Mat Vogels (00:13)
Welcome everybody. I am joined here by Philip and Sam, and we're here to talk a little bit about the data center revolution, I think is maybe the right way of calling it. I don't go, I feel like a day without hearing about the importance in building out our database or sorry, data center infrastructure. and how that touches onto everything from AI and energy. And we'll probably cover a lot of that in this episode, but let's start with some introductions. Sam, lead us away with who you are and what you're building.
Sam Mendel (00:45)
Yeah, well, I'm Sam, CEO, co-founder of Network Ocean. And we build these underwater and floating barge data centers. And this is primarily so we can take advantage of free seawater cooling. And we can be much more power, water, and cost efficient for specifically these GPU data centers.
Mat Vogels (01:02)
Love it. Philip, over to you.
Philip Johnston (01:05)
Yeah, I'm Philip. I'm the co-founder and CEO of Star Cloud, and we're building data centers in space initially to serve other satellites with compute and then later to address the AI energy challenge by drawing enormous amounts of power from solar in space.
Mat Vogels (01:18)
Fantastic. And you guys were both Y Combinator companies that we just covered that you were at the same batch together. So that's exciting as well.
Philip Johnston (01:25)
We were, we were. There's one thing I remember about Sam. I hope he doesn't mind me saying, I think it was either you or your co-founder who on the first day he wore this like, was it you that wore this like need GPUs t-shirt? I? He like stuck this sign to him that said need GPUs question mark and like was walking around the back. I was like, I was like, this is hustle. This man has hustle.
Sam Mendel (01:26)
Correct.
Yeah, yeah, that was my co-founder. So he went all out.
Mat Vogels (01:44)
I love it. I love it.
Sam Mendel (01:46)
Yeah.
Mat Vogels (01:49)
Ha
Sam Mendel (01:51)
Peace!
Mat Vogels (01:52)
That's fantastic. think that going into an accelerator and going through that program, I the relationships you build out from it is great. And I think it's also great when you have other partners or other founders that are building in a similar space. And so this conversation, I think, is kind of a cool combination of that as well. I'd love to maybe dive a little bit deeper, Sam, into the ocean in this case, on what specifically led you to wanting to build data centers on
in the ocean versus maybe on land. we kind of start from even maybe taking a step back of why data centers in general, what brought you to wanting to start this.
Sam Mendel (02:30)
Yeah, yeah, that's a good question. So I'll give you a little personal background. I've been obsessed with ocean infrastructure specifically since I was in middle school. So this really started on the power side where I was building these underwater generators. These were MHD generators that would operate in ocean currents with no moving parts. So that was the big goal. It was very prototypy, very researchy. But essentially, I was just driven by my definition of impact, which is maybe different than
the colloquial definition, but it's doing something that's not necessarily going to happen unless you're the one that pushes it. And it kept getting driven towards the ocean opportunity, just being absolutely massive. And we were just, at the time, really regulated to low-powered devices, everything out in the ocean. So my real goal was trying to evaluate what scale technology is run better in the oceans and on land. And I was looking at a ton of these different...
possibilities there. Even back in 2015, there was Microsoft that had these experiments with their own underwater data centers. And at the time, I'm like, aha, this is it. This is going to bring the demand. All I just have to focus on is the power out there. But that didn't really pan out. And for a few reasons we can get into later.
you know, essentially it was until, you know, two, two, three years ago when, you know, this demand inside the power density inside of these data centers just absolutely skyrocketed. you know, traditionally refined with air cooled systems. Now you need liquid cooling. All the emphasis is put on this cooling system and current infrastructure is not like nowhere near efficient, from, you know, power consumption, water consumption, and just the, the capex cost of building this thing out. so, you know, around a year and a half ago is when we, when we started building this. and, yeah, I mean, that's where we are.
was the first time where it was economically viable to build these things at scale. And obviously, just like in space, it's a harsh environment. There's lots of things to figure out. But it was definitely the first time where you can start building out this infrastructure that would enable huge capability off of it.
Mat Vogels (04:14)
We had an episode earlier on Pirates Only where we talked about just building in the ocean being really, really difficult. And there was a little bit of a rivalry going on in that building in the ocean is actually harder than building in space and that salt is actually harder to do. We won't try to ruffle any feathers here to make that argument, ⁓ but at least can understand that building in the ocean is very, very difficult and it adds a whole other layer on top of it. the benefits, we'll go into maybe some of those differences.
Sam Mendel (04:31)
You
Philip Johnston (04:32)
Ha ha ha.
Mat Vogels (04:43)
A little bit later, maybe as a quick kind of summary, how big is the team? Have you raised any capital? We mentioned going through Y Combinator. And you said it. How many years have you been building through it? Just some of the basic questions there.
Sam Mendel (04:54)
Yeah, yeah, just around a year and a half, maybe a little bit over that. We're based in San Francisco. We have five people on the team and a bunch of expert contractors that we use part-time here and there. And yeah, we've raised around $2.5 million to date.
Mat Vogels (05:00)
Nice.
That's awesome. That's awesome. Philip, kind of the same question over to you. What led you into data centers in general? And then maybe what caused you to try to go up into orbit instead of trying to build them here on Earth?
Philip Johnston (05:20)
Yeah, for sure. So my background, spent the first five years of my career on the engineering side and then went to move to the commercial side of things. So spent some time with McKinsey working with the space agencies of various governments. I think that's where I noticed the very rapidly declining launch cost was enabling lots of different things to become possible that were not previously possible. One of those that we started looking at, so I went down to Starbase, Texas in
early 2023, just have a look around. I was blown away by the scale of what they're building in terms of both the future reduction in launch cost and the future increase in capacity, which is potentially we might get 100x reduction in launch cost and we might get 10,000x increase in launch capacity as in total mass to orbit per year. And that's because Starships compound on each other. If you build a
one Falcon 9 every day for a year. At the end of the year, you have one Falcon 9. If you build one Starship every day for a year, at the end of the year, you have 365 Starships. And they're building a factory that can produce three Starships a day. It's absolutely mind blowing. But anyway, so we started looking at space-based solar, which is not a new concept. know, that was around in the seventies, people were talking about this. And the whole thing is just a calculation on the launch cost equation. So we ran the numbers and said, okay, what would the launch cost need to be for space-based solar to make sense?
The number we came to is around $50 a kilo. And that assumes that you have a 95 % efficiency loss or 90 to 95 % efficiency loss getting power from space to earth. Because you're using microwave or laser technology to beam the power down. Then we were looking at, well, all new terrestrial energy projects are being built basically for data centers. So let's rerun the calculation, assuming we had a cheap way to get the data center to space. What would the launch costs need to be for us to be competitive with terrestrial energy costs, including the launch cost?
And the number we came to when we rerun that calculation was around four or $500 a kilo. And so that was the basis of a white paper we put together. Um, and my co-founder Ezra and I grew up in the same place in the UK and then we were put, he has a PhD in engineering and spent the last decade building satellites. Um, you know, most recently with Airbus defense and space, uh, working on NASA's lunar pathfinder mission. And then my third co-founder Adi was at SpaceX, um, as a principal software engineer working on Starlink.
and before that 20 years at Microsoft on the data center side. So basically we put together this white paper and that was the genesis of getting going on it.
Mat Vogels (07:49)
I love it. Maybe we kind of double click into both of the technical aspects of it. Could you explain maybe why building in space or in that environment does actually produce a more efficient transfer of power or energy? And then we'll go into some of the reasons why it's so critical for us to try to hit these numbers, but maybe just explain for some of the audience folks to understand the benefits of building in space.
Philip Johnston (08:14)
Yeah, for sure. I mean, maybe I can illustrate it with a full Starship payload bay of compute chips and solar panels and radiators. So one full Starship payload bay is about a hundred tons. This is, and by the way, I should say that we're the initial business we're starting with is serving other satellites that need edge compute, let's say. And for that, we can do it with the Falcon 9 launch cost. But if we're going to compete with terrestrial data centers, we do need Starship launch costs.
So 1,4 Starship payload bay, 100 tons. It's what we expect it to be about a 40 megawatt data center. So including the chips, solar panels, radiators, and satellite structure, we expect to able to fit 40 megawatts in 1,4 Starship payload bay. If you run, the chips are going to last about four years. So if you run that treasurily for four years, then you're, and let's say you're paying 10 cents per kilowatt hour, which is, know, for a new energy project is pretty standard. Your energy cost alone over that four years is $140 million.
versus what we're doing, you can launch it for potentially as low as $10 million, maybe $5 million at the $50 a kilo mark. Your solar panels are about another $5 million. And then all of the other costs are roughly nothing out. So the cost of the coolant loops is similar. The cost of the chips is similar. And so rather than $140 million on energy terrestrially, you've got $10 or $15 million for launch and solar. And the real advantage is you can then scale it almost indefinitely with no permitting cycle. So if you want to build a 200 megawatt data,
Mat Vogels (09:37)
No, are.
Philip Johnston (09:38)
Yeah, let's say you want about 200 megawatt data center trust rate. Like you're looking at a 10 year lead time on that building that energy project in terms of permitting. You can launch five of these 40 megawatt modules very rapidly within a month, locate them physically in the same place or the same orbit and space, and then be up and running within a month with a 200 megawatt data center. And you don't have to stop there. You can scale that to multiple gigawatts.
yeah, so we started about a year and a half ago, January last year, we had 10 employees raised a $21 million seed in September, December, September and October last year. And before that did a $2.5 million pre seed. So, $23.5 million total so far.
Mat Vogels (10:24)
Congrats. That's awesome.
Philip Johnston (10:25)
with NFX
like that.
Mat Vogels (10:29)
So kind of staying on the space really quick and then we'll go on the ocean side. I never thought of the benefit, at least one of the benefits being that the scalability of it and the lack of regulation and those types of things, I think that's pretty incredible. At a high level, as if you're to say, guess right now, are you saying that today's product is essentially being a data center
to the objects that are already in space. Is that kind of the best way to think about it? And then as costs get lower for you to be able to then be a data center for the objects that are terrestrial and on earth. I love that. That's awesome. That's such a cool, it's such a first principle way of thinking about it and building it up too, where you can service. My guess is the demand for space compute is also still high and growing as well. So even that alone being a business is a great place to be.
Philip Johnston (10:50)
Yes. Yes.
Yes. Yes.
It is.
It is it is growing incredibly fast because of the launch costs having come down so rapidly and the number of objects that collect data in space going up so rapidly. So yeah.
Mat Vogels (11:25)
Sam, kind of the same question on your side. Can you talk, I guess that cooling temperature, there's probably a lot of benefits there, but can you give some specific examples of why building these data centers at the bottom of the ocean is the better alternative than on Earth or on the surface?
Sam Mendel (11:42)
Yeah, yeah, absolutely. So I guess to caveat this too, we have two of these units. One are these underwater vessels that sit on the seabed. it's typically, these are near shore. Both of these are typically at industrial waterfronts, like ports, shipyards, places that have accessibility and large power infrastructure already. So this is typically where we operate. The main focus for us right now is deploying our
our actual pilot unit, is a 200 kilowatt data center barge. So it's a floating platform with the data center on top of it we get free liquid cooling underwater. It's much easier on a few of these different sides, namely maintenance and scalability and being able to operate it like a normal data center essentially.
Mat Vogels (12:17)
Lack of salt water.
Sam Mendel (12:18)
Yeah, yeah. But we still get the same advantages of cooling. And that's honestly the most straightforward answer here. It's operating the oceans. You do have a lot of space, but the biggest thing is you have the largest heat sink you can get. So that's where we can operate. Our first unit is this 200 kilowatt unit with some GPUs. And it's pretty clear to see all in the cost benefits compared to these on-land GPU-specific data centers.
and we'll scale from there.
Mat Vogels (12:45)
So explain to me again, what's on the barge versus what's on the bottom of the ocean there? what's the separation, technically?
Sam Mendel (12:51)
Yeah.
Yeah, so for the underwater vessels there, they're like pressure vessels. They're essentially like these submarine capsules that sit there. They're housed with a bunch of compute cooling systems. A lot of it is passively cooled through the shell of the actual vessel and some heat exchangers on the outside of the vessel. But you can supplement it if need be, which usually you don't need to, but with an active cooling system where you pump seawater. It's nowhere touching the chips. There's separate cooling loops. You have a closed loop cooling system for your hardware. And then you essentially reject the heat into the
the seawater. And it's a similar system for the uploading except, mean, obviously it's at the surface, it's usually sitting at a berth. You can plug in with power with internet, operate like a normal data center, and we get free cooling in a similar manner. So to shed a little background on if some of the viewers are not super aware of what normal data centers look like on land for these new AI data centers.
Around a third of the entire build out of the data center in terms of space and in cost of this, the facilities is cooling. So there's these massive cooling towers, evaporative chillers, large pumping systems that you need to actually reject the heat. there's systems that, you
Mat Vogels (13:50)
Thank
Sam Mendel (13:56)
take the heat away from the chip itself and then move that somewhere else. And then you need to reject that heat out of the facility. So instead of those massive cooling towers, evaporative chillers that are very water inefficient, some of these other systems that are very power inefficient, they're all very expensive as well. So for us, we can eliminate a lot of that and make it just a much simpler cooling system by building some of these seawater components.
Mat Vogels (14:19)
Do you have an idea of like from the cost in energy like consumption that those like on land like is it equal cost saving I guess in a way or can you talk a little bit about that?
Sam Mendel (14:31)
versus floating in underwater or versus...
Mat Vogels (14:32)
Yes, exactly.
you mentioned, mean, the size disparity or least being able to split out the size and have the cooling piece at the bottom of the ocean versus on land. But my guess is that the energy cost for the cooling systems of those massive or the energy for the massive cooling systems has got to be huge. And do you have an idea, I guess, what the cost savings or differential is specifically on one of your data centers versus an equal size one on land?
Sam Mendel (14:53)
Yeah.
Yeah, yeah, good question. So I guess the main metric in the industry is PUE. this is like basically how power efficient is your facility. mainly this hinges on the cooling system by and large. industry average for a while has been 1.5, 1.4, 1.5, something like that. Some of them are cutting that down a little bit. essentially this means that if you're powering all of your servers with one unit of compute, you need half of that more to run all like essentially
essentially the cooling system and the rest of the components in the facility. And so for our underwater unit in early tests, we've seen a PUE of 1.07, which drastically cuts down this entire power usage of the data center. And a lot of those savings are just, for us, passed on to the customer. And we just want to scale this out.
Mat Vogels (15:41)
That's incredible. I love it. Philip, you mentioned, I mean, one of the big benefits being the ability to easily scale. Are there any other additional benefits that you want to touch on from the space side?
Philip Johnston (15:50)
Yeah, I mean, the longer term, one of the, so in the short term, cooling in space is actually quite a challenge because it's a vacuum. And so you need to build very large deployable radiators, which are going to emit this heat and infrared out into the vacuum of space. In the longer term, if you, if we double the, or let's say we 10 X the total amount of power that we're using currently either on data centers or in general, there comes a point somewhere between about 10 X and a hundred X where the amount of
that we're kicking off from these data centers and just in general from power usage is enough to warm the planet itself. So if you're using fusion reactors, for example, or if you're using any other power source, you're just, you know, we hope fusion reactors work, but you're literally just pumping heat out to the atmosphere. So it's not necessarily from fossil fuels. It's just literally from the waste heat. And that there comes a point where that is actually prohibitive to scale any further.
And it's within our lifetimes. So between 10x and 100x is probably within our lifetimes. And at that point, the only place you can build more data centers is in space, because that's the only place you can push more heat out into the vacuum of space. So in the longer term, that is one advantage of space. So yeah, we've got a little way till we get there, but it's coming quicker than we think, I think.
Mat Vogels (17:07)
Yeah, we'll definitely we'll dive in a little bit on some of the demand and I'm sure that you guys are seeing even from your own pitch decks and thinking about it. But we'll dive into that. One of the other questions that people were interested in related to each of your businesses is diving in specifically to some of the difficulties of what it means. Maybe sticking with you, Philip, on building in space, you the additional capex of actually like getting into orbit and doing all those things. What are some of the other difficulties and thinking about building a data center?
in space and the technicalities that go along with that, that maybe a typical startup that's not building in space, but certainly maybe not a typical data center.
Philip Johnston (17:45)
Yeah, I mean, I'd say half the team is working on this problem I just described, which is the cooling problem. So half the team is building very large, low cost and low mass deployable radiators. The other half of the team is working on making chips work in a high radiation environment or working to mitigate that through shielding or through algorithms. So yeah, those are the two big engineering challenges for us is thermal management and radiation shielding.
Mat Vogels (18:11)
It's funny, it's like radiation in space and then it's salt water in the ocean. ⁓ Sam, kind of the same question over to you. What are some of the difficulties of building in the ocean that maybe some other founders may not realize?
Philip Johnston (18:17)
Hahaha.
Sam Mendel (18:26)
Yeah, pretty similar. There's two. There's corrosion from the seawater, and then there's biofouling. So things in the ocean just like to grow on things. It makes it a lot less efficient. They eat into a lot of what you build out there. So those are two main challenges. There's a few others. But I think by and large, they're relatively solved problems, just in the marine industry. There's a cost in...
maintenance cycle that you have to follow and to implement and build these things.
Mat Vogels (18:57)
Is the regulation an issue? That was something that was brought up as well was the regulations in space sound like those are a little bit lighter, but regulation in the ocean tends to get difficult. Certainly as it relates to, you know, mining, energy production, those types of things, obviously different industry and what you're doing, but is regulation going to be a friction point or something that could prevent or hurt you as you keep going?
Sam Mendel (19:18)
Yeah, that's a really good question. And I think one thing is it's very different from location to location, but it's also like it lines up pretty nicely with the locations we look at, especially this industrial waterfront. So a lot of these, you know, ports, shipyards, places like that, because there are typically very dirty things around there and there's plenty of unused space. So as a comparison, you know, there's, you know, ferry ships, cargo ships, you know, cruise ships that are, you know, all they have engines that are actually, you know, super high power.
and liquid cooled in a similar manner, multiples higher than the heat that we would dissipate with our facilities. So it's quite easier to make that comparison early and there's definitely some differences and it's slightly quicker than actually deploying a full data center of similar size on land. So it's pretty interesting from that front.
Mat Vogels (20:05)
Yeah, and my guess is that it's not like you're putting these on like pristine coral reefs. there's not like you're trying to or even put these in the middle of the ocean where there's no infrastructure already in place. If you can put those to your point in some of these areas that are already inhabited by machinery and that there's already disturbance, I guess, going on from the ecosystem level by you adding on to that. It's not that you're doing additional harm, so to speak, or any harm, but
Sam Mendel (20:10)
Yeah.
Mat Vogels (20:32)
because it's already so busy in those areas. It's not that you're at any risk in doing anything there.
Sam Mendel (20:38)
Correct, correct. yeah, mean, one thing is like there's also like a temperature gradient that you look at from, you know, the temperature you get the water in at the coldest point and then the hottest that you outtake. And for us, it's, you know, within a few degrees Fahrenheit. And, you know, obviously in water it dissipates really fast, very high specific heat capacity as well. So, I mean, you know, obviously there's tons of testing and surveys that we're doing, but yeah, everything looks pretty good for now.
Mat Vogels (20:43)
Yeah.
from like a just like a maybe a little bit more boring kind of level, but how do you have to go down and like scrape and clean them off? Or like, what does the process like actually look like in keeping some of these clean and maintaining them? And I don't know what a high level maybe give it like a range of time of how often you'd have to go and maintain them and how long they would last down there.
Sam Mendel (21:11)
Thank
Yeah, good question. Good question. So underwater is a more interesting one. I'll start with the floating version. The floating barge is fairly similar. These things operate for decades and they're very commonplace. And essentially for our pilot for sense of scale, it's roughly two of these ship containers worth of equipment on our end that's sitting berths on this barge at Porter Industrial Waterfront. The underwater, so that's like much, much easier to do maintenance. And the big thing to consider is that, especially working with these GPUs, there is a
a
very high failure rate of the chips themselves, So networking equipment chips themselves, and you need someone there to actually make these switches to get these things back online. So that's a big reason why our main focus for the pilot is these floating units. It's much harder to do that underwater. So you have to just bake in this whole kind of maintenance operation, which has to happen fairly frequently. So the operation with that is you're looking at how many chips go down. At a certain point, it makes sense to let's do the switch.
So you float it back up to the surface. And typically we're in shallow waters, like also at industrial waterfront. And then we take it on land. We do the switch out, the fixes, and then we put it back underwater. So it's an interesting thing that we're looking into. It's kind of similar where, you know.
The large scale demand is definitely something we can attack with those floating data centers. And there's some of these niche markets where underwater definitely makes a lot more sense. And in the longer term, I don't have high conviction that this will happen in the next few generations of chips, but we'll continuously monitor if the...
failure rates of these chips start getting much better. There's also some very interesting research going on with these underwater data centers and Microsoft's early experiments back in 2015, 2016, they had one eighth of the failure rate of their servers in those underwater capsules. So that is theorized to be because of it's a very sealed, clean environment. also actually use nitrogen and some other
kinds of gases for a lights out data center instead of needing normal corrosive oxygen. So there's some interesting experiments we're doing on that front. But by and large, our full focus is for the scaled floating facilities.
Mat Vogels (23:32)
that's that's incredible. Philip, you mentioned on the radiation side, kind of a similar question, maintaining those and continuing to do that. Is it in space? Is it typically that you're launching these objects up? Is there a maintenance process or are they just coming down and you're just like relaunching up? Or what does the process look like of maintaining your product in space for you guys?
Philip Johnston (23:53)
Yeah, that's a great question. So initially, we will not have access to them. And so it's a bit like any satellite, like the Starlink satellites or any others, where the critical systems will have to have redundancy. And then we will have over-provision on things which will degrade over time, like the solar panels. Over time, the industry is moving towards robotic maintenance. And so over time, I'm expecting within five years, pretty
significant and substantial robotic maintenance operations in space. In fact, one sort of interesting sci-fi concept is I expect within five years, probably we're going to see terrestrial data centers being maintained with humanoid robots. And humanoid robots actually surprisingly work pretty well in space. If you put them in a space suit, that takes care of radiation and thermal. So it sounds a little bit sci-fi, but we could see humanoid robots maintaining space data centers at some point.
Mat Vogels (24:25)
Yeah.
I mean, it I don't think it sounds too sci-fi at all. I think it's it's a it's a no brainer. And one of the questions I was going to ask was what other companies? Yeah, exactly. What other companies do you know of that are, I guess, operating or helping companies like yours in space right now from like a support level? Because I feel like that's a whole other industry you're providing right now, the data center and those pieces to the space economy. But there's so many other pieces in the space economy that are going to start emerging.
Philip Johnston (24:52)
We are not totally. I'm glad.
Mat Vogels (25:17)
And think that's just one of them. don't know if you have any others top of mind that you're thinking through that could be exciting as well.
Philip Johnston (25:24)
Yeah, I mean, there's tons of robotic maintenance and maintenance companies coming up. Starfish Space a month ago just launched their second spacecraft. I think they'll do the first commercial space rendezvous and docking. Then in our batch, there was one called Spacium, and they're doing space refueling. Yeah, I think there's going be a whole bunch of these.
Mat Vogels (25:28)
Yeah. That's right.
Yeah, the space economy is, the ocean economy and space economy, both I think are going to be incredible to watch over the next decade or so, which kind of brings into the next phase of conversation here. It's kind of a, it's a fun one. Like I mentioned earlier, I feel like I don't go a single day where data center isn't a topic of conversation. I hear it from folks on the Capitol Hill side and the political side as it relates to how do we start thinking about
data centers. How do we think about regulation and the size of the energy consumption? And it's always in this weird conversation of we need way more of them, but we also don't have the energy to supply way more of them. And there's this balance and chicken and egg thing. Then meanwhile, we have an incredible surge in demand over the last couple of years, certainly. But if you look over the last even five, 10 years, it's almost exponential and it's continuing to go that way. Maybe Philip's saying with you, could you talk a little bit about
What you're seeing from a demand side, mean, probably a lot of the same conversations that you would have with your investors on why what you're building is so critical and what the demand is gonna look like projecting forward over the next 10 years or so.
Philip Johnston (26:54)
Yeah, for sure. And I'll just quickly touch on it because you mentioned the Capitol Hill conversations. So we had a representative from Senator Round's office reach out saying he's very interested. And so we went to meet him in DC, my co-founder, Ezra and I. And he views it as a national security priority that we find massive ways to generate massive amounts of new energy to power AI data centers.
purely because places in China, for example, they're building, deploying enormous amounts of energy to power mainly to power data centers. And if the future of national security is on who has the best AI, then that is certainly a cause for the US government to be concerned about. yeah, but in terms of the other demand, the first
bucket of demand is coming from other satellite users. And to be frank, in the initial stages, that's the conversations we're mainly having with DOD customers. And they have lots of programs to fund R &D upfront. So we're partnering with them and we'll be running some demonstration customer workloads with them on the first satellite that's going up next month. In the later stages, we'll be partnering with large data center builders terrestrially.
And we're hopefully going to sign an MOU with one of the largest data center builders, I won't say who it is, for energy offtake. So yeah, that's where things are going in the end state.
Mat Vogels (28:16)
Could you explain, maybe dive a little deeper on the conversations that you were having in Capitol Hill, you mentioned national security issues, which I think a lot of folks would agree with. What particularly is it simply that we're trying to geopolitically compete with China and what they're doing and if AI is considered there, but is it also, a lot of the conversations I have as it relates to that is just the infrastructure required in order to build that, maintain that is also kind of at a national security risk. If we need to,
continue to progress on the AI side, we're gonna need more data centers and we want more data centers, we're gonna need more energy production. Was there any more like conversation on how they were explaining as they're thinking through the industries around data centers that also I think need to be updated dramatically over the next decade?
Philip Johnston (29:03)
Yeah, 100%. I mean, you hit the nail on the head with the need to build huge new energy projects. And the thinking mostly has been we need to speed up regulation. We need to speed up permitting cycles for things like nuclear. But as people are starting to look for more, you know, for new and innovative ways, they're starting to think, OK, maybe there's there's something to this building, huge data centers in space. And I think Eric Schmidt recently was giving a testimony on Capitol Hill.
And he said, you know, the demand for energy is going to go up by something crazy number. And later he was asked about that and he recently bought Relativity Space, a launch vehicle company. And he said that the reason he built, the reason he bought Relativity Space was to build large data centers in space, which is, thought, I thought was very telling because he's one of the people that knows the most about it, given his time at Google.
Mat Vogels (29:53)
Yeah, it's, I think the whole space is going to be in for a wild ride. We'll share some charts and numbers, but when you look at the growth of data center demand, it's almost like this weird inflection point that obviously hit around, you know, right when AI and ChatGDP and started growing, but then it gets like more more steep as we continue to grow through. It's like, it's not a, once we get to like 2029 and then 2030,
It just continues to go up higher and higher.
Philip Johnston (30:26)
Exactly.
Yeah, I sometimes get asked, why can't we just do this terrestrially? And my answer is basically, if I thought that we were going to double data center energy consumption over the next five years, and then it was going to stop growing, I would say, okay, it's an enormous effort to build data center in space. Let's just figure out terrestrially. That's not what's going to happen. It's going to
double over the next five years and then it's going to quadruple over the five years after that and then it's going to eight times over five years after that and it's just going to keep going and going and going and going. There's basically no end in sight.
Mat Vogels (30:54)
Yep. Yep.
Yeah, Sam, kind of the same question on your side. Are you seeing and feeling some of that demand as you were going through the fundraising process and talking with customers now? What does that kind of pull, whether it's geopolitically, politically, revenue side, whatever it might be that you're seeing from that?
Sam Mendel (31:14)
Yeah, yeah, absolutely. I mean, of course, there's unprecedented demand. For us, with our pilot unit, it's mainly for these NeoCloud-type companies. So GPU-specific clouds that liquid-cooled data center space to cool these next-gen chips are super, super in constraint right now. it's pretty much like, not only can we build this and offer this for you, but...
Philip Johnston (31:29)
Take care.
Sam Mendel (31:37)
It's much more cost efficient, power efficient, water efficient on all those fronts. They can expect to save somewhere between 20 and 30 % using our kind of barges versus these online data centers. So quite significant demand. And I think just to put the actual scale in perspective, I grew up in a local IT family business. So I was setting up some of these servers into small and local businesses. That server was maybe
takes like 100 watts or so of power, maybe the whole rack, if it's filled is like 5 kilowatts or something like that. I mean, now, like with Nvidia's NBL72s, we're looking at like 120, 130 kilowatt racks now. And this is like only projected to increase in density like crazy, right? So, know, 600 kilowatts soon, megawatt racks soon, and it'll just keep increasing. So, I mean...
And on the usage side, I mean, there's these massive training clusters that are set up and we're really just starting to see like this inference demand start to scale, right? So, I mean, there's no question in mind we're going to have over a billion people using these, billions of people using these AI tools and billions of agents going around as well. So the inference demand is just going to be insane and we have to build that. And for us, it's obviously like we have to get there as soon as possible so we can build this infrastructure much more efficiently than
what's
currently out there. So there's a race internally in the team to get this going. Every time we see a new headline for a massive data center being built, we're like, we need to be there. Every second we're waiting here, there's another data center that's going up that could have been built much more efficiently.
Mat Vogels (33:08)
Could either of you maybe shed a little bit of light on, you know, for the folks that don't understand necessarily why or how the demand is growing so quickly. Obviously, AI being a big piece to that. But Philip, you kind of mentioned it, which is true. It's like we need to double by 2030 quadruple by 2035 and then eight times and then 10 times. And then it's going to grow and grow. Why is it that we're seeing like that type of growth? I don't know if you could add or shed some light on
on why AI is like so power consumptive and what we're going to be doing. And we'll project this into maybe what the world looks like in when those things become a reality. But do you have any any kind of clarity you can add to folks that maybe don't understand why the demand is just getting so high? Like what is what are we getting out of out of that?
Philip Johnston (33:54)
Yeah,
yeah, yeah. mean, a very crude and simple way of looking at it is that people are writing more charge of T queries and each charge of T queries and enormous amount of energy. A more nuanced and accurate description is that the world is going to be the world is and will be populated with agents, AI agents doing lots of things. AI agents are basically almost continuously running.
inference queries on these models. And they just go back to back to back. They're one run query that leads to another query leads to another query and it leads to another one. And they can essentially replace human workers. So then the question is, many workers do you want? And the answer lots of companies are coming to is an absolute shitload of workers of these AI agents. And when you say what's the most number of like
digital workers you could use. Oftentimes the answer is there is no upper limit in the number of sort of very low cost digital workers I could use. And so that is essentially, I think the key driver of it because each one uses just an enormous amount of power.
Mat Vogels (34:56)
Yeah, I think when I when I've asked people that before and I get the answer of like, well, it's just more and more people that are going to start using AI and they look at it as almost like a Google search where it's like, well, more people are going to start asking, you know, chat, GDP questions. But you're exactly right. I think the AI agent side is is going to continue to compound and people don't realize that you ask an agent to do something with the agents are doing now is like replicating themselves. So it's like I'm going to replicate myself as an agent.
Philip Johnston (35:09)
Hmm.
Mat Vogels (35:21)
four times and that agent's going to be able to replicate themselves to handle specific tasks. And it breaks down into this massive for maybe one query. Like you may ask it one thing and it ends up breaking off into this, this chain effect of all these different AI agents running. And maybe another kind of question, have we in your guys' opinion, are we, are we limited today? Like is there, is there, where's the, the, you know, chicken and a cat and mouse type of effect of
progress in AI as it's related to the energy consumption The chips that we're trying to put in these data centers like are there what are the blockers today? It feels like they're all kind of leveled up But they're all you know leveling up kind of at the same time But is that which one is going to be the blocker like all of a sudden are we gonna hit a GI and then now all of a sudden? A GI is limited because of the fact that we don't have the energy to run a GI for everybody yet like where where does the
Do the limits kind of come in?
Philip Johnston (36:17)
I can talk my own book if you want. mean, without too obviously talking my own book, I do think the fundamental limit is in energy. I mean, that to me does sound like the fundamental limit. But I mean, that is also the problem we're solving.
Mat Vogels (36:19)
Please, no, Philip, go ahead.
Sam Mendel (36:20)
Go for it.
Mat Vogels (36:38)
Exactly right. Yep.
So I think if you can, yeah, go ahead, go ahead, Sam.
Sam Mendel (36:40)
Yeah, yeah.
Yeah, I think off of there, a lot of the capabilities that the current AI models have shown, you know, past these thresholds of economic viability. We've already seen this in multiple industries. I think that the next step are, you know, actual breakthroughs in scientific discovery. think that's, you know, super, super exciting and game changing in terms of, you know, human understanding, human capability. But, you know, in terms of like more short term build outs, it seems like most of these main hurdles have been crossed. But a main limiter definitely seems to be, you know, power and
financing, right? So like these data center facilities are incredibly expensive and we've already been tapping, know, like at least started to tap some of these, you know, large pools of money. But, you know, I think that'll continue to grow for sure. Cooling is definitely a thing, you you can't use air-cooled data centers. These are getting incredibly more dense. You need to look at new cooling technologies. And think there's like some very cool technologies coming kind of across the stack to make that work.
Mat Vogels (37:12)
Yes.
Sam Mendel (37:34)
down to the chip level, all way to the facility level. And then on the power side with nuclear power coming online. And it'll be exciting to see what happens.
Mat Vogels (37:45)
I think it'll be fun as those things unlock, hopefully it unlocks companies like yourselves where you end up feeling like you have unlimited power to continue to scale and grow through. And I think that we'll meet the demand. I get the question asked a lot of whether in five years are we gonna be satisfied with where we're at and then it levels out. Like when does this curve slow down? And I don't think it will, at least maybe in our lifetimes, just because we'll find ways to meet that maximum and continue to produce more, want more.
And even then like the AI agents and robots and that's another world that I'm going into on the on the data side is the physical world is so much more complicated than than a digital one in some ways. And then when you add that into the mix, I think it just it compounds on top of that. The other kind of questions that we get asked a lot and we like to ask a lot of the founders is if your company is successful, everything that you want to do over the next 10 years, you hit the milestones with ease and maybe even surprise yourself.
What is, what is the, what does star cloud look like in 10 years? If everything that you want to have happen happens maybe as a company, but then we'll talk about maybe the, benefits of the world. Yeah. Let's, do that. mean, wherever you feel like 10 years or maybe 20 years, whichever. Oh yeah. Let's do, let's do 20. Yeah. Yeah.
Philip Johnston (38:53)
10 years.
20 years.
So in 10 years, think almost all new data centers will be being built in space. The one exception to that will be where you need extremely low latency and many cities are built next to the sea. I can imagine, Sam, that there's going to be people wanting to do what you're doing for the latency. But for things that just want enormous amounts of raw power, think in 10 years, almost all new data centers will be built in space.
Mat Vogels (39:18)
There we go.
Philip Johnston (39:27)
That means that there will be a replacement rate over the next 10 years after that. And by 20 years, I would expect the half of all data centers to be in space or like at least. and then because data centers have, you know, like a 20 year lifetime, it's going to take time to replace them. Um, and then beyond that, we're, we're building Dyson spheres and Matrioshka brains.
Matrioshka brain is a Dyson sphere hooked up to as much compute as a star system can physically support. Now that's going to take probably about 100,000 years to build a Matrioshka brain.
Mat Vogels (39:58)
Wow.
100,000 years, is that what you said?
Philip Johnston (40:05)
Yeah, so you surround a star with a swarm of solar panels, ⁓ yeah, that's where we're heading. That's the future of StarCloud.
Mat Vogels (40:08)
⁓ yeah, yeah, exactly. That was gonna say that you put that
Sam Mendel (40:11)
You
Awesome.
Mat Vogels (40:15)
in the last
page of your pitch deck. That's perfect. If you tell, yeah, set Elon on it, maybe it'll be a thousand years instead of a hundred thousand. to wait and see. But do you have an idea, what's the economy of that? Like, what does the economy look like of data centers in space then in 20 years? I mean, could you even project it? Like, does the line get, like, it gets too steep almost, I feel like.
Philip Johnston (40:18)
Yeah.
Maybe.
I mean, I think almost in the end state, pretty much the entire physical economy will be data centers in space. That's that's more than 20 years. That's. Yeah, yeah, that's more than 20 years. That's probably more like hundreds of years, but yeah, I mean, if you look, if you take the current total data center revenue and at 10 % per year out for 20 years, I think that's roughly the.
Mat Vogels (40:44)
Yeah. And the robots maintaining them and yeah, yeah.
Philip Johnston (41:03)
Revenue potential of data centers in space. So talking 10, 10, tens of trillions of dollars.
Mat Vogels (41:06)
Yeah, that's good.
Yep. And it's going to continue to be, I think, critical national security for forever, for as long as we can, we can project it. The other question that I get in space, that's what I was going to say it is, it could become a whole planetary type of thing as well. Yeah, especially as we think about climate and all the issues that we kind of have as a whole. But the other question that I get, maybe this is one of those dumb questions that I get asked a lot and I don't have a good answer for, but maybe you've thought a lot more about it.
Philip Johnston (41:20)
National and planetary security.
Mat Vogels (41:41)
is what are we running out of space and space like eventually do we run out of space and space every now and then I see a picture and I'm like that's a lot of satellites up there are they just not like running into each other like constantly and that is you project forward and have all these data centers out there like do we run out of space in like certain orbits like what does that look is that a limit
Philip Johnston (41:52)
Ha ha ha ha.
Yeah, that's it.
So these maps are incredibly misleading. And the reason is that each of those blobs is usually the width of California. Like they're literally the width of California. And you're talking about a satellite that's like this wide that's represented as the width of California. space is actually, there's a lot of space in space. I mean, if we actually start to run out of space, then we can start putting stuff in a sun orbit or in the Lagrange points or in a lunar orbit. So there's basically limitless space in space.
Mat Vogels (42:06)
I know, yeah.
Philip Johnston (42:31)
much more space in space than there is space on Earth.
Mat Vogels (42:34)
Absolutely,
no, absolutely and it's one of those ones I try to like look it up and you're exactly right the visuals Don't do a very good job of showing it because yeah, I see the here all the satellites the SpaceX satellites and I'm like They're covering the whole earth. There's no space anywhere. Like what are we doing? And then? ⁓ Yeah, when you do a little bit of napkin math, you realize that there's there's a lot of space up there
Philip Johnston (42:48)
Ha ha ha ha.
Sam Mendel (42:49)
Thanks.
Philip Johnston (42:54)
Yeah,
Mat Vogels (42:55)
Sam,
kind of thinking on your side, same question. You're projecting 20 years out. What does it look like for Network Ocean? But what does it look like, I think, as a whole back here on Earth, whether it's in the oceans or even on the terrestrial side? But yeah, specifically to your company as well.
Sam Mendel (43:10)
Yeah, mean, even 10 years, I think it gets really, really exciting. we're going to have gigawatts of data center deployed, or underwater data centers deployed. Hopefully, we get to the point where it starts to make a real dent in terms of the average efficiency of the total data center infrastructure globally.
But what's really exciting after this is we're going to be starting to look at pushing out further from this kind coastal infrastructure. A little bit further we can look at probably start deploying or start at least looking at some of these nuclear-based options for power. And enabling us to go further out is actually huge because not only...
Mat Vogels (43:44)
further off into the middle of the
ocean. ⁓
Sam Mendel (43:47)
Exactly. And
you have obviously a lot more space. You still get a ton of cooling regulatory side. It's easier on some of those fronts.
But really, the main goal that I was talking about at the beginning of our discussion is that I want to push forward the capabilities of humanity in the oceans. And that really starts with something where you can access grid-level power out in the oceans. And there needs to be something that drives that demand. In other major infrastructure build-outs in humanity's time scale, there's always been some kind of demand driver. And for the oceans, could be data centers. So it'll be really interesting where you start to see other infrastructure being built on top
of our power and data center infrastructure out in the ocean. So I mean, I'm super excited to see where that goes. It'll start with transforming some current ocean industries. There's plenty of new unlocks that I'm very excited about and can talk about forever. But yeah, think essentially if you have good level power out there, capabilities drastically increase all the way to the point where you can have people living out there and really unlock the 70 % of their planet that we're basically just tapping.
Mat Vogels (44:36)
So, thank you.
Sam Mendel (44:51)
and topping the service of. So that's what I'm excited about.
Mat Vogels (44:54)
To talk to give some examples of that so it's almost like you're exactly right if we start needing to build infrastructure out into the oceans There's just other economies that will be naturally built out from those You said you have a bunch of those order some of the the ones that you you get excited about the most
Sam Mendel (45:11)
Yeah, I mean with traditional industries, mean, you can see electrification of shipping, can see large scale aquaculture, fishing, deep sea mining, basically just like large data collection and just huge macro engineering projects, what we're talking about. And you really have everything you need to have like a human colonies on the ocean. So I think there's some really interesting developments kind of across the board, some of these other companies, but
Yeah, I'll stop there, but
Mat Vogels (45:36)
We this is the second time that water rolled the movie has been brought up on on this podcast but that's the first thing I kind of think of is if we could eventually live out on the on the ocean will have these like countries that are that are floating out there so to speak and And maybe it starts with the data centers data centers are the pathway to living in the oceans
Sam Mendel (45:40)
Yeah.
Yeah, mean, there's a, you know, it might be a very, very clear path here, you know, kind of this kind of this golden path, right, so that we can have this first opportunity where we can build something at scale in the ocean that operates much better. And, you know, obviously, like they have, you know, stable, very, very high returns and massively skilled, you know, more than any other other industry on Earth. you know, using that, you can definitely build out the backbone of, you know, large scale power out in the ocean. And yeah, I mean, that's just like the most
Mat Vogels (46:15)
Yeah.
Sam Mendel (46:23)
where humanity finally actually unlocks the ocean so I'm very excited for that.
Mat Vogels (46:26)
Awesome. Last question before you kind of go into some CTAs and what you guys are working on or need help with, but what other companies that are either in your space or not, it could be completely unrelated industries or sectors, are you most excited about right now? Not including yours or each other's companies. And Philip, maybe we start with you.
Philip Johnston (46:46)
I'm super excited about human order. I invested in a company called Proception in the last batch. They've just got sued by Tesla because their engineer used to work there. Yeah, yeah, I think I recognized it. OK, amazing. He's absolutely awesome. That team is absolutely amazing. I went in to see them recently. So yeah, that's the thing I'm by far the most bullish about.
Mat Vogels (46:49)
Wow, yeah.
They were a Black Flag company as well. And we're having him on the pod here soon. I think you did.
We had a Ben over from K-Scale on a couple of weeks ago as well and talking about what they've done with K-Bot. I've said that the humanoid robots is probably the biggest thing that I've had my 180 thought on, where two years ago, if you would have said, you know, when are people gonna have robots in their homes or have robots doing mundane tasks, like, you coming in to fix like a drywall in your house or fix a leaky pipe or something like that. And I would have been like, like decades, if ever.
People don't want that, like we don't want that at all. And now I'm like, it could be next year. Like it's gonna be so fast that ⁓ we're gonna, the same way that you have a car, you're gonna have a humanoid robot in your home. And obviously there's incredible layers to it in the commercial space, whether it's in space or in factories and those, but I agree. think that humanoid robots is gonna be, that's why people are saying it could be one of the largest industries.
Philip Johnston (47:40)
Yeah.
Yeah.
Mat Vogels (48:06)
ever and not even be close and I could could believe that. ⁓ So yeah, I think that's a that's a really good one. So you mentioned any other specific companies that that you think about in that space that you're excited about besides perception.
Philip Johnston (48:08)
Yeah. Yeah.
That's the one I'm most excited about. Yeah, not really a comment.
Mat Vogels (48:23)
Yeah, perfect. Sam, what about you, Eddie? What are the industries, maybe even outside of the data center side that you're really excited about right now?
Sam Mendel (48:31)
Yeah, actually a few guests that you had on your ocean episode. I think what Will is doing with Ulysses, what Poseidon is doing, in terms of unlocking, it's just a very symbiotic relationship in terms of being able to unlock capabilities in the ocean. think that's awesome. And there's a real push that's starting to form. So that's really cool. Actually, on the data center side, think there's tons of cool cooling technologies across the stack that are happening.
I guess one that I'll highlight is Akash system. So they're essentially making these like synthetic diamond cold plates. They're like essentially the layer that sits right on top of the chip, like for direct to chip cooling. And it's this kind of like interface layer. This is heat spreader before you get to the cold plate itself or the heat sink. And that's just like one single bottleneck in terms of like increasing heat capacity. Cause I have a belief, think there's a few.
pieces of example of this, but I think Jensen really wants to do more. He wants to increase the heat of these chips. wants all these like cooling and data center people to start upping their game and he's pushing that on a few different fronts, but we really have to...
Mat Vogels (49:38)
Mm-hmm.
Sam Mendel (49:39)
start improving this and it's a little like call to action for the industry to start working on novel new technologies across the stack. what they're doing kind of solves that one bottleneck and then a new one will appear and then we'll have someone attack that. yeah, mean, I think that's, it's really exciting to see.
Mat Vogels (49:55)
days.
The chip race is going to be a whole other category that's going to be fun to watch over the next decade as well. It's why it kind of feels like it's this really cool race across the board and energy efficiency, data center efficiency, chip efficiencies, all while AI is like compounding. So it's like all these things that are trying to bring the cost and infrastructure pieces down exponentially as fast as we can, while AI on the other side is like really trying to outdo those efficiencies as best as we can.
My hope is that we just end up hopefully on the net positive side, but I think even if we're gonna end up even where we'll just continually like AI is gonna grow and then we'll try to match that demand with efficiencies. We'll see how that projects over the next decade or so, but I agree on the ocean economy stuff. There's some fun stuff happening in the middle of the ocean and we get a lot of startups that submit. We've had startups that have submitted like the cruise ship.
type of thing where it's like we're gonna build cities or ecosystems. We had an accelerator proposed that they should, they're gonna do a cruise ship. And so they're looking for funding to like start a cruise ship and then put a bunch of founders on it, sail around the world and do like an accelerator on the ocean, which could be kind of fun one too. yeah, there's a lot of stuff. YC should do one on the ocean. Actually Black Flag should because it's a little more nautical themed. Awesome.
Sam Mendel (51:10)
Yeah.
Philip Johnston (51:12)
Ha ha ha
Sam Mendel (51:12)
Let me know how
Philip Johnston (51:13)
ha.
Sam Mendel (51:13)
we can support. Maybe we'll do a compute for you guys.
Mat Vogels (51:15)
Yeah,
I mean, honestly, as soon as the unlocks start happening, it'll be a free for all. So that's great. Anything else you guys want to cover any topics or things that on the data center side that we should bring into light?
Philip Johnston (51:26)
I just wanted to say, it's awesome to chat with Sam again, because basically every single investor call I have, I get asked, why don't you build data senders under the sea? And so have to, have like an answer where I'm like, actually when Microsoft did it, the reason they shut it down, it wasn't necessarily that they didn't have good results from it. Like, cause my co-founder used to work at Microsoft. He said they actually produced some pretty good results from it. so it's a space that I think is super interesting and super cool. yeah.
Mat Vogels (51:37)
Uh-uh.
Philip Johnston (51:52)
It's cool to share with Sam again.
Sam Mendel (51:55)
Yeah, yeah, it's good to catch up. It's been a while since the bash. And I think likewise, I also get...
Mat Vogels (51:58)
And you guys are both in the Bay Area, right?
Philip Johnston (51:59)
Yeah, yeah.
We're in Seattle in Redmond.
Mat Vogels (52:02)
God, OK. Yeah, so yeah, so yeah, nice. So you're not going to run into each other in any sort of circumstance there. Are there like data center conventions that you can meet each other at and go to?
Sam Mendel (52:03)
I see. We're in SF.
I'm sure, I'm sure. Absolutely.
Philip Johnston (52:13)
Probably.
Mat Vogels (52:14)
Yeah, yeah, yeah.
Well, great. Thank you again, guys, for jumping on before we hop off. I want to give each of you a little bit of a platform to share any needs that you have right now, hiring, announcements, anything like that. Phillip, starting with you.
Philip Johnston (52:30)
Yeah, so we've got first launch coming up in a few months. We'll likely go out for a raise after that and towards the end of the year. And we'll look to bring on some more engineers. not hiring right now. We'll keep the team as it is. But probably towards the end of the year, we'll be looking to bring on some more amazing engineers and sign some more customer contracts.
Mat Vogels (52:46)
Bye bye.
Amazing. Yeah, we'll keep an eye out for the launch. What's the month or do you have an exact date?
Philip Johnston (52:54)
It's supposed to be mid August, we have a launch provider dependent. we'll see. There's something inkling it might get pushed back. We're ready. We've been ready since May, by the way. yeah. Anyway.
Mat Vogels (52:59)
Yeah, there you go. Exactly.
Yep. There you go. I was gonna say this is on your side. This is a logistical hurdles of trying to ship
something into space. Yeah.
Philip Johnston (53:11)
Yeah.
Mat Vogels (53:12)
Sam, how about you?
Sam Mendel (53:13)
Yeah, so mean, the team is fully focused on getting our 200 kilowatt pilot unit out. So this is on the floating data center barge. And we'll scale to tens of megawatts soon after that in terms of our deployments. I mean, yeah, if there's any interest in having the most power, water, and cost efficient liquid cool data center space on Earth, let us know. We're happy to get started with that and love the conversation.
Mat Vogels (53:39)
Love it. Thanks again, guys. Thanks for being on. I'm sure that we'll try to do a follow-up episode. then next year, hopefully you guys can do some voting for the next Black Flag 100. And we'll see if we get a whole new lineup of folks. My guess is that you might have a good chance of being on it again. We'll see. Both of you guys are doing exciting stuff. So thanks again for coming on. And we'll catch up soon.
Philip Johnston (54:01)
Thanks so much, Hamlin. Awesome. Thanks a lot.
Sam Mendel (54:03)
Awesome. Thanks a lot, Mat.
.webp)
