration or for sales really. That's about where we stand right now from a state-of-the-art supply chain.

Ken Miller (09:38):
Is the manufacturing creating a bottleneck in the supply chain process?

Jen Santos (09:43):
Yes. And as you think of it as a line, literally each step, phase one, two, three, four, five, 10, 12, there's 12 that we have, each one of those... There's 50 companies in the first phase, two companies in the second state. You see what I'm saying? But when you get to manufacturing, that's why the CHIPS Act is important, because the CHIPS Act is focused on manufacturing. The middle part, it's not on the design.

(10:07):
The design houses, although are important, we got to keep an eye that they don't go offshore because there's no capital investment into them, right? This investment from the government into the industrial base to shore up some of those vulnerabilities is really focused towards the middle part of the development phase.

Geremy Freifeld (10:30):
Yeah, and that is the bottleneck and that's where we have the problem, because having two, even three companies being able to produce the entire world's state-of-the-art chips is obviously an issue. We saw that during COVID. Although they don't use state-of-the-art chips, but it's still hard to buy new cars nowadays because there's a limiter for chip supply both at state-of-the-art nodes and at the state of the practice nodes.

(10:58):
And that limiter, for example, it gets sold. Back to the spectrum thing, it gets sold. For example, Apple has bought the next three years production from TSMC at a certain node. Even if somebody else wanted to produce chips at that node, they could not because Apple has already consumed the entire capability there.

Jen Santos (11:23):
I think it might be good for us to define what is legacy, present, and state-of-the-art, the three different sectors.

Geremy Freifeld (11:30):
Sure, sure. Well, right now, I would say state-of-the-art, this is obviously evolving. It used to be less than 10 nanometers. Now I would say it's about seven nanometers or under for state-of-the-art, none of which is really produced onshore, although Intel will debate a little bit about seven nanometers itself. None of which is produced offshore. Then there's state of the practice, which is that 12 nanometer node still in very high volume production, say 22 to 10 state of the practice, where that exists in several different worldwide locations. The plants are still running at capacity.

(12:11):
And then there's legacy nodes, which for logic would for computer chips is probably larger than 28 or so. Although I do want to point out, and this is part of the issue with microelectronics, is the memories themselves. Those are usually at legacy type node sizes, which is 45 nanometer, 28 nanometer for flash. Those are the legacy products, which are typically into the more integrated products like cars. They would typically choose state of the practice or legacy. They wouldn't really be state-of-the-art. State-of-the-art is like laptop, cell phones, graphic cards, gaming console type things.

Jen Santos (12:53):
Most of legacy chips are in national security issues. Now, you can start to see, if you're building a plane, a ship, a submarine, you're probably using legacy microelectronics, which gets back to a supply chain issue with not having suppliers available to continue to still build legacy microelectronics. The state-of-the-art, which is the other extreme, which is the most sophisticated, as well as faster, are typically in more commercial end items.

Geremy Freifeld (13:29):
But we see that changing over the next one to two decades. That's very important that we start integrating state-of-the-art components into weapon systems at a much faster rate. I have a little anecdote. I'll go off here. DARPA had a program called AlphaDogfight. I don't know if you've heard of this or saw this, but they basically pit AI pilots against human pilots in the same aircraft. This is an ongoing program. It's also DARPA ACE, A-C-E, Program. But in the first round of the battle scenarios, humans lost every round. It wasn't due to the AI fancy maneuvering or anything like that.

(14:11):
The AI pilots flew straight at the human pilots and just locked down and shot them down before the human pilots could do anything. In the end, after five battles, the human pilot started pulling up and doing all kinds of crazy maneuvers, but the AI pilot didn't really do much, just continued to point his nose at them, acquired the target and shot it. What's going to happen pretty soon, especially with the autonomous things, autonomous vehicles, whether they're underwater, whether they're land-based, whether they're air based, space-based, what's going to happen pretty soon, and we see this in Ukraine today, is that you're going to have two drones flying at each other to deliver weapons.

(14:45):
The drone with the more powerful processor is going to win. There's not going to be any luck. There's not going to be any skill. It's whoever has the more dense amount of gates operating at the higher clock frequency will acquire the target and shoot it down first. That's it. If we don't have access to state-of-the-art, but our adversaries do, we lose. There's no fancy math to it. It's just we lose.

Ken Miller (15:09):
That's a great point. What do we have in place today that points that we're going in the right direction on this?

Geremy Freifeld (15:16):
Right now, a good bulk of our weapon systems have humans in the loop. Our most advanced drones, like Predators, have humans on the ground with Xbox controllers doing a lot of the target acquisition and ordinance deployment. There are always humans in the kill chain currently. But you see that very slowly eroding to where there's more autonomy in the kill chain. I think that that's where the danger becomes, that we're really, really going to lose out. Even if we don't do it, even if we don't automate our kill chain, our adversaries will. The computer will be faster than the human every time.

Jen Santos (15:54):
I'll say this a little bit differently. From an acquisition perspective, the government buys things from industry. Plain and simple. They need a plane, a ship, or a sub. I'll just keep using those. They go ask industry, "Can you go build me a plane, a ship, or a sub and put whatever technology is in it?" Now, you probably are very aware that a lot of our weapon systems are old. The old weapon systems, to go put in a state-of-the-art microelectronic chip into an old weapon system means they have to do a spiral development and it's many, many years away. Is it even a requirement? Do the requirers that build the plane, the ship, or the sub say, "I need the latest state-of-the-art microelectronic in my weapon systems?"

(16:40):
No, because they don't buy microelectronics. They buy weapon systems. Part of the change in this conversation, and this kind of goes back to the last conversation we just had, is educating our industry executive and legislative branch on the value and importance of this so that they're better decision makers or driving where the technology's going. Because it's much easier to just stay with the processor that you have in your weapon system. It's much easier. And then we'll put on a new EW capability, we'll put on a new radar, we'll put on a new engine, but we're not going to go change out the guts, the brains that drive that speed.

(17:23):
As you design new capabilities, if you design it and say, "I need state-of-the-art," which some weapons systems are absolutely going to, but you have this whole history of we launch and we leave. We build a submarine. Five years later, you do modernization to it. Go ahead, Geremy.

Geremy Freifeld (17:41):
A good analogy for this, what we're talking about, and I don't remember the current name, is the Phalanx capability that we deploy on chips. They switched them on. Those are totally autonomous. If something's flying or cruising towards them, they can take it up. It may be a little bit, but the bullet technology has not improved that much. But what the Navy keeps doing is upgrading the internal guts of the Phalanx. It has a lot less sensor dead zones and a lot faster acquisition and time on target. I think that that's the way that war fighting is just going to happen in general as the 21st century gets rolling. If we don't keep pace, we're going to be left behind.

Jen Santos (18:20):
Or you even look at that with critical infrastructure. Critical infrastructure in this country is sitting on old legacy nodes. It's not at the state-of-the-art. Similar, bringing together all the powers of the executive branch to say, "All right, we're going to go together and we're going to grow together and drive those capabilities, encouraging industry to invest. If TSMC and Samsung are the two leaders, as Geremy said, we want to have two or three in the United States as well," you're going to have to have driving that capability to that same level, so industry will invest and then come to the table.

Ken Miller (19:05):
In our last episode, we talked a lot about the aggregation of demand. We're talking about here today the bottleneck in manufacturing. Of course, then you have the legacy to more advanced technology. There's a lot to address, but what I want to get talk about a little bit I guess is just the issue of capacity. You can talk about chips for new advanced weapon systems or evolution of capability, but how can you affect the capacity or the sheer number of what you need to buy to convince industry that, yeah, this is where we need to be going? How does that play into this? Because it seems like if you're trying to move from legacy to advance, you're only talking relatively a small piece of the market.

David Hagerstrom (20:07):
Again, I'll go back to what I said about four areas of electronics manufacturing. Because when you're talking about capacity, you're talking about capacity to manufacture these things. You have to build a silicon, and that's very expensive. The way state-of-the-art silicon is made today, it's on a very high volume basis. You need about $10 billion to set up a factory, and then it has to operate... Or more. It has to operate nearly 24/7 to get a return on investment. That's a problem area. The other three that I mentioned, the assembly of those chips into a package, the building of the circuit boards and the final assembly of circuit boards, they actually scale very nicely.

(21:04):
You can put a plant in place for any one of those operations for much, much, orders of magnitude less money, and it can be right sized for the kind of demand we see from the DOD or from the government. I think there's ways to go after those three, although it still needs attention, but it's not healthy today, but I think there's ways to go after those three and the silicon is a challenge.

Ken Miller (21:36):
Can you talk a little bit about that? Well, you've reached the limit of my knowledge a while ago, but talk a little bit about what goes into the production of silicon.

Geremy Freifeld (21:49):
Oh, certainly. There's a lot of processes, but you can think of them as kind of like layers or steps. You start with a plain silicon wafer, and then you either add or subtract material to it. You dope it, so you put other things besides silicon within the silicon itself in order to change its electrical properties.

Ken Miller (22:08):
What types of substance would you be using for that?

Geremy Freifeld (22:12):
Oh, for doping? I mean, usually you can think of noble gas type thing, like boron. We do fluoride, boron. But in the really state-of-the art advanced nodes, we start doping with three five type materials, which are gallium arsenide, gallium nitride, silicon carbide. There's a lot of different materials. It's a very, very complex standpoint. I say this to my colleagues a lot, but it is the most complicated human endeavor yet realized. It is the most complex engineering feat that humans have been able to do is to produce these state-of-the-art microelectronics. The production of them is in the highest, tightest controlled environment known to man.

(22:59):
The foundries themselves, one of the reasons they're so expensive is not only because of the tooling that we put into them, but also the environment that they're manufactured is the most tightly controlled environment that mankind has yet created. From just a pure engineering perspective, it is the pinnacle of human engineering to date. We can talk about moon landings, which I'm a very big fan of. Draper, we helped, not me, but my predecessors helped develop the Apollo Guidance Computer. That was huge of the day. But today, the moon landing type engineering is happening in the silicon foundries themselves.

Jen Santos (23:36):
I mean, that is the sophistication of this. It's not just the supply chain and the industry partners that do it. It's those investments into those industry partners to be able to maintain those qualification standards to be able to produce.

Ken Miller (23:51):
You mentioned gallium arsenide, gallium nitride in use of production of the silicon.

Geremy Freifeld (24:01):
Nowadays, we think of as a transistor, which is the base building block for complementary metal-oxide-semiconductor, CMOS. It used to be just all silicon with some different noble implantation to change the transconductance of the silicon stuff. But nowadays, there's so many... Like David said before, there's thousands of steps and each of those steps is adding or subtracting something to that silicon. A lot of that is new material technology and different things. We still think of copper, especially in the defense industrial base, as a gold standard for passing signals.

(24:43):
But at the very deeps of micron state-of-the-art foundries, they're replacing copper with different cobalts and zirconium and stuff like that at this point. It's very advanced engineering and material science stuff that goes into it. A lot of the people that get physics degree that used to go into particle physics and things like that are now really working on the microelectronic fabrication problems.

Ken Miller (25:16):
We've had this conversation in the past at AOC about the resource availability, mining of rare earth elements and so forth, and the availability of the materials that go into this and the limited supply that we have there to even start this whole thousands of step process. Is there a next step where we're not as resource constrainted in the materials that we're using?

Geremy Freifeld (25:47):
There's a published roadmap, but I don't think it really addresses the raw material supply issue. I mean, the public roadmap is really focused on continuing to scale the transistor to allow higher transistor density for state-of-the-art applications. But I'm going to flip that on you and say an issue is that the manufacturing today, even at TSMC'S perspective, the question is, is it economically sustainable? That's the real question. At the seven nanometer node, we did a calculation, 80% of the cost at the seven nanometer, which is kind of one or two generations ago in the iPhone, 80% of that cost is equipment depreciation and maintenance.

(26:34):
As Dave mentioned, he said 10 billion, I would argue that at the three nanometer, we're talking a 30 billion startup cost for a foundry. You have to have such a large market share before you build the foundry captured in order to make that foundry sustainable, that it's really not economically viable to stand up true competitors in the state-of-the-art space with the current manufacturing paradigm. That's the fundamental problem that we have. The CHIPS Act, the people on the advisory board and the commerce department need to understand that it's not going to get a good return on investment if we're just doing like for like, and that's where we talk about at Draper a kind of domestic differentiation.

(27:17):
I think that the thing that we are trying to educate the various policymakers on is that the way forward is probably not at the end item manufacturer, like the Intel, TSMC or Samsung type thing, but at the industrial tooling and design phase to try to change the manufacturing paradigm to allow a lower cost of entry and a higher speed of technology development. Right now, TSMC, to build a new plant in Arizona, it's going to take them three years just to get the first product off of the thing. That's fine from a commercial perspective because they have their roadmap, they know what they're doing. But if we want to overtake foreign advances in technology, we have to iterate faster than them.

(28:04):
We're going to have to do it with a different paradigm, with a new industrial tooling in order to realize what we call low-volume high-mix, which means we can use the same or similar tool sets to develop legacy components for older weapon systems and use those tools to realize the highest technology implementations as well.

Jen Santos (28:32):
Geremy did a really good job describing that. We call that being disruptive. You have to disrupt that early phase and think differently and operate differently to expect a different outcome. Using the same materials, the same tooling, the same design to beat a competitor isn't the answer here.

Geremy Freifeld (28:49):
That already has a pretty substantial lead on you.

Jen Santos (28:51):
Yes. The disruption has to begin at the early part of this development process.

Ken Miller (29:00):
David, looking at the evolution of the chip, obviously there's this concept of Moore's law and it's getting smaller, an order of magnitude smaller. Where is the evolution of the chip taking us? What are some of the advances that are both most exciting to you or maybe developments that are maybe keeping you awake at night thinking like, here's where chip technology is going, here's the point in the future we need to be meeting it at?

David Hagerstrom (29:35):
I would say you can think of it in two paths. One path is heterogeneous integration. In other words, there's a lot of products that combine a lot of functions into a microelectronic module. Those functions don't all need to be built at the state-of-the-art. Memory, depending on what the product is, you might be able to do just fine with memory that's built from a different semiconductor node. There's a great deal of attention going into heterogeneous integration, and that's really coupled with the idea of advanced packaging. How do you assemble these heterogeneous pieces of silicon into a very efficient, very high performance combination in the package?

(30:26):
That's one direction that's very exciting. It's getting a lot of attention, and it's one where we need to play catch up in order to be competitive. The other is at the silicon or semiconductor level. Moore's law has been talked about for many, many years. It seems like there's still some life in Moore's law with three-dimensional processing and some of the other innovations that we need to pay attention to.

Geremy Freifeld (30:57):
Certainly. The government, basically DARPA, and industry is very focused on the next what. There's carbon nanosheets, what we call gate-all-around structures. As our computational physics become more advanced because of the new computers, we can develop more advanced structures with which to manufacture, and then we develop new tooling with which to manufacture those structures. What we're saying is we need to flip that a little bit and go to the how and focus a little bit first on the tooling to make the tooling economically viable from a high-mix low-volume application so that we can swap tooling out and achieve technology more quickly.

David Hagerstrom (31:39):
Now, part of that paradigm might be simply having a branch, and simply is obviously a misnomer, but having a branch in the equipment business model which allows for a smaller scale copy of the same thing that's being built for the big commercial companies. If you're putting a 300 millimeter, a large piece of equipment in the market in order to support the high volume business, might there not be a market for a scaled down version of that, it would benefit from the R&D that's done for the commercial unit and be much more affordable.

Geremy Freifeld (32:25):
Well, right. Industrial tooling make these pilot tools themselves, but they usually don't make them commercially available because of the economics of development. I think that where we would steer the policymakers is to directly investing into the tooling itself so that we can gain access to those early pilot tools in order to construct new manufacturing processes.

Ken Miller (32:52):
We talked about partnerships, industry, government, obviously there's the nonprofit world, associations like the AOC. What role do you see as associations like the AOC playing in terms of encouraging these partnerships, but especially in the STEM world in interacting with colleges and universities and engineering programs? What can we do to help this along from a private, nonprofit perspective?

Jen Santos (33:18):
I'll say Association of Old Crows has been around for how long?

Ken Miller (33:21):
About 60 years.

Jen Santos (33:25):
You have an opportunity to educate. You have brilliant folks associated with association that are high tech, understand electronic warfare, electronic components. You have this opportunity to educate, hosting these type of sessions and then sharing this with the younger workforce, not just with the association. Bringing this to light in a way that's spoken in I joke and say it's in English and not in government speak. Having the English conversation around this and not the government speak I think can bring a lot to light. Go ahead.

David Hagerstrom (34:01):
Well, I'll reiterate my thousand points of light. My granddaughter is doing Lego Robotics. It's just a great thing, and it's a very local grassroots kind of thing. She's in a girls program after school. It's all girls learning robotics. Any one of your members, anyone really listening to this podcast could think about, can I reach out at that local level and start to incite the young children to have an interest in a field like this?

Jen Santos (34:31):
Because they grew up with laptops at their fingertips. I mean, my kids, as soon as they could fig... I actually had my son take computer classes before I had him work on his handwriting, truthfully. I said, "You can type way faster than you can write things and you'll use that." They are learning, but the knowledge and skills that you have, they can transfer to those folks. I mean, robotics classes, yes. There are so many opportunities in those spaces to educate the future of our national and economic security.

Ken Miller (35:10):
We're running out of time here. Jen, any closing thoughts in terms of wrapping everything up that we were discussing or any points that we did not get to?

Jen Santos (35:20):
I'll tell you that, one, it's really a pleasure for us to hopefully bring education, insight and intellect to a problem that our company spends a lot of time investing in. I mean, over a third of our engineers work distinctly on the microelectronics problem from a safety and security issue. Giving us that opportunity is fantastic and hopefully we brought some education to your listeners. But we need to challenge ourselves, industry, executive branch, legislative, to address the questions we've been just discussing. We need to partner with friends and allies to amplify our collective competitive advantages.

(36:04):
We need to instill a sense of responsibility to the national and economic security problem, because rapid changes in technology shapes every aspect of our lives and national interest. We must reinvest in retaining our scientific and technological edge, working alongside our partners to seize those opportunities and advance national security. Microelectronics is a perfect example of this. We need the United States' emerging technology to boost our security and our economic competitiveness. We're honored to be able to bring this education to light in the vein that we have, and we look forward to continued conversations over the next years.

Ken Miller (36:42):
I appreciate you taking time to join me here in person having this conversation. As always, we could always spend twice as much time talking about this, and hopefully this will be an opportunity where we can have you back and continue this conversation, not just with our podcast, but through our other programs that we have, whether it's webinars or conferences. I think our community needs to hear from your expertise on a regular basis. Looking forward to that opportunity as well, but do thank you for joining me here on From the Crows' Nest and it's great to have you on the show.

Jen Santos (37:14):
Thank you.

David Hagerstrom (37:14):
Thank you, Ken.

Ken Miller (37:16):
That will conclude this episode From the Crows' Nest. I want to thank my guest, Jen Santos, Geremy Freifeld, and David Hagerstrom for joining me here for our two-part conversation. Again, you can download part one, which was released on January 18th. You can learn more about Draper via their website at Draper.com. Also, don't forget to review, share, and subscribe to this podcast. We always enjoy hearing from our listeners, so please take some time to let us know how we're doing. That's it for today. Again, you can follow me on Twitter @ftcnhost. Thanks for listening.