The Biggest Money-Making Opportunity Since Bitcoin?

[0:00] 100,000 robots for $50,000 each. That's
[0:02] around $5 billion of revenue. Now, okay,
[0:05] what if I sell a million robots? That's
[0:07] $50 billion. You're getting to almost
[0:09] the scale of some of the biggest
[0:10] companies in the world. A million robots
[0:12] is really nothing, right? Cuz Amazon has
[0:14] millions of workers and Walmart has
[0:15] millions of workers. And what if I sell
[0:17] tens of millions? Okay, that is now 500
[0:19] billion. And so, you can see there's a
[0:22] really clear path. Trillions of dollars
[0:24] of revenue and that would imply tens of
[0:26] trillions of dollars of market cap. And
[0:28] that doesn't even consider stuff like
[0:30] Jeban's paradox where what's going on
[0:32] guys? Today we have a very important
[0:34] conversation with Andrew Kang. Andrew is
[0:35] one of the best investors in the crypto
[0:37] world. But the last couple of years he
[0:39] has spent a majority of his time focused
[0:40] on the robotics industry. Today he
[0:42] serves as the CEO of Robo Strategy, a
[0:45] publicly traded closed end fund
[0:46] specifically focused just on investing
[0:48] in the robotics industry. In this
[0:50] conversation, we talk about how big the
[0:52] market is, why traditional venture
[0:54] capitalists haven't been interested in
[0:55] it, why Andrew thinks that this is a
[0:57] place that his capital can compound over
[0:59] a long period of time, how he personally
[1:00] built the conviction to actually invest
[1:02] a large sum of money into these
[1:04] companies, and then why he chose to use
[1:06] a public vehicle to go and invest in
[1:08] these companies. This conversation, I
[1:10] think, is one that you guys will look
[1:11] back for years and years to come. I
[1:13] highly suggest that you pay attention. I
[1:15] think that Andrew is a very special
[1:16] investor, but I also think the strategy
[1:18] in the industry that he's going after,
[1:20] it's something that people don't yet
[1:21] understand, but something that I'm
[1:22] personally very convicted in, and I
[1:24] think that there's a lot of opportunity
[1:25] here. Here's my conversation with Andrew
[1:27] Kang. Andrew, you're one of the best
[1:29] crypto investors in the world. You've
[1:31] had an incredible track record over the
[1:32] years, but now you're spending a lot
[1:34] more time on humanoid robots and
[1:35] robotics in general. Why the shift in
[1:38] attention and capital to a market that
[1:40] maybe most people don't understand that
[1:42] well yet?
[1:43] Yeah, I mean there's a lot of reasons
[1:45] we're we're finally going to have
[1:46] robots, right? It was always, you know,
[1:48] this big dream in sci-fi. You'd see
[1:51] robots that were shaped like humans
[1:53] doing everything a human could do. And
[1:56] we're finally getting to the point where
[1:57] that is that's possible. It's not a
[1:59] 50-year thing. It's not a 20-year thing.
[2:02] This is probably more so like a 3 to 5
[2:05] year thing, right? Where you start to
[2:07] see humans humanoids in everyday life.
[2:11] Now, when you think about these humanoid
[2:13] robots, I always go to there's different
[2:15] ways that people envision these fitting
[2:18] into our lives. There is like humanoids
[2:20] that are in factories. There is videos
[2:22] uh whether they're AI generated or real
[2:25] of demos where they're doing things in
[2:26] our homes. Then there is the like maybe
[2:29] Jetson style like they're like walking
[2:30] down the street walking your dog or
[2:32] doing stuff in, you know, society.
[2:34] >> What is your vision for how humanoid
[2:36] robots are going to actually fit into um
[2:39] kind of our future?
[2:41] I I think you know almost all of the
[2:43] above is possible in the future, right?
[2:46] Um you're going to have humanoids act
[2:48] the same way and participate in jobs the
[2:50] same way that humans do. Some of them
[2:52] they might be focused on a specific job.
[2:55] They might be a service worker at a
[2:57] restaurant. Some of them they might be
[2:59] working a specific factory job. Some of
[3:01] them might be your personal assistant at
[3:02] home. Right? I think the beauty of
[3:04] humanoids is that they're adaptable to
[3:08] the everyday world. Right? They can go
[3:10] up and down the same buildings that we
[3:11] do. They can use the same tools that we
[3:13] can do. And they in general are just
[3:18] very multi-purpose and very adaptive,
[3:21] right? And that's that's the difference
[3:22] between a humanoid robot and say a more
[3:25] application specific robot. And I think
[3:28] that's why you see you know uh you see
[3:32] the prevalence of similar concepts or
[3:35] similar items in everyday life like your
[3:38] your smartphone for example right it's
[3:40] very general purpose I can play music on
[3:43] it I can use it as a map to navigate the
[3:45] world I can call people with it and in
[3:47] the same way a humanoid has such great
[3:51] multi-purpose functionality that it
[3:53] makes sense that it can permeate so many
[3:55] different aspects of of of life in
[3:57] general.
[3:58] >> So when you think about the total
[3:59] addressable market, like what do you
[4:01] think it looks like for humanoids?
[4:03] >> Yeah. I think to understand that you
[4:05] have to look at what is the total market
[4:08] for human labor. It's something like 50%
[4:10] of GDP. It's, you know, some people put
[4:13] it at $40 trillion, $60 trillion. Seems
[4:16] like a massive number, right? Um the way
[4:18] that I would think about it is a
[4:21] humanoid can do everything that a human
[4:24] can do in the future, right? but it's
[4:26] doesn't need a rest. It doesn't need to
[4:27] take breaks. Doesn't need to go on
[4:29] vacation. It's not going to quit on you.
[4:30] So, you don't need to hire and, you
[4:33] know, retrain somebody else. A lot of
[4:35] these like jobs in factories, they suck
[4:37] and there's really high turnover. And
[4:39] that is a really high overhead cost to
[4:42] having physical uh um you know, labor.
[4:45] And you you don't have any of that with
[4:47] with robots, right? And so, one robot
[4:49] could theoretically perform uh the
[4:52] function of maybe three humans, right?
[4:54] uh assuming it's able to work multiple
[4:56] shifts. And so if we take that
[4:58] perspective of well um you know robot
[5:02] robots are going to replace most of
[5:03] physical labor then you can also maybe
[5:06] do a bottoms up analysis where you say
[5:08] hey look one robot $50,000. Uh how does
[5:12] that compare to a human? Uh well a human
[5:15] I have to pay you know every single
[5:17] year. Uh robot that's a onetime fee.
[5:20] Maybe I have to pay a little bit for
[5:21] electricity. Maybe you have to pay a
[5:23] little bit for maintenance. And but when
[5:25] I look at it from a per hour basis,
[5:29] right, and I I expect this robot to last
[5:31] multiple years, that math gets you down
[5:33] to around $2 per hour. And in America,
[5:36] you know, you take the all-in cost of,
[5:38] you know, your average worker and, you
[5:41] know, their bonuses, their insurance,
[5:43] and everything else associated with it,
[5:45] you know, you get to something like $35,
[5:46] $40. And so it's it's a no-brainer in
[5:49] terms of the cost structure. Um, but
[5:51] even if you compare it to, you know,
[5:52] low-income countries, uh, like India or
[5:56] the Philippines or Indonesia, it's
[5:59] honestly it's hard to beat $2 an hour
[6:01] there as well, especially with all the
[6:02] benefits that, you know, we discussed.
[6:05] Um, and then so I think you can come to
[6:07] an understanding that look, it's going
[6:09] to be a lot of robots. We sell billions
[6:12] of cell phones every single year. We
[6:15] sell hundreds of millions of PCs and and
[6:17] cars, right? And so probably sell a
[6:21] similar amount of of robots. And so but
[6:24] like what if we just sell 100,000 robots
[6:26] for $50,000 each?
[6:29] That's around $5 billion of revenue. I
[6:31] mean you you you're making that much
[6:33] revenue. You're already I would say like
[6:35] a hundred billion plus business
[6:37] potentially. Um assuming your margins
[6:39] are good. Um, and
[6:42] now, okay, what if
[6:45] I sell a million robots?
[6:49] That's $50 billion. It's, you know,
[6:51] you're getting to almost the scale of
[6:53] some of the biggest companies in the
[6:54] world. What if I sell But a million
[6:56] robots is really nothing, right? Because
[6:58] Amazon has millions of workers and uh
[7:01] Walmart has millions of workers. And
[7:02] what if I sell tens of millions? Okay,
[7:04] that is now 500 billion. And so you can
[7:07] see there's a really clear path in terms
[7:10] of like how we can potentially achieve
[7:12] achieve trillions of dollars of revenue
[7:15] and that would imply tens of trillions
[7:16] of dollars of market cap. And that
[7:18] doesn't even consider stuff like Jeban's
[7:21] paradox where you're lowering the cost
[7:24] of labor of of of the product. You're
[7:27] actually increasing the market size,
[7:29] right? Because now things that were too
[7:30] expensive to do before are now
[7:32] economical. They might make sense. And
[7:35] so that top down type of sizing is I
[7:38] think really puts it into perspective if
[7:41] you look at but if you look at something
[7:44] like Apple right like people would have
[7:45] thought you were crazy if you told them
[7:47] in 2006 that this would be a $3 trillion
[7:49] company today cuz before the iPhone came
[7:51] out it was like what $50 billion company
[7:54] and that you know 3 trillion would have
[7:55] been bigger than like most of the big
[7:57] companies combined back then. And so I
[8:00] think people really underappreciate the
[8:02] amount of value creation that can happen
[8:03] with really transformative uh techn
[8:07] technological jumps.
[8:08] >> Now there's a lot of people who I would
[8:09] say are kind of Monday morning
[8:10] quarterbacks, right? Or they'll sit and
[8:12] they'll say, "Oh, I think humanoid
[8:13] robots are going to be big." And they do
[8:14] nothing about it. You have taken an
[8:16] immense amount of personal capital over
[8:18] the last couple of years and you have
[8:20] put it into some of the leading private
[8:22] companies in the humanoid and robotic
[8:24] space. And when I say uh you know
[8:26] immense amount of personal capital, I
[8:28] think some of the investment sizes are 8
[8:30] figure personal checks that you wrote
[8:32] into these companies. And so just talk
[8:34] about to get that level of conviction.
[8:37] Is it this is for sure going to happen?
[8:39] Is there some level of asymmetry? Like
[8:41] what was the thing that convinced you,
[8:43] okay, these humanoid robots are actually
[8:45] going to be here and they're going to be
[8:46] as pervasive as you're saying? Yeah,
[8:48] it's it's it's interesting because when
[8:50] I first started investing in humanoids
[8:53] and robotics in general, it was
[8:54] definitely not consensus.
[8:57] I found out about figure AI around 2023,
[9:01] late 2023, and I had no experience
[9:03] investing in the space, right? And so I
[9:06] thought it was really exciting because
[9:08] Chachi PT had just come out and it was
[9:10] clear that the path of AI development
[9:13] was going to dramatically accelerate.
[9:15] And so, not only were we going to get
[9:17] digital AGI, but we were going to get
[9:19] physical AGI. And the intelligence
[9:22] portion was always the bottleneck to
[9:23] making robots work. And that was that
[9:26] was going to be solved. Um, and so with
[9:30] that in mind, I went out to a bunch of
[9:32] my traditional, you know, VC network and
[9:34] I asked them, hey, like what do you
[9:36] think about this company?
[9:38] All of them told me not to invest. Um,
[9:41] and I think a lot of the reason came
[9:44] down to the fact that the robotics
[9:47] industry historically has not produced a
[9:49] lot of winners, a lot of big venture
[9:51] scale outcomes.
[9:53] Um, there's a lot of challenges with
[9:54] developing robotics. Um, the hardware
[9:57] iteration cycle is very long. It's very
[9:59] expensive and even after getting them to
[10:02] work, the deployment can still be very
[10:05] messy and expensive. Um,
[10:09] I think what a lot of investors failed
[10:11] to appreciate was that the development
[10:14] of physical intelligence was going to
[10:15] change all of that, right? And so
[10:21] the more I kind of dug into this
[10:22] investment,
[10:24] the more I kind of
[10:27] realized how underappreciated it was
[10:29] because I tried I, you know, I didn't
[10:31] even invest directly at first. I
[10:33] invested through four SPVS. Uh, and so
[10:35] when you're getting all these SPVS
[10:37] thrown at you, right, you're getting
[10:38] access to a deal which is supposedly
[10:40] supposed to be a really great deal, it
[10:43] you think, oh wow, am I getting ripped
[10:45] off, right? Like,
[10:46] >> why am I so
[10:47] >> why? Yeah, why why am I getting so much
[10:49] access? Why are the top VCs in the world
[10:52] not picking this up? And the more I dug
[10:55] into it, the more I realized it was more
[10:58] so people, you know, had certain
[11:00] industries that they were comfortable
[11:02] investing with, certain thesis, and
[11:05] deviating from that is it can be very
[11:07] uncomfortable. You have to take a lot of
[11:09] risk. But there was no
[11:12] I don't think there was any question
[11:13] that there's product market fit for
[11:15] humanoid robots once these actually
[11:16] work, right? It was just a question of
[11:18] like when when are these actually going
[11:20] to work and is this a team uh is this
[11:23] the right founder to make it happen. And
[11:26] so the more I dug into it the more I
[11:28] actually wanted to increase you know my
[11:30] initial investment from a million to 5
[11:33] million eventually to around 19 million.
[11:35] And at the time I had not even spoken to
[11:39] to Brett the founder. uh you know I had
[11:41] spoken to some people on his leadership
[11:43] team and I had spoken to a lot of other
[11:46] investors in the round but I think
[11:51] doing all the research and understanding
[11:53] the background of Brett that the team he
[11:55] had assembled which were some of the
[11:57] most world-class roboticists
[12:00] um understood that this was one of the
[12:02] teams that was very likely going to make
[12:03] it happen.
[12:05] >> Let's talk a little bit more about
[12:05] Figure AI. So, you wrote a $19 million
[12:07] check initially, you know, kind of a
[12:09] collection of different investments. Um,
[12:11] what was it about them that they're
[12:13] doing differently, right? What I've seen
[12:15] the demos online. I think a lot of
[12:17] people have seen those now. I've talked
[12:18] to Brett before. Um, but what is the
[12:20] thing that you think they either do
[12:22] uniquely well or the thing that you
[12:23] think will continue to be a moat for
[12:25] them as they go to build this business?
[12:27] >> There there's so many things. I mean, I
[12:29] I think the the first is the caliber of
[12:31] the team that that Brett assembled. Um,
[12:34] you know, there there were a few other
[12:36] humanoid robotics companies that have
[12:37] been around for longer than I figure by
[12:38] the time that they were incorporated,
[12:40] but their rate of iteration and their
[12:42] their speed of execution was completely
[12:44] unmatched in terms of what I was seeing
[12:46] in terms of progress uh month after
[12:48] month, quarter after quarter. Um, if if
[12:52] you look at if you just think about the
[12:54] the challenge that's in front of you in
[12:55] terms of how how do I solve general
[12:57] purpose robotics, right? It's one of the
[12:59] most difficult challenges in the world.
[13:02] You need PhDs in computer vision. You
[13:04] need PhDs in uh robot behavior. You have
[13:07] PhDs in um uh you know uh hand
[13:12] engineering or I don't think there's
[13:13] even a PhD for that, right? But you need
[13:15] experts in hand engineering. Uh you need
[13:17] experts in uh developing fleet
[13:19] orchestration software. And so you have
[13:22] all of these different
[13:24] special fields that there are maybe only
[13:27] a hundred of people that are really
[13:29] competent, right, in in these specific
[13:31] fields. And you need all of them within
[13:32] one company. And to do that is a huge
[13:34] challenge. And so you need a founder
[13:36] that that can do that and to raise the
[13:38] billions of dollars of capital that you
[13:40] need to make that company a success. And
[13:42] so Brett was really one of the only few
[13:44] founders I had found at the time that
[13:46] was able to do that. You know, I had
[13:47] looked at the other humanoid companies
[13:49] in the space. And you know, outside of
[13:51] Tesla, right, nothing nothing was really
[13:53] comparable. What's interesting to me is
[13:54] humanoid robots seem to be very similar
[13:56] to the drone industry where there's a
[13:58] ton of um you know uh experimentation
[14:00] maybe even speculation at the earliest
[14:02] stages of the technology. There are a
[14:04] couple of companies that are able to get
[14:05] some sort of innovative breakthrough but
[14:07] then those companies really struggle to
[14:09] commercialize the innovation that
[14:11] they've done. So you either have kind of
[14:12] very technical teams that are able to
[14:14] work on the hardware and software of a
[14:15] drone but then they have a hard time
[14:17] actually building a business around it.
[14:18] What I think is happening in humanoid
[14:20] robots is we are now starting to see
[14:22] teams that not only are good at the
[14:23] technology but they also have a track
[14:25] record and experience of actually
[14:26] building a business as well. Yeah.
[14:28] >> Is that your experience in in terms of
[14:30] why maybe now is the moment uh why these
[14:33] are actual companies not just kind of
[14:35] cool Boston dynamic robot you know demos
[14:37] or something?
[14:37] >> Technical experience is important. Yeah.
[14:39] Don't don't get me wrong, but I think
[14:42] sometimes um investors can overindex on
[14:46] maybe you know like a PhD or professor
[14:49] type of talent uh which I think is
[14:52] instrumental to have on the team but if
[14:55] you look at Brett's background for
[14:57] example you know his first company Veter
[15:00] was a company focused basically on
[15:02] recruitment right and so he understood
[15:05] and built the skill set of how do I find
[15:07] and attract the best talent in the world
[15:10] and the second company Archer which he
[15:13] took public I think it's still around
[15:14] you know5 to10 billion um after he took
[15:17] it public in a spack a few years ago
[15:20] I mean he was innovating a whole new
[15:22] type of complex machinery that really
[15:24] hadn't existed before commercially and
[15:27] that type of skill set to create you
[15:31] know almost like a stepwise change in
[15:34] um you know what machines can do is
[15:37] really unique in America
[15:39] I I think you see that in some other
[15:40] parts of the world, but America has
[15:42] outsourced a lot of their hardware
[15:44] design and their manufacturing, right,
[15:46] to places like China. And you know, you
[15:49] you look at some of the biggest hardware
[15:50] companies in America like Apple, right?
[15:53] Like the phone, iPhone, it's getting
[15:55] better, but only very marginally
[15:57] year-over-year. It's not a really big
[16:01] change in terms of th this is like a
[16:04] completely new invention. Mhm. Now you
[16:06] mentioned China. Obviously uh anyone
[16:08] paying attention to the space has seen
[16:10] incredible demos and videos coming
[16:12] China. I have seen half marathons run. I
[16:14] think they had a robot Olympics where
[16:17] they were doing all kinds of different
[16:18] events. Uh China is known to be very
[16:20] good at building hardware as you
[16:22] mentioned. They have a great supply
[16:23] chain. They've somewhat become the
[16:25] manufacturer of the world in many
[16:26] people's eyes.
[16:27] >> Why do you think the United States or
[16:29] any American company can beat China at
[16:31] this game?
[16:32] >> Yeah. Oh man, the the US versus China
[16:34] robotics topic I think we can talk about
[16:36] for a while, right? Um
[16:39] look, China is undoubtedly uh undefeated
[16:43] in in manufacturing. Um and I think they
[16:46] will continue to be. But I would
[16:47] evaluate robotics companies on three
[16:50] competencies. Uh one is their ability to
[16:53] execute on high rate manufacturing.
[16:56] Second one is their hardware design. Uh
[16:59] and the third one is their AI
[17:01] capabilities. Right. And we already
[17:05] covered manufacturing. I would say
[17:06] China's very strong at. There are some
[17:08] companies like Tesla and Figure that are
[17:10] getting very strong at that in America.
[17:12] Uh and when it comes to hardware design,
[17:16] you have,
[17:18] you know, I would say Figure and Tesla
[17:21] at the top in America. And in China, you
[17:24] have, I would say, you know, maybe like
[17:26] 100 plus different companies. and
[17:27] they're all in various degrees of I
[17:30] would say excellence in terms of how
[17:32] good their their hardware is. If you
[17:34] look at Unitry, for example, their
[17:36] hardware is great for research and for
[17:38] entertainment purposes. You see them
[17:40] doing back flips, uh you see them
[17:41] dancing, but you can't take that same G1
[17:44] robot and and put it in in factory and
[17:47] have it lift 30 lb payloads, right? It's
[17:50] going to fall apart. And so it's not
[17:53] like and but it's also not like Unitry
[17:55] can't design robots that are more
[17:58] durable. It's just not the market that
[18:01] they've decided to pursue at at first.
[18:03] Um that being said, um the AI piece is
[18:09] is is I think what is underappreciated
[18:11] that the US is ahead on in terms of
[18:12] physical intelligence. uh the top
[18:15] physical AI labs in America I think have
[18:18] some meaningful edge over the leading
[18:20] Chinese labs um and without the robot
[18:23] brain right the the robot's pretty
[18:25] useless and so you know if that trend
[18:29] continues then maybe we do see some
[18:31] dependency from the Chinese companies on
[18:34] US physical intelligence I I think what
[18:37] is also underappreciated
[18:39] is that
[18:42] the amount of technology that's
[18:44] developed is not always onetoone with
[18:46] the amount of value that's created. And
[18:49] so you can have really amazing robot
[18:52] technology come out of China. You can
[18:54] have that industry flourish. And it
[18:58] doesn't always mean that's results in
[19:00] the biggest market cap or the biggest
[19:01] win for shareholders, right? And you've
[19:04] seen the same dynamic play out for cars,
[19:06] EVs, and for cell phones. China makes
[19:09] amazing cell phones and they make
[19:11] amazing electric vehicles. If you look
[19:13] at BYYD,
[19:15] I think a lot of people have a lot of
[19:17] great things to say about their cars.
[19:18] They sell more cars than Tesla, but
[19:21] their market cap is, you know, onetenth
[19:24] 120th of Tesla, right? And their margins
[19:27] are significantly lower. And that's an
[19:29] issue across all Chinese, I would say,
[19:32] hardware companies in general is that
[19:35] they play in an ultra competitive
[19:36] landscape. And that is partially kind of
[19:40] encouraged by the government, right?
[19:42] They're putting out these subsidies and
[19:44] they're giving support to these
[19:45] companies and they want a wide field of
[19:48] these players to exist.
[19:51] But the end result of that I would say
[19:54] is more so to benefit society and the
[19:57] consumers within the nation rather than
[20:00] the companies themselves. Obviously the
[20:01] companies have to be successful. They
[20:03] have to they have to grow economic
[20:06] value. At the same time, it doesn't mean
[20:09] that they're going to be as big as say
[20:11] the biggest American companies, right?
[20:12] Like Apple's the biggest phone company
[20:14] in America, even though Huawei and
[20:16] Xiaomi make great phones.
[20:18] >> What what's interesting to me is there's
[20:19] not only the capital markets that you uh
[20:21] participate in, but there's also the
[20:23] regulatory environment that you uh uh
[20:25] participate in, and it seems to me like
[20:27] uh in the United States, there is a very
[20:30] um big conversation happening around,
[20:33] you know, self-driving cars or other
[20:34] forms of robotics and autonomy. We
[20:36] obviously see lots of people yelling and
[20:38] screaming about data centers and it just
[20:40] feels like there's a lot of scrutiny and
[20:42] I would even argue that maybe society is
[20:44] fractured a little bit on is this stuff
[20:46] good, is this stuff bad, whether it's in
[20:48] software or hardware form.
[20:50] >> I don't see a lot of that conversation
[20:51] coming out of China. I don't live in
[20:53] China though. So I I don't know kind of
[20:54] what the the temperature is if you will
[20:56] in the local communities or in society
[20:58] in general. But what I do think is
[21:00] really interesting is the success of the
[21:03] companies in America is almost despite
[21:06] all of that debate, all of that uh
[21:09] scrutiny and it feels like the Chinese
[21:11] companies are getting immense amount of
[21:14] uh subsidies and and lots of help from
[21:16] the government. And the reason I kind of
[21:18] paint this picture of, you know, two
[21:20] different uh maybe relationships between
[21:22] the public and private sector
[21:24] >> is I do wonder if the American companies
[21:26] benefit in the long run from having to
[21:29] do the hard thing versus having the
[21:31] Chinese companies, if there is
[21:32] subsidies. Like I've heard rumors that
[21:34] the Chinese government is actually
[21:35] subsidizing people to buy the humanoid
[21:37] robots.
[21:38] >> So, hey, if you want a factory full of
[21:40] humanoid robots, we'll give you some
[21:41] money to subsidize. I don't know if
[21:42] that's true or not. Maybe you have some
[21:44] insight into it. But it does feel like
[21:46] maybe in the US there's less government
[21:48] support and therefore the companies have
[21:50] to be able to actually innovate and
[21:52] actually be successful in the free
[21:54] market much more than a Chinese company.
[21:55] Would you say that's true?
[21:57] >> I would say there's there's some truth
[21:58] to it but I would say
[22:01] government support would not you know it
[22:04] would not be negative for the industry.
[22:05] I I would say it's very positive for the
[22:06] industry and I think we should have more
[22:08] government support. And there are also
[22:10] there are also a lot of initiatives
[22:12] being pushed by lawmakers and lobbyists
[22:15] to uh create more support for the
[22:17] industry. Um the the interesting thing
[22:20] about what's happening in China is that
[22:23] yes there's a little bit of what you
[22:25] described in terms of you know companies
[22:29] forming JVS almost with local
[22:31] governments and you know them
[22:33] establishing say a center to do robotics
[22:36] research collect training data and you
[22:39] know that center might also be a big
[22:42] purchaser of of robots from you know one
[22:45] of the companies that had helped
[22:47] establish that JB and so there's a
[22:49] little bit of circularity going on. Um I
[22:52] wouldn't say that is
[22:55] um misleading in terms of their sales
[22:57] cuz those robots are being sold for a
[22:59] good purpose, right? They're sold for
[23:00] collecting training data. I think
[23:01] there's a question around how useful is
[23:03] that training data and how effective are
[23:05] they at at collecting it. Um but I think
[23:09] more importantly, it's just when you
[23:11] have say 100 players competing against
[23:13] each other, it you're you're going to
[23:16] squeeze down margins. you're going to
[23:17] have a bunch of IP transfer as well in a
[23:20] in and you know an environment like
[23:22] China and so there's going to be a lot
[23:24] of innovation at one company but then it
[23:25] might quickly leak out to all the other
[23:28] companies right from talent flowing back
[23:29] and forth so much whereas if you only
[23:31] have a few companies in America and
[23:33] maybe they're geographically separated
[23:35] you're going to have less of that um an
[23:38] important point I also want to bring up
[23:39] since you mentioned regulation is that
[23:43] there is
[23:45] there there's I I think been very
[23:47] obviously a big push to re-industrialize
[23:51] America, right? To bring back
[23:52] manufacturing
[23:54] for us to be independent on our own
[23:56] supply chains to build things in America
[23:58] generally. And to do that, we we need
[24:01] our own domestic robotics industry. We
[24:02] can't rely on other nations. And if
[24:07] you've seen what happened with EVs,
[24:08] right, we don't really have Chinese EVs
[24:10] in America. Why is that? Well, the
[24:13] reason is because, well, one, Biden put
[24:16] 100% tax on Chinese EVs during his
[24:18] administration. And then two, the FTC
[24:22] also outlawed, you know, vehicles from
[24:24] certain countries that have, you know,
[24:26] software from those countries. And they
[24:29] have telecommunications devices, right,
[24:31] from those countries. Think about what
[24:33] are robots? Well, they're things that
[24:35] can see and hear everything in the
[24:36] world, and they're filled with
[24:37] telecommunication devices. And so I I
[24:40] have a hard time seeing the American
[24:42] government allowing, you know, robots
[24:45] from other nations
[24:47] be the dominant player in America. Uh
[24:49] and there are certain bills that are
[24:52] trying to be passed right now that are
[24:54] centered around this, right? That are
[24:56] explicitly banning robots from certain
[24:58] countries um from at first being
[25:02] purchased by fedally funded
[25:04] organizations, but you know there's this
[25:06] concept that that will expand more
[25:08] broadly to you know everyday life. How
[25:11] much of those bills are targeting uh we
[25:14] want to encourage domestic manufacturing
[25:16] and and help American companies? Yeah.
[25:18] Versus they are explicitly saying
[25:20] there's a security concern from XYZ
[25:24] nation internationally and that's what
[25:26] we're doing right because those are you
[25:28] know related but maybe two different
[25:29] things. It's one thing if you say hey we
[25:31] just want our companies to flourish.
[25:33] That's kind of what tariffs are very
[25:34] protectionist type you know approach.
[25:36] It's another thing if maybe uh the US
[25:38] government's relationship with Huawei
[25:40] where they explicitly said we believe
[25:41] that there are security concerns and we
[25:43] want to target this. How do you look at
[25:45] the regulation that's getting put
[25:46] forward or legislation? I
[25:47] >> I I think both are important things that
[25:50] legislators currently care about. I mean
[25:52] you you saw it with Tik Tok, right? I
[25:55] mean that originally being from a
[25:58] Chinese related parent was a really big
[26:00] deal and
[26:02] the US government didn't want certain
[26:06] governments to have that sort of power
[26:08] or potential surveillance capability.
[26:11] And so it made them give it up. And so
[26:15] that that is 100% going to be a big
[26:18] concern is this national security risk.
[26:20] if you have robots not just in the
[26:22] Pentagon but in people's homes or in
[26:24] important businesses that it's a risk.
[26:27] Um and so the other part you asked about
[26:31] was around uh encouraging and
[26:35] incentivizing the current domestic
[26:37] industry, right? There's a lot of ways
[26:39] to do it. Uh you know you can invest in
[26:42] the companies themselves like the
[26:43] government has done with Intel. Uh you
[26:45] can give them loans at v very favorable
[26:47] terms. you can uh establish uh more
[26:51] prevalent education programs for people
[26:53] to get involved in you know robot
[26:55] related trade uh or manufacturing
[26:58] related trade and I think all of these
[27:00] are really important. I I would I would
[27:03] say
[27:04] we should be as aggressive as possible
[27:06] because in terms of policy we are
[27:08] lagging a little bit behind. Uh but I
[27:11] think we're going to get there very
[27:12] soon. What do you think the US
[27:13] government can do to encourage more uh
[27:16] success in the human robot space?
[27:18] >> I think they should just be directly
[27:20] investing in some of these robotics
[27:22] companies. I mean, the reason why we
[27:23] started Robo Strategy was because we
[27:25] felt like one, there was a need to
[27:28] invest billions of dollars of capital
[27:30] into the most important robotics
[27:32] companies of the future. And two, there
[27:34] was an opportunity to do so as well. And
[27:38] the thing is, we've never had companies
[27:40] or venture capital firms with the
[27:42] experience to do this. And so, it's a
[27:45] completely new underwriting game for
[27:46] them. They have to step out of their
[27:48] comfort zone. They have to get
[27:49] comfortable with all the execution risk,
[27:52] with all the dilution risk of potential,
[27:54] you know, future capex buildouts. And
[27:57] it's it's something that if the US
[28:01] government doesn't step in, it could
[28:03] take an elongated amount of time for the
[28:05] de for the industry to develop as as
[28:07] fast as it should.
[28:09] >> Let's talk about uh general purpose
[28:11] versus specialized uh kind of
[28:13] applications. And in my opinion, if we
[28:15] look at software AI, we're seeing this
[28:17] play out now, right? There are the
[28:19] general purpose models, the large
[28:20] language uh labs. They are spending
[28:22] immense amount of resources to go and
[28:24] really have one model that can do a lot
[28:27] of different things. I don't think
[28:28] there's anybody in the world who is
[28:29] denying that they are having success.
[28:31] >> At the same time, there are very smart,
[28:33] very capable, very wellunded people who
[28:35] are saying, well, we actually think that
[28:36] there's a huge opportunity for
[28:38] specialized workflows using AI.
[28:40] >> My general take on the software side is
[28:43] that both are going to have a place and
[28:45] both will be successful depending on
[28:46] what the end use case is.
[28:48] When I look at the robotics space, I'm
[28:50] nowhere near as uh uh kind of down the
[28:53] rabbit hole as you are, but I know of
[28:55] companies like physical intelligence,
[28:56] which maybe is a little bit more
[28:57] specialized application compared to a
[29:00] figure oric or a 1x that are much more
[29:02] focused on, you know, kind of general
[29:03] application.
[29:05] >> One or the other wins, they both will be
[29:07] successful. How do you look at general
[29:09] versus specialized?
[29:10] >> Yeah, I I think this is a dynamic that
[29:12] has played out in a lot of different
[29:15] areas of history, right? If we look at
[29:18] our phone, we talked about this earlier,
[29:20] they replace your GPS, your MP3 player,
[29:23] uh you know, your recording devices,
[29:25] etc. Um, but those devices still exist
[29:27] to some extent, right? You know, the
[29:29] watch industry is still a thing even
[29:30] though you can tell time on your phone.
[29:33] Uh, and another parallel I like to look
[29:35] at is the the GPU versus the ASIC
[29:37] industry. Both both are massive. Both, I
[29:39] would say, are in the trillions of
[29:40] dollars. But
[29:44] I I think some people they kind of go to
[29:47] this um line of thinking where they
[29:50] believe that in a perfect world
[29:52] everything should be specialized for a
[29:54] specific purpose and I I don't think
[29:56] that's the case and you you can look at
[29:57] these previous examples to kind of um
[30:02] you come come to that similar
[30:03] understanding right but the the the
[30:07] thing with having something general
[30:08] purpose is that while it might not be
[30:11] the most efficient from, you know, what
[30:14] one certain metric. It is something that
[30:17] needs to be made at an incredible scale.
[30:19] And so you you have these massive uh
[30:22] unit economics that are working in in
[30:24] your favor that you wouldn't have for
[30:26] something that you're producing for one
[30:28] specific application, right? And that
[30:30] means I can produce something maybe at
[30:32] 80% cheaper cost than I would otherwise.
[30:35] And that is something that I think is
[30:37] really underappreciated. Um
[30:40] I I think there will still be
[30:42] specialized robots. I think we'll have
[30:44] specialized robots for welding, for
[30:46] example. We invested in in path
[30:47] robotics. Uh there are going to be
[30:50] specialized robots for you know home
[30:53] construction in different areas. And so
[30:56] there's there's an area for for both of
[30:58] these and I think they're both going to
[30:59] be humongous markets that you know we're
[31:01] going to play. And I think it it really
[31:02] depends on the context which robot makes
[31:05] sense. Let's talk about the training
[31:07] data that goes into actually creating
[31:09] these. Um, I think a lot of folks have
[31:11] seen some of the demo videos where a
[31:13] human is doing something and then the
[31:15] humanoid is essentially mirroring that
[31:17] action. Um, we've also seen there's a
[31:19] startup here in New York City that now
[31:20] is offering to come and clean your home
[31:22] for free. Yeah.
[31:23] >> But they want to wear cameras and
[31:25] basically collect all of that training
[31:26] data while they're cleaning your home.
[31:28] So, you get a free home cleaning, they
[31:30] get the data, they'll then use that. How
[31:31] do you look at the training data and
[31:32] where this is going to come from? Is is
[31:34] it possible to go back to the last
[31:35] question because I think there's a
[31:36] >> good
[31:36] >> there there's an important point which
[31:38] is the the world is never static which
[31:42] is why general purpose devices are so
[31:45] useful like GPUs right algorithms change
[31:48] all the time AS6 you have to develop a
[31:51] completely new ASIC for it but the GPU
[31:53] is still going to be useful and the same
[31:55] thing is with anything that involves
[31:58] human labor if I'm working in a factory
[32:01] I might have to produce a completely a
[32:03] different type of product next year,
[32:05] right? And that involves a different
[32:07] manufacturing process and sometimes I
[32:09] can use the same machines. But the
[32:11] reason why humans are still used in
[32:12] factory environments today, even though
[32:14] a lot of the labor is repetitive, is
[32:16] that humans can adjust for different
[32:18] workflows and the world's always going
[32:22] to be changing, right? We're always
[32:23] going to have different environments,
[32:24] different objects that we have to
[32:26] interact with and different situations.
[32:29] And so you you need something that that
[32:30] is maximally adaptable and you know that
[32:32] looks like a humanoid or maybe something
[32:35] that is similar to a humanoid. Maybe
[32:37] it's stationary. Maybe it's just a fixed
[32:39] general purpose arm, a cobbot industrial
[32:41] arm or a wheeled humanoid, right? But
[32:43] something that is flexible and can do
[32:45] multiple different tasks. Hello
[32:47] everyone. Sorry to interrupt this
[32:48] conversation, but I wanted to tell you
[32:49] about CFO Sylvia. This is an AI product
[32:52] that we built to help you better manage
[32:54] your finances. I had a problem. I wanted
[32:56] to use the latest artificial
[32:57] intelligence technology, but it didn't
[32:59] have the context of my personal
[33:01] portfolio. That's why we built Sylvia.
[33:03] Today, you come in, you can attach your
[33:05] accounts, you can start to ask Sylvia
[33:06] questions, and the more you use Sylvia,
[33:08] the data suggests, the faster you will
[33:10] grow your net worth. CFO Sylvia is the
[33:12] most powerful personal finance product
[33:14] that I've ever used. And our team is
[33:16] hard at work making it better every
[33:17] single day. Go check it out today at
[33:19] cfossilia.com.
[33:21] Again, that's cfossilia.com.
[33:23] And if you start using this product and
[33:25] you let me know how we can improve it,
[33:26] we'll make the changes right away. All
[33:28] right, guys. Let's talk about something
[33:30] that's actually moving the needle in
[33:31] crypto right now. And that's my guys
[33:33] over at Arch Public. These guys are
[33:35] killing it. They continue to lead the
[33:37] way in everything that's got to do with
[33:38] aentic trading. And everyone knows this.
[33:40] Nothing else is being talked about in
[33:41] the market other than let your AI agents
[33:43] trade whatever asset you want. Now, with
[33:46] the addition of Archai, the Market Wave
[33:48] algorithm, and their tax harvest tool,
[33:50] they have turned this into a true
[33:51] one-stop shop for automated trading.
[33:53] They have now over 25,000 users. That
[33:56] includes several corporate treasuries,
[33:57] asset managers, and businesses. And so,
[33:59] the testing phase for Arch Public, that
[34:01] is over. They have perfected how to
[34:03] automate advanced trading strategies
[34:05] without ever taking custody or asking
[34:07] you for your private keys. It's a big
[34:09] deal, obviously. So, Arch Public is an
[34:11] action speak louder than words type
[34:13] company. That's what I like to see. They
[34:15] give every user access to their platform
[34:17] absolutely free. You get to test it. You
[34:19] run the strategies. You watch the
[34:21] performance in real time. And then only
[34:23] after all of that do you put real money
[34:25] in to go put it to work when you're 100%
[34:27] comfortable. That's it. No pressure, no
[34:29] hidden fees, no gotchas, none of that
[34:31] nonsense. Go to archpubic.com right now
[34:34] and you can get started for free. Once
[34:36] you see how you can accumulate and
[34:37] manage crypto this way, you'll never
[34:39] trade the same again. archpubic.com. Go
[34:42] check them out today.
[34:44] Now when you look at the training data
[34:46] um let's talk about where that comes
[34:48] from right we've obviously seen examples
[34:50] where humans are standing in front of a
[34:51] humanoid they do some task the humanoid
[34:53] then mirrors it um we also have seen
[34:56] though there's a recent company in New
[34:57] York City that now is offering to come
[34:59] clean your apartment for free
[35:01] >> but in exchange for you getting a free
[35:02] apartment cleaning they wear cameras and
[35:04] they collect live data and then they
[35:07] basically are using that for training
[35:09] data
[35:10] >> is this just going to become like a
[35:11] global scavenger hunt for whoever can
[35:13] find the best training data or do you
[35:15] think that there will be more specific
[35:17] we are trying to create data for these
[35:19] humanoids? Yeah, I mean this data topic
[35:22] is is really interesting because there
[35:25] used to be this kind of whole train of
[35:27] thought where people were comparing it
[35:30] to LM right and LM were trained on this
[35:33] corpus of internet um the entire corpus
[35:36] of internet history um and people right
[35:40] made the comparison that hey like we
[35:41] don't have this equivalent for robotics
[35:44] and we have to recreate all of that from
[35:46] scratch and it's going to take a lot of
[35:47] time it's going to take a lot of labor
[35:50] And I I think some recent de development
[35:54] developments in the model world have
[35:56] changed my thinking on this one is the
[36:00] models that have basically um started to
[36:04] use video generation models as a
[36:06] backbone like sometimes also called
[36:08] world models and these models are
[36:10] basically just trained on internet video
[36:12] data for the most part right uh think
[36:14] about you know Sora or uh Juan for
[36:18] example It's the one of the best Chinese
[36:20] open- source video models. Um,
[36:24] while these aren't the same type of data
[36:27] you would get from, you know, someone
[36:28] putting a camera on your head and and
[36:30] cleaning your room, this is also really
[36:34] important data to train a model that has
[36:37] an understanding of of the world, right?
[36:39] And has an understanding of physics, has
[36:41] an understanding of, you know, fluid
[36:43] dynamics. So, if I tip this glass over,
[36:45] right, like how does the water move and
[36:47] how objects interact with each other and
[36:51] we we have a lot of that data on the
[36:53] internet and that has sh been shown to
[36:56] be really really useful as a as a
[36:58] backbone for robot foundation models.
[37:00] And so that doesn't mean necessarily
[37:03] that that's the only data that we need
[37:05] and we don't need anything else. We
[37:06] still need a lot of robot environment
[37:09] specific data. We need, I would say,
[37:12] data that doesn't exist on the internet
[37:15] of people doing certain tasks or in
[37:17] certain environments that are not
[37:20] usually recorded, right? So, if I'm
[37:23] uh producing some like a book in a
[37:26] factory, uh there's not really much
[37:29] video of that on the internet. And so,
[37:32] some of that data is going to be needed
[37:34] is is going to need to be collected. I
[37:36] think there's a question around what is
[37:38] the scale that needs to be collected. I
[37:41] think it's it's hard to say, but it it's
[37:43] a lot less I think than what previous
[37:45] people would have previously thought.
[37:47] >> Now, when I think of these humanoid
[37:49] robots, it seems like there's already uh
[37:51] somewhat of a Trojan horse into many
[37:53] large companies. Right? If I look at a
[37:55] Tesla car being built, they don't look
[37:57] like humanoid robots, but it is pretty
[37:59] much a robotic uh assembly line, and
[38:03] that's how those cars get created. If I
[38:04] look at Amazon, Amazon has about 1.5
[38:07] million humans that they employ, but
[38:09] they report that they have 750,000
[38:11] robots at the company. And my guess is
[38:13] that there's going to be more robots
[38:14] than humans at some point in the future.
[38:16] >> The videos that I've seen online, I
[38:18] don't see any humanoid robots walking
[38:20] around yet.
[38:21] >> I see a lot of robots that are
[38:22] autonomously moving things within their
[38:24] different warehouses and and doing
[38:26] different tasks. But then if we go to
[38:28] like self-driving cars, again, it seems
[38:30] like there's a rise of what I would
[38:32] consider a version of a robot,
[38:34] >> not a humanoid robot. And so, is this a
[38:36] thing where robots end up being kind of
[38:39] the the Trojan horse? It pushes through,
[38:41] people start to adopt this technology,
[38:42] and then actually humanoids are the last
[38:45] but biggest market, or how do you see
[38:46] the relationship between, you know,
[38:48] nonhumanoid robots versus the actual
[38:50] humanoids themselves? I I think look um
[38:55] designing a robot to do a specific task
[38:57] is I think one specific task is going to
[38:59] be a lot easier than u training a robot
[39:02] to do 50 different tasks and the
[39:05] humanoid robot you would expect to be
[39:07] able to do a lot of different tasks to
[39:09] kind of fulfill its its kind of desired
[39:11] purpose and so from that perspective yes
[39:14] I think you're going to see some more
[39:15] adoption of these um application
[39:19] specific robots a little bit earlier
[39:21] than for humanoids, but I I think the
[39:25] development of humanoids is it's it's
[39:28] it's exponential. Um,
[39:32] >> will humanoids be making other
[39:34] humanoids? And so, it like legitimately
[39:36] is exponential.
[39:38] >> That is happening in the very near
[39:40] future. I think multiple companies have
[39:42] hinted at that before. You know, Tesla
[39:45] has hinted at that before. I think
[39:46] they've shown some videos of Optimus
[39:49] organizing, uh, you know, battery pieces
[39:52] within their facility. Uh, you know,
[39:54] figures hinted at it before. Uh,
[39:55] Electronic has talked about that before
[39:58] with their partnership with with Jable.
[40:00] Um, and so that's going to be part of
[40:01] it, right? It's the kind of like the the
[40:04] recursiveness of it is you with AI or
[40:08] with LMS, right? You have the the AIs
[40:12] being able to help with AI research and
[40:14] here you have the robots able to help
[40:15] with robot development and then you also
[40:17] have right the latest models opus 4.8
[40:22] being able to help with robot AI
[40:24] training as well.
[40:25] >> Let's talk about these humanoids in
[40:27] someone's everyday life. You know, one
[40:29] of the things that I continue to say is
[40:30] I personally believe
[40:32] >> humanoids are going to be cleaning in
[40:33] your home.
[40:34] >> They're going to walk your dog. They may
[40:36] even watch your children one day. You
[40:39] know, I I have young children. Uh
[40:41] finding a babysitter sometimes is hard,
[40:43] right? Um maybe I wouldn't be
[40:45] comfortable if they were awake, but if
[40:47] they're asleep or something, right?
[40:48] Like, but just talk about like what do
[40:49] you expect the average person in, I
[40:52] don't know, 10 years, their interaction
[40:54] with these to be, and how much are we
[40:56] going to trust them to replace tasks
[40:58] that maybe today we would only allow
[40:59] another human to do?
[41:01] >> Yeah. I mean this this kind of goes back
[41:03] to also some of the debate around teleop
[41:06] right somebody controlling a robot from
[41:09] say you know a remote place versus fully
[41:11] autonomous
[41:12] >> for those that don't know tea operation
[41:14] is essentially there is the true
[41:16] autonomy where the robot has
[41:18] computational power on it uh actual
[41:20] device it's moving throughout the world
[41:21] there's nobody who can externally
[41:23] control it or or tell it to do certain
[41:25] things operation is the exact opposite
[41:27] right it's almost like a remote
[41:28] controlled robot that somebody else is
[41:30] controlling
[41:30] >> right right maybe somebody has, you
[41:32] know, an exoskeleton of of a humanoid
[41:34] and they're moving around in, you know,
[41:35] some remote work center and then they're
[41:37] controlling the robot and there are that
[41:41] is happening in some use cases as a way
[41:44] for people to kind of bootstrap adoption
[41:46] or gather training data. Um, and
[41:50] the the issue with that having that in
[41:52] your home though, right, is you have
[41:53] somebody that can see and hear
[41:54] everything that's going on. You kind of
[41:56] lose your your personal space. And so
[41:58] I'm a little bit skeptical of that
[42:01] approach for the home first,
[42:02] >> the teleoperation,
[42:03] >> the tele operation approach for the
[42:05] home, which there are some companies um
[42:07] that are kind of leaning on it a little
[42:09] bit more. Um but for when these things
[42:12] become autonomous, I think it's going to
[42:13] be similar to, you know, how people
[42:15] interact with uh these chat these chat
[42:18] AI apps today with chat GBT and Claude,
[42:20] it's right like a lot of people tell
[42:21] them their personal information, some
[42:23] things they wouldn't even tell their
[42:25] significant friends. And so I think
[42:27] people will eventually have that type of
[42:29] relationship with their robots as well.
[42:31] >> I do think it's interesting for like a
[42:33] child to grow up uh in a home where
[42:35] there's a robot that they depend on for
[42:37] certain things. You know, it does enter
[42:39] this world where what is the
[42:40] relationship between man and machine uh
[42:42] starts to get blurred a little bit more
[42:44] than uh than not. And I I know because I
[42:46] have friends who will tell me that their
[42:48] kids sit and talk to the Alexa.
[42:50] >> Mhm. and just talk, you know, talk and
[42:52] Alexa will answer and it's kind of, you
[42:54] know, a proxy for a human and maybe it's
[42:56] not the, you know, most intelligent
[42:57] conversation or uh something where the
[43:00] the Alexa is actually able to replace
[43:02] the human conversation, but for a young
[43:04] child, you know, it does what it needs
[43:06] to do. And so you could easily see
[43:08] humanoid doing the same thing in, you
[43:10] know, kind of the physical realm, right?
[43:11] >> Yeah. And in some way, maybe they do it
[43:13] better than a human, right? Because, you
[43:15] know, they're trained to be to, you
[43:17] know, be very engaging, right? to uh
[43:20] understand us and to say things that we
[43:24] enjoy hearing. So yeah,
[43:27] >> now let's talk a little bit in terms of
[43:29] uh these companies. We obviously have
[43:31] talked about uh figure AI. I think
[43:32] that's probably the largest private
[43:34] company. Many people have probably heard
[43:36] of the company if if they don't uh
[43:38] already know it uh to some degree of
[43:39] familiarity. Uh Tesla is another
[43:42] company. They're in the public market.
[43:44] uh they have both the self-driving cars
[43:46] and they have uh the humanoid robots
[43:48] that they're going after. My external
[43:50] view seems that really what they are
[43:52] trying to do is build these AI models,
[43:54] machine learning, and just help robots
[43:56] see or think. And whether that's a
[43:58] self-driving car or the humanoid robot,
[44:01] that is kind of the pitch of that
[44:02] business. What do you think about Tesla
[44:04] and do you think that they can win the
[44:06] market? I
[44:07] >> I think undoubtedly Tesla is going to be
[44:10] a big winner in humanoids. I mean, I
[44:12] would never bet against Elon. I just
[44:14] think he just he's a winner. Um, at the
[44:17] same time, um, you know, you look at the
[44:20] Optimus program. Um, I think they have
[44:23] hit a few speed bumps over the past one
[44:26] or two years. Uh, things aren't going as
[44:28] fast as expected, but you know that this
[44:30] is a really hard problem to solve. Um,
[44:33] and
[44:35] I I would say on the AI side, um, they
[44:38] haven't shown as much as some of the
[44:41] other companies. I think they'll be able
[44:43] to get there because Elon is going to be
[44:45] able to do it. But for example, for
[44:49] figure, right, they showed this demo of
[44:51] the robot sorting packages for what was
[44:54] it? At first, it was eight hours and
[44:56] then they extended it to multiple robots
[44:58] to work something like nine days
[45:00] straight. Mhm.
[45:01] >> Um that is, I would say, a really
[45:04] impressive feat of their autonomous
[45:06] capabilities that very few other
[45:08] companies have shown. Um and knowing
[45:12] Elon,
[45:14] if he's got something great to show, I
[45:16] think he he'd show it, right? And so I
[45:18] think because we haven't seen that yet,
[45:19] maybe they're not there yet. Um on the
[45:22] other hand, no one's going to beat him
[45:24] in manufacturing um in America. And on
[45:27] the hardware design side, I I think the
[45:30] like taste and style is going to be
[45:32] really important outside of, you know,
[45:34] your functional requirements. And he's
[45:36] going to be able to sell a premium
[45:37] product based on that. And um, you know,
[45:41] there are some, I think, issues with the
[45:43] hand. Not issues, but I would say need
[45:47] for redesign, right? The hand is super
[45:49] complex, one of the most complex parts
[45:50] of the robot. And you need to make sure
[45:53] that even though you have a lot of
[45:55] really small components that they can
[45:57] withstand wear and tear for years and
[45:59] years and years and that they can be
[46:01] very precise because everything you're
[46:02] doing with their hand is very very
[46:03] precise. And so, you know, I think
[46:06] they're still going through design to to
[46:09] really perfect that. Um whereas maybe
[46:12] some companies are a little bit ahead,
[46:13] some are a little bit behind. But that's
[46:16] kind of how I would characterize Tesla
[46:18] at the moment. What's interesting to me
[46:20] is I've always talked about humans have
[46:22] an advantage over the robots for now,
[46:24] which is these five fingers. You know,
[46:26] it is the one thing that every single
[46:28] company I've talked with that are
[46:29] building these humanoid robots, it's the
[46:31] hardest part in many of their view. And
[46:33] to a human, it's natural. You know, you
[46:35] just pick something up, you don't even
[46:36] think about it. And so, what at what
[46:39] point do you think that humans should
[46:41] start to worry about maybe the negative
[46:43] impact of this? Do you think there's a
[46:44] negative impact on jobs or the roles
[46:46] that humans will do in the employment
[46:48] sense?
[46:49] >> Yeah. Yeah, absolutely. I I think some
[46:52] tech CEOs might sugarcoat this a little
[46:54] bit and you know they might relate it
[46:56] back to the history, right? This is what
[46:58] we kind kind of constantly hear as a
[47:00] talking point that hey look, every time
[47:02] in human history where we had this big
[47:04] technological revolution that put people
[47:06] out of jobs, there was always something
[47:07] new that kind of came about, right? I
[47:09] think the difference here is that
[47:12] previously
[47:14] you just had to move up the cognitive
[47:15] stack. You had to do something that
[47:16] involved more thinking or more planning
[47:19] that you know whatever had replaced you
[47:21] wasn't able to do. And now with AI
[47:25] basically replacing the complete
[47:26] cognitive ability and physical ability
[47:28] of a of a human there there's no place
[47:30] to kind of expand out to. Right? I don't
[47:33] think there's going to be a certain task
[47:34] that you can say, hey, look, um, a human
[47:36] can do this, but a robot won't be able
[47:38] to do. And I I think from that point of
[47:41] view, we need something like universal
[47:43] basic income. Uh, I think there's going
[47:44] to be a lot of people out of jobs, which
[47:46] is unfortunate, but there's also nothing
[47:48] we can do to stop technology. And so, we
[47:51] we're going to need safety nets from
[47:53] governments around the world to deal
[47:55] with this. And is your thought process
[47:56] that if humans are going to get
[47:58] displaced by humanoid robots that UBI or
[48:02] something like that is the solution?
[48:04] >> It's hard to say what's the optimal
[48:05] solution, but it's one of the best
[48:06] solutions I think that um people have
[48:08] come up with so far.
[48:10] >> Do you think that it is more likely
[48:12] software AI or hardware AI replaces
[48:15] human workers?
[48:18] >> Well, I think you look at the makeup of
[48:20] um you know, white collar work versus
[48:22] blue collar work, right? It's, you know,
[48:24] for people like you and me, maybe we're
[48:25] in in these in these bubbles and we
[48:27] think, okay, it's it's mostly white
[48:29] collar jobs, but there's this huge swath
[48:31] in group of people that are doing more
[48:33] physical labor jobs. Um, I would say
[48:35] that's that's that's probably bigger
[48:37] than white collar jobs. Um, especially
[48:39] in other parts of the world. Um,
[48:42] and
[48:45] so I I think it's it's going to be huge
[48:47] for both. It's it's you know, I don't
[48:49] think it's important to say which one's
[48:50] going to be bigger. just going to
[48:52] destroy a lot of jobs.
[48:53] >> What I find very interesting is uh we're
[48:55] seeing on the software side people are
[48:57] starting to now write maybe their
[48:58] technical documentation and have it
[49:00] optimized so that AI agents can read it
[49:02] not just humans or you see other aspects
[49:05] where people are actually building
[49:06] whether it's their products or you know
[49:08] checkout flows or or whatever so that
[49:10] agents are able to interface with it
[49:12] easily.
[49:13] >> In the physical world we do not yet see
[49:16] people changing anything because of
[49:19] humanoid robots. My guess is that that
[49:21] will actually happen though. And so if
[49:23] we think of a factory right now, it's
[49:25] pretty much there's a human factory and
[49:27] then there's some that have robotics.
[49:29] >> Yeah.
[49:29] >> Do humans and robots work together? Do
[49:32] we get like a human, you know, facility
[49:34] and then there's an entirely different
[49:36] robotic facility? How do you think about
[49:38] the physical world really conforming and
[49:40] changing and evolving because of the
[49:43] rise of human robots?
[49:44] >> Yeah. So I I mean we've had robots
[49:46] working alongside humans for decades. Um
[49:50] some of them you might not really think
[49:52] about as robots cuz you're so used to
[49:54] them, right? Like coffee machines for
[49:56] example. Um other ones that look a
[49:59] little bit more like robots.
[50:01] Um they're essentially right called
[50:03] cobots. They're just large industrial
[50:05] arms that are a little bit smaller and
[50:09] they're programmed in a way that is safe
[50:12] to work around humans. And that has
[50:13] existed for I would say 20 years or so.
[50:17] Um and these exist across a lot of
[50:20] factories in the world. But the issue is
[50:22] that that the adoption hasn't been high
[50:24] right because
[50:26] um yeah I mean if you look at just all
[50:28] the amount of robots that are installed
[50:30] on an annual basis right now it's around
[50:31] 500,000 globally. Um a big chunk of that
[50:35] is is in China. um and a very small
[50:38] fraction of that is is in America even
[50:40] though we still do have a lot of you
[50:42] know factories and places where people
[50:44] are doing physical labor in in America.
[50:46] It's just the issue with programming a
[50:48] robot um creating the environment around
[50:52] it and making sure that it's working
[50:54] reliably right you know around the clock
[50:58] for the entire year is really really
[50:59] expensive. Oftent times it's more
[51:02] expensive than the cost of the of the
[51:03] robot itself. But when we make these
[51:05] robots smart, right, that cost of
[51:07] deployment goes down dramatically and
[51:10] these robots become a lot more
[51:11] economical and they make a lot more
[51:12] sense and a lot of different places in
[51:15] the world.
[51:16] >> Let's talk about what you're doing in
[51:17] the space. We talked earlier, you've
[51:19] made some very large personal
[51:20] investments, but uh you recently
[51:22] announced that now you are uh taking
[51:24] those personal investments, you're
[51:26] contributing them into a closedin,
[51:28] publicly traded fund uh called Robo
[51:31] Strategy. talk a little bit as to why
[51:34] step out of the shadows to some degree
[51:36] uh and go down this path of operating uh
[51:39] this entity.
[51:40] >> Yeah, I just want to clear up that I've
[51:42] contributed some of of my investments in
[51:44] robotics. I think it's something around
[51:47] 15 to 20% of my figure AI exposure and
[51:51] um you know some fraction of my
[51:53] electronic exposure but my exposure
[51:57] outside of the fund is is currently more
[52:00] than the exposure from the NAV basis
[52:02] inside the fund. Um, but I don't hope
[52:05] that to be the case, right? I hope I
[52:07] grow this as big as I can. Um and one of
[52:11] the reasons why you know I went down
[52:13] this path you know I could have
[52:15] continued to just invest privately was
[52:18] because when I went down this journey in
[52:21] the beginning you know 2 or 3 years ago
[52:24] I saw that the venture space just wasn't
[52:26] appreciating the industry and that there
[52:28] was not only going to be a really big
[52:30] need for more capital allocators to step
[52:32] in um but it was an opportunity to do it
[52:36] at really big scale and so you saw What
[52:39] happened with OpenAI and Enthropic,
[52:42] right, is these companies just went
[52:45] completely nuts over the period of a few
[52:47] years. And I think you're going to have
[52:50] that same vertical takeoff happen for
[52:52] robotics as well. You're going to be
[52:54] able to put tens of billions of dollars
[52:56] to work in a very high risk adjusted
[52:59] manner, right? If you had just put all
[53:02] your money into the top few AI companies
[53:05] a few years ago, you would have
[53:06] outperformed the best seed funds in the
[53:08] world. Um and so
[53:12] this is a very I think special
[53:14] opportunity in time and
[53:18] you know we're going to have the figure
[53:20] or we're going to have the anthropic and
[53:21] the open AI for robotics as well. Um and
[53:25] I think the only way to be able to reach
[53:26] that scale is to do it through the
[53:28] public capital markets. Um you know we
[53:31] had never raised um capital in the
[53:34] private markets before for you know an
[53:36] actively managed fund. you know, we've
[53:38] kind of kicked the idea around a little
[53:41] bit, but and you know, like maybe after
[53:44] some effort, we could get, you know, one
[53:46] or two billion dollars, maybe a little
[53:48] bit more if we tried really hard. Um but
[53:53] I I think the opportunity in in the
[53:56] public markets is I would say in terms
[53:58] of demand for robotics exposure, high
[54:00] quality robotics exposure, it's probably
[54:02] on the scale right now of tens of
[54:04] billions, very soon hundreds of billions
[54:06] in in the future, trillions, right? And
[54:09] there's no place for all that capital to
[54:11] get exposure. And so, you know, we over
[54:15] the past few years have been talking
[54:16] about our views on markets, our views
[54:18] on, you know, this industry. And we were
[54:21] bombarded with people that were asking
[54:22] us, how do we get exposure to the space?
[54:24] What stocks and what companies should we
[54:26] be looking at? And there was no answer
[54:27] except for Tesla. And so we thought, why
[54:30] not create this vehicle, right, for not
[54:33] only for people to get exposure, but for
[54:35] us to be able to scale this into what I
[54:39] would consider potentially, you know,
[54:41] one of the biggest investors, robotics
[54:43] investors in the world, but also
[54:44] potentially one of the biggest VC funds
[54:46] in the world, right? VC funds have
[54:48] traditionally not been public. There was
[54:51] a very brief history of them being
[54:52] public in, I would say, like the 1960s.
[54:55] Uh there were some issues with it, but
[54:56] then for the most part it just went all
[54:58] private. And then you had these
[55:00] companies, right, that before they went
[55:03] public at a few billion dollars. Now
[55:05] they wait until they're trillion dollar
[55:07] plus tens of billions of dollars in
[55:09] revenue to go public. And all that value
[55:11] creation right is is going to in the
[55:14] hands of a few venture capital funds
[55:17] their LPs and you know their network
[55:19] which is really smart small part of the
[55:21] world but
[55:23] there's right the public capital markets
[55:25] which is much much bigger and it it has
[55:28] interest right it wants to participate
[55:30] in in this value creation and ride along
[55:32] in an investment journey and so why not
[55:35] take a investment fund public and so I
[55:39] think this model.
[55:42] If we're able to execute on it
[55:44] correctly, we we could kind of reshape
[55:46] the whole paradigm of venture capital
[55:48] investing as well. I think you're going
[55:49] to see other venture funds start to go
[55:51] public. There are a few funds that went
[55:52] public earlier this year. Um but I think
[55:55] and they they've done well in terms of
[55:57] demand, right? They're all trading at
[55:59] premiums to NAV. uh the the difference
[56:03] is I don't think a lot of them have
[56:06] really zoned in and figured out this
[56:08] model of how do I take that demand and
[56:12] put it back into the vehicle in a way
[56:14] that's accretive for shareholders and to
[56:17] compound that over time and then rein
[56:19] continue to reinvest in in the best
[56:21] companies in the world and do it on an
[56:22] active basis right because if I want to
[56:24] put billions of dollars to work I need
[56:26] to have the relationship with the
[56:27] founder I need to have underwritten the
[56:28] deal that I'm comfortable putting that
[56:30] amount of capital to work. Uh I need to
[56:32] have um you know been with this company
[56:36] and following it for for many many years
[56:38] already. Uh and you know I want to I I
[56:42] also need to be somebody that these
[56:43] founders want on on the cap table as
[56:45] well. uh not just me, you know, the
[56:47] entire team that we've built at Robo
[56:49] Strategy and and so that is something
[56:52] that we're doing a little bit
[56:52] differently from uh some of the other
[56:54] publicly traded venture funds that that
[56:56] have launched and you know you might see
[56:59] some people try to copy in the future.
[57:01] >> Now let's talk about um as you have this
[57:03] vehicle uh let's go to the maybe atomic
[57:06] unit the most important thing which is
[57:08] the decisions that you are making. Which
[57:09] company are you allocating to? Um what I
[57:12] find very interesting about your team is
[57:14] you obviously are a great investor.
[57:16] You've got a track record over time of
[57:17] investing in some of these innovative
[57:19] type uh industries. Um but you've
[57:21] assembled I don't know one of the top
[57:23] robotics teams in the world in order to
[57:26] actually go underwrite this stuff. Talk
[57:28] a little about the team and what you
[57:30] guys do from a diligence standpoint when
[57:32] you guys are actually evaluating whether
[57:33] you're going to invest in a company or
[57:34] not.
[57:35] >> Yeah. Um, so look, to I think make the
[57:38] best investment decisions in robotics,
[57:41] you you need experience and I I can try
[57:43] to get as smart on the industry, you
[57:45] know, as I as I can. And I think I've
[57:47] done a pretty decent job at it. Um, you
[57:50] know, reading,
[57:52] you know, multitudes of of research
[57:53] papers, talking to people across the
[57:55] industry obsessively for the past two
[57:58] years. Um, but it doesn't replace
[58:01] decades of real industry experience,
[58:03] which is something that I think other
[58:05] venture capital funds, they might have
[58:07] in their specific domains, but I don't I
[58:10] haven't really talked to anybody else
[58:11] that has that in in the robotic space so
[58:14] far. Uh, and they need to understand
[58:16] what are the pitfalls, right, that these
[58:17] companies can go through. What are uh,
[58:21] you know, some key areas that they need
[58:23] to operate uh, to be successful? uh and
[58:27] uh you know who who who is the right
[58:29] talent? Like are these people that
[58:30] they've brought onto the team actually
[58:31] legitimate? Are they going to contribute
[58:33] to this company being a big success? And
[58:36] so you know I without understanding I I
[58:38] went out and I you know I found Scott
[58:40] Walter uh who is incribable talent. He's
[58:43] a little bit of a mini celebrity in the
[58:45] robotic space. Um, with the content that
[58:48] he posts on ex in and YouTube and lot
[58:52] almost every major humanoid CEO knows
[58:55] knows Scott uh has a lot of respect and
[58:58] some of his the content that he posts
[58:59] online has even informed some of the
[59:00] mechanical designs of um some of the top
[59:03] players in the space and you know he he
[59:06] founded two robotics companies. He sold
[59:08] the last one to Cuka. He sold both of
[59:10] them. the last one he sold the Kukah and
[59:13] uh you know 40 years experience is more
[59:15] than you know some people have their on
[59:17] their entire team. Um his his background
[59:19] is in uh what is it robot offline
[59:23] programming uh human mechanical humanoid
[59:26] mechanical design um manufacturing
[59:28] simulation. So these are all incredibly
[59:31] tangential you know fields of expertise.
[59:34] also have Jack Pearson on the team who
[59:36] you know was also a founder operator
[59:38] himself with a decade of experience in
[59:40] the indream and both of them have this
[59:42] kind of unique combination of both
[59:44] technical and commercial competency and
[59:48] founders find that really useful as well
[59:51] and you know we we're we're trying to
[59:52] structure the platform in a way that we
[59:56] provide we we believe we can provide
[59:58] value to the founders that we work with
[59:59] as well and so that is something that
[60:01] we're going to continue to do. We're
[60:02] going to continue to recruit more
[60:04] robotics veterans so they can help with
[60:07] you know supply chain with design of
[60:10] their robots for recruiting validating
[60:12] the right people uh etc.
[60:14] >> Now let's talk about the actual entity
[60:16] itself. Um a lot of people don't know
[60:18] this but uh closed end publicly traded
[60:20] funds have some pros and cons I think
[60:22] compared to other vehicles. Um and so
[60:24] one of the examples I've always used is
[60:26] Bill Aman. He has a closed end publicly
[60:29] traded fund in Europe
[60:30] >> and most people see he you know he
[60:32] managed $18 billion but majority of the
[60:34] capital is actually in the closed end
[60:35] publicly traded fund
[60:36] >> now in the United States you cannot
[60:39] charge carry y
[60:40] >> whereas with a traditional private
[60:42] venture capital fund you know if you
[60:44] were just a a regular VC you would say
[60:46] hey I'm charging you know 2 and 20% 2%
[60:49] management fee 20% carry an LP gives you
[60:52] the money you go and you invest and when
[60:54] you sell positions you're taking 20% of
[60:55] the profits. Yep,
[60:56] >> it's a pretty big number, right? Um,
[60:59] >> in the publicly traded closed end funds,
[61:02] you can only charge a management fee and
[61:04] so there is no carry there, but it's
[61:06] also permanent capital and so I think it
[61:08] changes the incentives. It changes the
[61:10] way that people think in terms of time
[61:12] horizon.
[61:12] >> Talk just a little bit as to why you
[61:14] chose the actual closed end publicly
[61:16] traded fund versus maybe doing this in
[61:18] some other format.
[61:20] >> I mean, there are only a few other
[61:22] formats that would work, right? One is
[61:23] we raise capital privately. that
[61:25] wouldn't get us to the scale that you
[61:27] know we we we want to get at. Um there
[61:29] is and that would I think just almost be
[61:32] just as hard. Uh whereas you know kind
[61:34] of like our competencies actually come a
[61:36] lot from from marketing. uh all of our
[61:39] team members have their own I would say
[61:42] you know large or at least have a
[61:44] meaningful presence on on social media
[61:46] and kind of understand how to generate
[61:48] more attention and awareness which is
[61:50] really important for a close-end fund
[61:51] structure because you're attracting
[61:53] capital from the public capital markets
[61:55] uh and then the other option right is is
[61:57] an ETF but with the ETF now you can
[61:59] actually have private assets in ETF but
[62:01] I think they restricted to around 15%.
[62:03] >> Right? And so we're dealing with illquid
[62:06] securities right and so the only way it
[62:09] really works is in this clos fund
[62:11] structure
[62:13] >> now what we've seen happen with many of
[62:15] these closedend funds is when you put
[62:17] private investments that are good
[62:18] highquality assets people want access
[62:20] right if you think about uh in the early
[62:22] days of Bitcoin people wanted to buy
[62:24] Bitcoin in the public market they
[62:26] couldn't do that and so the Bitcoin
[62:27] trust
[62:28] >> traded at a premium this is a story as
[62:30] old as time um talk a little bit as to
[62:33] the premium of which you guys trade at
[62:35] today is sometimes three 400% above NAV.
[62:40] What does that allow you to do? What
[62:41] what is kind of the plan here?
[62:43] >> Yeah. So I think a lot of people look at
[62:45] the premium and they get afraid, right?
[62:48] Because there is a risk which is the
[62:50] premium could compress um and but at the
[62:55] same time the premium premium could
[62:56] expand. Um I think what's important for
[62:59] us is that there's this aspect of
[63:02] premium crystallization which I think is
[63:03] not very well understood. Um which means
[63:06] that I can issue shares and I I I can do
[63:10] it in a way that is actually accreative
[63:12] to shareholders. And so for for example,
[63:16] right, if I have, you know, $100 of
[63:19] robotics equity in my vehicle and it's
[63:22] trading at a 3x nav and I issue
[63:26] 10% new shares, right now my fund size
[63:29] goes to $130. I issue one new share and
[63:33] so there's some dilution, but taking
[63:36] that into account on a per share basis,
[63:39] my per share value actually goes up 18%.
[63:41] Right? So net of the share issuance,
[63:44] you're actually accreating value. And if
[63:46] you do this over and over and over,
[63:49] right, that can compound and that can
[63:51] build over time. And that's what we've
[63:54] seen with um structures like Micro
[63:56] Strategy and Menoplanet where they did
[63:59] this at really large scale. I think
[64:02] Micro Strategy their NAV per share went
[64:06] from around $2 to $4 depending on how
[64:09] you account it to it's it's well the
[64:12] price is pretty volatile but around $150
[64:15] uh dollar now per share. There's around
[64:18] maybe 30 of that that comes from
[64:22] referred equity and so some credit but
[64:25] or there's around like 120 left over
[64:28] that just purely comes from the the the
[64:32] premium crystallization compounded over
[64:34] and over and over again over time. And
[64:37] some people think they look at the
[64:38] structure and they think oh no Michael
[64:41] Sailor he just levered up bought a bunch
[64:43] of Bitcoin it went up and that's where
[64:45] the value for Micro Strategy came from.
[64:47] If you look at their average Bitcoin
[64:48] acquisition cost, it's around 75,000.
[64:51] >> It's like right where we are.
[64:52] >> I think it's actually now above where we
[64:55] are, right? Bitcoin is like what 71,000
[64:57] today. And so they've actually lost
[64:59] money on their on their Bitcoin
[65:01] holdings. And so that is like that that
[65:04] is negative to the to the nav per share
[65:07] but their nav per share is still much
[65:10] much higher from when they first started
[65:13] because of the accretive nature of share
[65:15] issuance at a multiple. Um, and this
[65:19] type of structure is is actually people
[65:22] think it's foreign, but it's not very
[65:24] different than say the private to public
[65:27] company rollup model where you see
[65:30] companies like I don't know if you're
[65:31] familiar with Transdime, right? Big
[65:33] aerospace supplier or Constellation
[65:35] Software,
[65:36] >> Waste Management, a bunch of these guys.
[65:38] Yeah.
[65:38] >> Right. like they they have a company
[65:42] where they have earnings that trade at
[65:44] maybe a 15 to 40x multiple and then
[65:48] they find private companies that trade
[65:50] maybe at say a 3 to 8x multiple. They
[65:52] acquire them maybe with their own cash
[65:54] flow, maybe with equity issuance, maybe
[65:56] with debt, but that cash flow and that
[65:59] enterprise value, it's immediately
[66:01] rerated, right, to whatever the earnings
[66:03] multiple is on the public market
[66:06] >> uh cash flows. And so and then they do
[66:08] this over and over and over. And so if
[66:10] you actually look at the growth of a
[66:11] company like Constellation Software over
[66:14] time, only around 10 to 20% of that
[66:17] growth has come from the actual
[66:20] underlying business growth that's
[66:22] organic and the other 80 90% comes from
[66:25] basically M&A. It's them taking
[66:29] advantage of I would say not taking
[66:31] advantage that's the wrong word but them
[66:34] selling assets
[66:36] getting cost of capital at public equity
[66:39] prices and then
[66:42] being able to acquire businesses assets
[66:46] at private market prices which are
[66:49] different which I think is a little bit
[66:50] difficult for people to comprehend right
[66:52] because they think hey
[66:55] um you know this is the value of a
[66:56] company. This was like the last round
[66:58] price and this is what it should be
[67:00] valued at. But value differs, right,
[67:04] depending on who are the set of market
[67:06] participants that are valuing that
[67:07] asset. And so one thing I like like to
[67:10] compare it to is like if you just took a
[67:13] factory worker or a tech worker from
[67:15] China and put them in America, they have
[67:17] the same skills. They they'd be paid
[67:18] four times more, right?
[67:20] um if you know we took this building in
[67:23] Manhattan right now and we said hey look
[67:26] only people within a 10 mile radius are
[67:28] allowed to buy this the price would be
[67:30] much different than you know the price
[67:32] of say you let everybody in the world be
[67:34] a potential buyer for for for this asset
[67:37] and so I think that is some of the
[67:39] complication that people have to think
[67:41] about when they're thinking about public
[67:43] versus private marks and what the
[67:46] premium means because with private
[67:49] valuations you're working with a very
[67:51] small constrained set of market
[67:52] participants with public markets is much
[67:54] much larger and often times there's
[67:56] different valuations.
[67:58] >> If I was to describe someone what you're
[68:00] doing I think you know so a friend
[68:01] texted me I would basically say robo
[68:04] strategy is a publicly traded venture
[68:06] fund that's focused on investing in the
[68:08] best humanoid companies and related
[68:11] private companies. M the idea is that
[68:14] it's permanent capital and they can
[68:15] continue to recycle this into the best
[68:17] businesses over time. When it trades at
[68:19] a premium, they will then be able to
[68:21] take that premium, monetize it, raise
[68:24] more capital and deploy that back into
[68:26] the private markets and over time they
[68:28] should be able to compound capital by
[68:30] specifically focusing on capital market
[68:32] strategy and humanoid robot industry. Is
[68:35] that like a fair way to describe this or
[68:36] would you change anything?
[68:38] >> Yeah, I'd say it's relatively fair. only
[68:40] relatively we're we're focused on more
[68:42] than just humanoids, right? I mean, our
[68:44] north star is investment returns within
[68:46] the robot robotics space. And so that
[68:49] includes other robotics companies that
[68:52] are not building humanoids. One example
[68:54] is standard bots, right? Uh another
[68:56] example is Dino Robotics with their
[68:58] portfolio. I mean, they're building both
[69:00] hardware that look like yield humanoids
[69:02] and they're also really um they're one
[69:05] of the best researchers that that are
[69:06] building the robot brain. Um, and so and
[69:10] and there's all this stuff within the
[69:11] supply chain as well. I I think most of
[69:13] the value is going to acrew to the end
[69:14] of the supply chain, but there are
[69:17] definitely a lot of circumstances where
[69:20] there's going to be some big companies
[69:21] that are producing components. I mean,
[69:23] if you look at actuators, for example,
[69:24] right, they make up around 30 to 50% of
[69:27] the bomb cost of of a humanoid. And that
[69:29] space really hasn't significantly
[69:31] evolved for 50 years. So, there's a lot
[69:33] of area for innovation in certain areas.
[69:36] And you know we're keenly looking at
[69:39] pretty much everything that is
[69:40] tangential to the space to understand
[69:43] where we can make the best investments.
[69:45] >> Uh you mentioned standard bots. What do
[69:47] they do?
[69:48] >> So they make they make cobots. They make
[69:51] industrial arms. They're also building
[69:53] you know a semi-w wheeled humanoid right
[69:55] now. Uh I would say they're really the
[69:57] premier and only reasonably sized
[70:00] industrial arm manufacturer in America.
[70:05] And that is incredibly important if we
[70:07] think it's a big priority for America to
[70:10] re-industrialize, right? And so Evan is
[70:13] actually around New York. Um they have
[70:17] what I would say one of the most
[70:20] freshest takes on developing um robot
[70:23] arm technology because you know there's
[70:26] a lot of robot arms have been around for
[70:27] a long time, right? uh you know you have
[70:30] Fenuk uh you have Yasawa you have all
[70:32] these big robot arm guys but I would say
[70:35] they're stuck in this old school of
[70:38] traditional robot programming which
[70:40] looks very different from AI native
[70:43] robots of the future right and so some
[70:45] of the hardware needs a little bit look
[70:47] different uh maybe you need to have more
[70:49] torque sensing ability um maybe
[70:54] um you know uh you're you're you you
[70:57] have the programming interface not be so
[71:00] I guess complicated and it could be as
[71:02] simple as say I talk to the robot I tell
[71:05] it what what I wanted to do and it's
[71:06] able to do it um and and and so
[71:11] you know standard bots I I think that is
[71:13] going to be an an incredible company um
[71:16] should have them on
[71:17] >> let's talk about uh electronic um I've
[71:19] talked with the team there before I
[71:21] think it's one of the larger holdings uh
[71:23] in uh the fund that you have what is
[71:25] maybe their differentiator why are you
[71:27] excited about that
[71:28] Yeah, I mean if you look at Appronic,
[71:30] it's one of the longest standing uh
[71:33] humanoid companies out there. They're
[71:34] OGs. They've been at this for, you know,
[71:37] 9 plus years. Um and a lot of the more
[71:41] innovative actuator technology came
[71:43] from, you know, some of the researchers
[71:45] that were part of the founding team or
[71:47] currently work at Aptronic. um they
[71:50] actually were um contracted to create
[71:53] some of the prototypes for some of the
[71:55] leading uh humanoid companies that
[71:57] everyone knows about today. And uh you
[72:00] know then they decided hey look this is
[72:02] going to be such a great opportunity. we
[72:03] have all this experience, why don't we
[72:05] just build the humanoid and
[72:07] commercialize it it ourselves and and
[72:09] and so
[72:11] if you look at their death of experience
[72:14] then also you know the CEO leadership I
[72:16] think Jeff just recently hired uh the
[72:18] previous uh CPO of Whimo uh some top
[72:23] executives from um what was it Amazon
[72:27] and and some other big places and
[72:32] we talked about this before, right?
[72:33] Manufacturing and hardware are some of
[72:36] the two kind of competencies that we
[72:38] evaluate. And there are few other
[72:41] robotics companies even, you know, after
[72:43] we've invested in these companies 2
[72:44] years later that have come out and have,
[72:47] I think, been able to give the
[72:50] impression that they'll be able to
[72:52] actually commercialize millions of
[72:54] robots that actually work, right? that
[72:56] they're not going to break after uh 6
[72:59] hours of of work that they're going to
[73:01] be able to carry what they say they're
[73:03] supposed to be able to carry. Um and you
[73:06] know they also have a partnership with
[73:08] Google Deep Mind which I think is really
[73:10] special right cuz you know building the
[73:12] robot brain is going to be incredible
[73:14] tough incredibly tough and deep mind is
[73:17] um they have one of the deepest
[73:19] experience experiences in robot learning
[73:21] out of you know any any company out
[73:23] there. Let's talk about some of the
[73:25] misconceptions maybe are the critiques
[73:27] of uh of what you guys are doing. Um you
[73:29] know, if I was to play devil's advocate,
[73:31] uh one of the things I've seen people
[73:32] talk about online is like, uh you're
[73:34] just using the public as exit liquidity.
[73:36] You're taking some of your shares,
[73:37] you're putting it into this fund, uh
[73:38] you're hoping that they bid it up, and
[73:40] then you're just going to sell out of
[73:41] it, and you know, we're the dummies if
[73:43] we go and we buy this.
[73:45] >> I think it's just a very short-term
[73:46] minded um view set. I mean to do what we
[73:50] did in terms of um you know building
[73:51] building the team that we have and
[73:53] building the relationships with with the
[73:55] founders and just getting really smart
[73:57] on the space I mean it was an incredible
[73:59] amount of work and to make you know a
[74:02] few hundred million to take a company
[74:03] and then immediately sell your shares. I
[74:05] mean, there's a lot easier ways to make
[74:06] a few hundred million dollars, right?
[74:08] And um the opportunity that we're
[74:12] looking at here is is is to potentially
[74:14] create one of the largest venture
[74:16] capital funds in the world, right? And
[74:18] to reshape how venture operates as a
[74:21] whole. Like that that's what we're
[74:23] trying to do. Um and so that criticism,
[74:28] it just Yeah, it's it's it's it's it's
[74:31] hard to understand. I mean, there are so
[74:32] many ways that you you you can do it,
[74:34] right? could put your shares under like
[74:36] a different entity that is not
[74:38] associated with with your name at all.
[74:40] Um you could lock up all the other
[74:42] investors which you know we didn't do.
[74:44] Everybody was basically you know
[74:46] completely unlocked on on the first day.
[74:48] And the reason why we did it was because
[74:50] we saw all these other you know IPOs or
[74:52] companies projects going public and we
[74:54] saw what they did which is they locked
[74:56] up 90% of their shares right and what
[74:58] happens is it artificially inflates the
[75:00] price because price is a function of the
[75:03] supply and demand.
[75:04] >> Mhm. And but the issue is that supply
[75:07] has to come on the market someday,
[75:09] right? And it could be six months, it
[75:11] could be two years. And because people
[75:13] know that supply is going to increase by
[75:14] 10 times,
[75:17] they don't really want to buy. And when
[75:19] when people start unlocking, then they
[75:21] start selling. And then the share price
[75:22] kind of goes like this. And it's just
[75:23] impossible to kind of create any any
[75:25] momentum at all. And it's you're never
[75:27] going to be able to get over this this
[75:29] fear of, you know, extra shares entering
[75:31] the market. And so what we did was like
[75:33] why don't we just do the opposite of
[75:34] that
[75:35] >> right like just rip off the band-aid on
[75:37] the first day unlock all the shares and
[75:41] you know I would say most of the shares
[75:43] are held with the team right like that
[75:44] that is public and you know you'll be
[75:47] able to see if we move them or not there
[75:49] is maybe this you know this fraction of
[75:52] shares that are held with um private
[75:54] market investors and while we believe a
[75:57] lot of them are long-term aligned with
[75:58] us maybe there are some that believe
[76:02] that the moment a company company goes
[76:04] public, that's the moment that you take
[76:05] profit and um and they're free to do so
[76:08] on the first day uh or or the first
[76:11] week. But either way, that's that's out
[76:12] of the way if if if they've done that,
[76:14] right? And so that was something that we
[76:17] did very differently, I would say, from
[76:19] other companies, just kind of
[76:20] understanding from our understanding of
[76:21] the public capital markets.
[76:23] >> The last thing I've seen is uh what
[76:25] makes you think you're a good robot
[76:26] investor? What makes you think that
[76:27] you're good at that?
[76:28] >> Can I can I add a little bit something?
[76:29] Yeah. I mean, I I just also
[76:32] want to add that I have no intention of
[76:34] selling my shares. I don't have any
[76:35] plans to.
[76:36] >> Mhm.
[76:36] >> Um, if we don't like a company within
[76:39] the portfolio, I mean, we we we do a lot
[76:42] of diligence. We do a lot of work to
[76:43] make sure that the investments that we
[76:45] make are are really sound, but if the
[76:48] founder changes over, there's some, you
[76:50] know, board dispute or whatever, and
[76:52] there's a leadership change, and we're
[76:54] not on board with that, then we also
[76:56] have the ability to to sell our position
[76:57] as well.
[76:58] >> Mhm. Um, hopefully we don't. Hopefully
[77:00] we just hold these companies forever.
[77:01] And that's what our permanent capital
[77:02] model allows us to do, but we can always
[77:06] just change out exposure from one
[77:08] company to another one that we like
[77:10] better. Like I don't need to sell my
[77:12] shares ever, right? Because of that and
[77:14] because I mean I want to invest in
[77:18] robotics. I I do it in the vehicle.
[77:20] That's it's it's a it's a beautiful
[77:21] structure.
[77:22] >> Mhm. That uh that makes sense to me. Um,
[77:24] one other critique that I've seen is,
[77:26] uh, people say, "Well, what do you know
[77:27] about investing in robotics?" You know,
[77:29] what what makes me think that you're a
[77:30] great investor?
[77:32] >> I mean, I think we're one of the few
[77:34] investors that have a track record of
[77:36] investing in robotics. Um, and this is
[77:39] outside the fund where, you know, we've
[77:41] generated uh mid nine figures of uh
[77:45] unrealized returns. Um, and you know,
[77:48] have made those bets earlier than a lot
[77:50] of people were excited to make those
[77:52] bets. I think now it's becoming a little
[77:53] bit more excit acceptable and there's
[77:55] people getting excited about the
[77:57] industry. Um, but it's it's just pure
[78:00] obsession. You know, I'm here to win and
[78:04] to do it. It just means look, I I don't
[78:06] have a social life. Like I'm reading
[78:09] robotics research papers, listening to
[78:11] like robotics research podcasts. Uh,
[78:13] shout out to Robo Papers. um and uh you
[78:16] know talking with different companies
[78:18] and founders and trying to understand
[78:21] the industry in as holistic of a way as
[78:23] possible and I'm we're bringing on team
[78:27] members that think similarly
[78:29] >> right uh you know they're not here for a
[78:31] paycheck they really care and they
[78:33] understand the industry and we have a
[78:35] really high bar for for talent as well
[78:38] >> one of the things I find interesting is
[78:39] um you know when you buy a public stock
[78:42] you're hiring that management team to
[78:43] take care of your capital. When you buy
[78:45] a closedend publicly traded fund, you're
[78:47] doing the same thing, right? And so in a
[78:48] weird way, uh people who allocate
[78:50] capital, uh they are hiring you and your
[78:53] team to go find the best robotics deals
[78:56] and if they don't like what you're
[78:58] doing, they can always sell as well,
[78:59] right? It's kind of this weird dynamic
[79:00] that you don't get in the private
[79:01] market. I'm an LP in plenty of funds. I
[79:03] allocate my money to them. I kind of get
[79:05] to choose what to do with my money when
[79:07] they give it back to me. And that might
[79:09] be 10, 12 years, right? Um in the public
[79:11] market's a little bit different.
[79:13] >> Mhm. Yeah. I mean that's the beauty of
[79:14] our structure is that it this is liquid
[79:17] and so you know that also might be a
[79:20] reason for you know any premium that we
[79:23] get as well right when assets are more
[79:26] liquid they're more accessible to people
[79:29] have the option to trade in and out of
[79:30] them and there's there's value
[79:32] associated with that
[79:34] >> um and you know I hope people believe in
[79:37] our story and I hope they they want to
[79:39] buy and and and hold and take a very
[79:41] long-term view on what we're doing
[79:43] because that's that's how we're
[79:45] operating.
[79:46] >> Makes sense to me. Where where can we
[79:47] send people to find more about you or or
[79:50] find out more about Robo Strategy?
[79:52] >> We have our X account um at Robo
[79:54] Strategy. Uh my my X account and uh our
[80:00] website and Edgar Filings as well.
[80:02] >> All right. Well, Andrew, thank you so
[80:03] much for taking the time to do this. I
[80:05] think that you guys are really on to
[80:06] something interesting here. It's a
[80:07] massive market opportunity and I think
[80:09] that you have a very unique approach to
[80:11] it. I personally am a big fan of the
[80:12] closed end publicly traded funds and so
[80:14] I'm excited to see what you guys do in
[80:15] the next couple months.