Robert Murphy, CEO of Othersphere

Ashby and Dane welcome Robert Murphy to the TTI podcast. This episode unpacks how Othersphere helps developers and investors identify the best locations for data centers and other energy-intensive infrastructure by combining global spatial, infrastructure, environmental, and community data.

Transcript

SPEAKER_02 0:15

Technologized investor podcast. It's a beautiful day here in California. It's summertime at Stanford University. A little hot, but not as hot as Europe. So that's nice. I'm your host. I'm Ashby Monk. My co-host is Dr. Dane Rook. And Dane, this is episode number seven. I think I need to update the website because I keep telling people our brand new podcast, but I think we've been doing this for a year. Toddler stage, man. Toddler stage now. Toddler stage. Good call. Well, one of the things we were pitching this podcast about, maybe, you know, a few months back, was that we were going to help investors understand the robot invasion. And I have to say, there's some updates in that robot invasion that make it look quite credible, like the US government putting export controls on anthropic fable and mythos. Any takes? Are we about to have AGI or even ASI happen?

SPEAKER_01 1:10

That's such a wrinkled question because again, the benchmarks for what actually constitute AGI are so diverse and it's almost like you have the students writing the test in some respects, because the organizations and the folks at the organizations that are trying to build AGI are effectively some of the folks that wrote all of these various standards and tests for what AGI is presently. And in the current frothing environment of pre-IPO companies, where there is massive interest in being able to prove out AGI capabilities at some of these larger, very late-stage entities. I'm not quite sure we're going to get a totally objective slant on or view on things just yet. But I think that there are signals that I don't know. I am I am a bit of a pessimist in terms of the longevity of the current approach to LLMs getting us fully in AGI. And I think there are some signals in terms of high-profile departures at some major AI research outfits that are kind of indicative of that. But that's always blended with are these folks leaving because they actually have a different perspective than the mainstream, like Yan Lagoon, or is it that it's just so lucrative and we're such a bubble environment that them leaving to start their own unicorn company basically just on the back of their name is in their best interest. It's just there's so much murkiness in my mind right now in terms of what the tea leaves are portending for the next couple of months, if not next two years.

SPEAKER_02 2:36

I do appreciate the tinge of skepticism, Dane. I think I live in a world as a you know part-time on-the-side venture capitalist where many of the people I engage with are in quotes AGI pilled. I'm supposed to say that word apparently. I hate saying that people are pilled, but apparently, like coming out of the matrix, that's what we call people who have taken the pill and believe that this is all headed towards, you know, uh a simulation, which we should talk about another time on another podcast, by the way, because we might be in a simulation already. That's for another time. Look, I think, just for the historian's benefit, that we are at this really interesting moment where we had SpaceX go public a few weeks ago. We got Bernie Sanders demanding a new sovereign wealth fund to allow people to participate in the upside of AI. And at the same time, Bernie is out there saying, we got to stop data centers. We need to figure out how we're gonna actually power this AI boom. And so it is just magical. It's almost like it is a simulation that we can invite Robert Murphy from OtherSphere to talk to us about this insanity. Robert is the guru of energy, but also is building a platform to help capital allocators understand all things related to the digital infrastructure needed to power AI. And so we're really excited to have you, Robert, on the show. Welcome. Thank you very much for having me. Well, we have a standard format here, which starts with who are you? I was about to bring up that thing we mentioned at the at the at the top, which I'm actually not going to bring up unless you want to talk about that part. Oh, my trans hood and my sorted. The fact that you were on a terrorist, you're one of the few founders I know that may have been on a terrorist watch list.

SPEAKER_00 4:22

Uh yeah, like a light one, like a kind of a pre-serious version of that. Yeah, yeah, no.

SPEAKER_02 4:27

How does one let's go back to that? How does how does one end up on a terrorist watch list? Even if it is the lightest watch list.

SPEAKER_00 4:36

Um, it wasn't a huge well, actually, it was a little bit of effort. Well, first of all, I feel like I shouldn't say anything. I think that build out that pre-embolate intra Ashby. That's everything I will say now will be downhill from that. Oh no, I know you well enough to know this is just the beginning. Thank you. So yeah, I I'm Canadian originally, but I grew up bouncing around because my family worked in oil and gas. And so lived in Yemen for a number of years as a kid. And then when I was, I moved down to the US for grad school, and at the same time, I was doing a lot of work on like nuclear energy, nuclear non-proliferation. So ended up by hook and crook in Libya and Iran and North Korea all in the space of a year, but like as a tourist, effectively. And so that triggered sort of the pre-AI kind of the intelligent at the time databases, but today's uh by today's standard, probably pretty low. And that I was on the very long-form watch list, like the mid-2000s ones where there are lots of people there, and I just basically was randomly screened every time. And so randomness 100% of the time is not. And then one time I just happened to be lining up for a list in the or uh for a flight, and the one that counters like, Oh, you're on our list. I'm like, oh, watch list, watch list. He's like, Oh, the watch list. Uh I had a friend that DHS the time and he checked and he's like, Yeah, yeah, you're in there.

SPEAKER_02 5:37

Your name, I guess I've got to imagine by putting you on the watch list with a good what I have to imagine is a good Irish name. Seems like a very generic name.

SPEAKER_00 5:47

Robert Murphy is very bog standard. I could say like I I don't feel like my name is going to be any of my like big calling card or like sort of um thing I hang my hat on in life. And so, yeah, Robert Murphy, there's a few of us floating around, Irish people in various descent have had, you know, flirtations with the with the world of terrorism at various points of the year. So it's not shocking that I would that I would be there. And as I said before the call, yeah, I just started booking tickets with my middle name, and all of a sudden it went away. So the databases weren't that clever at the time.

SPEAKER_02 6:15

Just feels like there would be a lot of people caught up in a lot of Robert Murphy's caught up in that watch list, you know.

SPEAKER_00 6:21

Yeah, and I mean, in fairness to them too, I mean there was a bit of a profile there, like what that travel history, like that, I would want to have a second look at anyone who had been wandering around to those places for a while, too. Um, but it was very, but it was good for perspective because then you see a lot of you see, I think, and like living in the US as well, like seeing countries from the inside gives a lot of perspective than how other people look at them. So it was very educational, interesting in a lot of ways.

SPEAKER_02 6:44

So, how did you get into the data center world? Because I when we first met, you were actually the world expert, I'll say it why not, on hydrogen.

SPEAKER_00 6:56

Not well, yes and no. So I I got into this by a long path. So I started out at a firm, uh, oil and gas consulting firm in DC that's now part of SP Global. I was at the Cleantech Group in the Bay Area in 2008, just in time to watch Cleantech 1.0 go pop. It was very educational.

SPEAKER_02 7:12

I was here too.

SPEAKER_00 7:13

Yeah, it was interesting.

SPEAKER_02 7:14

Around the pop, just after the pop, actually.

SPEAKER_00 7:16

But yeah, it was interesting. A lot of related pops, and cleantech was, you know, global financial crisis, sort of collateral pop. But uh went back to DC, I was at the World Bank for a couple of years doing energy and mining projects, and then I spent some time at Enbridge, Canadian midstream developer. So a lot of big pipeline infrastructure, wind and solar uh sort of project footprint in North America. We were looking at international projects and building out their strategy and footprint abroad. Uh then I came back down to the Bay Area. I was a Chevron for about five, six years. Most of my time there, I was their lead economist for emerging markets. So, where is energy demand going? Where's you know, geopolitical issues going? If I was still there now, it would be in the thick of Venezuela and Iran and Straits of Hormuz questions, and it would have been fun, interesting? I mean, to a degree, but it was also like all these things are sort of known and they're boiling along for a long time. And so eventually when they happen, though usually they all wouldn't happen at once. Uh it does make things very interesting internally in terms of like how do you do capital planning for a company like that when you're really uncertain, like there's a huge range of volatility in terms of where oil and gas prices could go. So, yeah, then it was there for the five, six years, and then I moved over to the startup world. I was like, okay, I want to work on climate in particular. I want to sort of, I grew up around the energy sector. I'm totally sold on sort of all of the social and political and economic benefits. But then how do you do that and provide that, but do it better? And so looking at climate in particular was sort of where I decided to spend some time. So I was at a startup, uh, environmental data startup, Acclama in the Bay Area for about five years. We went through our series A and Series B rounds, learned a lot in terms of startup life, definitely went from chevron to startup was the a big shift. And then yeah, I landed somewhere in the middle after that. I was, I moved over to Breakthrough Energy, which is sort of Bill Gates' climate vehicle for funding a lot of energy-intensive and uh sort of commodity infrastructure and like how do we produce low-carbon hydrogen, steel, cement, and so on better and get those technologies out into the lab, out of the labs and out into the real world. And so I came, I overlapped with hydrogen there, and I had touched on hydrogen a little bit before, but it had always been on the edge of kind of that the more core question on energy and how do we provide global prosperity without, you know, breaking banks and everyone's pocketbooks, but also brought our systems at the same time. And then we met.

SPEAKER_02 9:28

And then we met. And I feel like I met you when you were first thinking about the spin out from Breakthrough, which by the way, we have deep connections at Stanford into the whole breakthrough energy, breakthrough energy ventures um community. So we knew a lot of the same people. And then, you know, I actually think you were still a fellow at Breakthrough when I met you, and you said, I think I'm gonna go and build a company to do siting of hydrogen. And it had just been after, I think maybe it was the Inflation Reduction Act had had kind of passed, or maybe before actually, it was even pre that.

SPEAKER_00 9:57

It was that was bubbling along. And it was definitely, I mean, so it was less it had been drafted, and then there was the period after it was released, and on things like hydrogen, there was a lot of time waiting for the actual rules to settle, and like what are they what are they applying and where what are the incentive cutoffs and so on. So it was right in the thick of that period. But well, I think we met beforehand, and yeah, I mean, the whole concept for others here was a lot of what I was doing for the teams at Breakthrough as an advisor to the startups uh through their fellowship program was where do you how do you scale? Like, okay, so I can help you think about fundraising and team building and so on, but really my bread and butter is like when you get to that kind of nth of a kind scale, you've got your cement process, you've got your sort of hydrogen production process, and you want to make that matter in real markets and you want to compete with incumbents. And a lot of that is like building in the right locations where you know you have off-takers and customers nearby, you know, power costs or energy costs are cheap, maybe especially relative to the incumbents, you kind of have a nice competitive advantage. Or for a lot of them, it's like be near the end customer so you can basically just beat the incumbents on shipping costs. You can do small modules or production of ammonia or something right next to the users as opposed to a long way away and having it shipped. Same thing for Cement. So that's what I was doing. And then we just started talking. I did not go to breakthrough to start a company because it's really not a great life choice in many ways. It's it's fun, but it's not sane. But yeah, they were really kind and we started talking about some ideas about doing what I was doing, but at real scale. And I'd been pondering the concept behind Other Sphere, really since Enbridge, in the sense of like how would you diligence all these projects? Like I was looking at window solar projects or pipeline projects there. How do we know they're the best opportunity? And how do we spend two months on each one? And then things just started to fall into place.

SPEAKER_02 11:34

So let's do it. What is Other Sphere? What does the company do? Both both because there was a slight pivot part way through. You can tell us the original story and then how that led to the pivot, and then what is it doing it now?

SPEAKER_00 11:47

Yeah. So basically, what we do technically, we break the world into about 180 million tiles. We basically pixelate the world. And each of those tiles, we're signing time series data, infrastructure data, environmental and human data. So each tile, then you look at and say, if I built any project there, whether it be hydrogen or a data center, how would it perform based on fundamentals? So now we can fully model a project. And for hydrogen and for other physical commodities, it gets interesting too, because then they have a really complex physical footprint. So, like, how is my power, my gas coming in? How do I ship out all my materials to my end users, my off-takers? And so you really have to model all of that well to know is that location a good idea? And so what everyone ends up doing is they get, you know, a real estate broker or a banker, someone will come to them say, I have some land, you should build something here. Uh, or they'll be out looking and they'll try to sort of narrow down a list to 10 or 20 or less sites. And then you have to go through a long process to sort of really diligence that understand, is it a good idea? Like, am I building in the right location? What are the hidden risks, the red flags, the long-term fit? And is it comparatively the best? Or is it just okay? And so what we really do is we pre-model the world so that anyone doing infrastructure diligence can make these decisions faster. They can review more opportunities, but then ultimately that means they're building a better pipeline of opportunities. More things are getting funded or built that are stronger in terms of fundamentals, and so they're gonna have better long-term performance. So it's it's two sides of the same coin, whether I'm a data center developer or a data center financier. I care about those things in just different orders.

SPEAKER_02 13:11

Definitely there's people listening to this that are sitting on infrastructure teams and pension funds or sovereign funds or insurance companies. What you're describing sounds like it's part of a you know an investment screening process. But it's also like you say, the other side is the developer probably wants to actually understand where to go do greenfield or which brownfield assets they want to convert. And so even though you started with hydrogen, you know, you probably had 95% or maybe it was 88% of what you needed to do data centers.

SPEAKER_00 13:42

Exactly. The whole concept actually always was like starting with hydrogen was a means to an end because you had to do hydrogen well to do like ammonia or methanol or steel. And so a lot of the downstream derivative or like sustainable aviation to a lot of a lot of commonalities there. So all of the sort of lower carbon and sort of sustainable call it commodities, hydrogen was a really good spar to start for many reasons. It also means it's very easy to go upstream to like on site or sorry, uh, ground-mounted solar, onshore wind, battery storage at grid scale, like all of those things lean on the same data sets, the same understanding. And then when it gets really interesting, how you chain them together and say, hey, is this site useful for three or four different things? Or I have a portfolio that includes power generation and data centers. Oh, actually, how do I put them together in the most complementary way? So the whole idea of pixelating it into that sort of standard index of data for infrastructure is so that you can look across multiple commodities very easily. So starting with hydrogen was just sort of a, for us, a means to an end. We went through a few different sectors, data centers we started working on about two years ago. Um, that was less of a pivot than I'd say more of like kind of a natural progression into honestly where the most money and interest was because as you mentioned, the Inflation Reduction Act, and so not just in the US, but globally, I think policy enthusiasm and certainty around climate has fallen. Inflation and like people's day-to-day pressures have increased. And so the tolerance and interest for doing things on climate that push some of these new technologies into sort of the money sooner is lower. Whereas data centers has the opposite problem of no shortage of froth. So there's very we we basically followed where the market was because uh you know, as a small company, you kind of have to. So that I mean then the other thing is yeah, we sort of focus on we brought, we did never had anything with generative AI, going to what Dame was saying about HGI there and where all this goes. We were using a lot of kind of quote unquote old school AI forever in terms of big analysis and creating data sets and so on. Generative AI, we were using for some coding in like gradually over time. And these tools two years ago were far less capable than they were a year ago, they're far less capable now. So that curve in terms of internal utility has definitely changed. And then about six months ago, we started releasing a new product, which is saying we've we've mapped the world, you can search these 180 million locations, and then we can dive in detail for sort of detailed site diligence, and that's where we wrap generative AI around our data to say, okay, what is a permitting environment, the community sentiment, a lot of the things that we're gonna do. I was just gonna ask in global data sets.

SPEAKER_02 16:04

You're doing all this spatial analytics, you know, and and it would be obvious for me that you're collecting information about permitting regulation, all that stuff. But the backlash from the actual people in communities is something I've been curious if you track. Like, was that part of the algorithm to help spot the best locations? And if it wasn't, is it now?

SPEAKER_00 16:27

It was we brought in the best we could in terms of proxies from even from back in the beginning. So even building a hydrogen plant or stable sustainable aviation facility, fuel facility, community sentiment, permitting realities, they always matter and they matter a lot. But the data there is very flimsy because it really isn't. There's surveys and there's sort of people, oh, there's trackers of like state level, county level moratoriums and bands and so on. But that fine-grained detail, like back when there's a Chevron, there were projects that I was sort of not I was aware of and sort of giving and feeding advice into on looking at like Twitter and Facebook uh signal for likelihood of municipal level fracking bans, for example. And that's kind of the level of detail you need to get down to, but it's hard. And and I think a lot of the information you would use to inform that has gotten like messier and messier over time. Like the Twitter and Facebook 15 years ago were pretty calm, sane places, I think, relative to social media now, in terms of you know, how much is reality there versus how much is sort of the different sides talking at each other. And so it's really hard to, I think, draw out good information. But the data center backlash is real. It's understandable for a variety of reasons. Some of it's driven by misinformation and misunderstanding, and that's a fundamental challenge, I think, with every bit of infrastructure we try to build. Like I think our biggest problem with climate is that we just can't align on what we're trying to solve for and agree on where we should build things. And it's a human problem, it's not really a technology or capital problem. And no, it's no different for data centers.

SPEAKER_02 17:51

Does outer space solve all these problems because we're just going to put data centers in outer space? And will data centers in outer space kill your business?

SPEAKER_00 18:01

Uh no. No, data. I I I think no across the board. Okay. I like I love the theory. I like it's intriguing space stuff. I've been in uh you can't say over one of the video, but I like at the picture of the far side of the moon behind me from one of the first was a Russian satellite that took some of the first photos. So I love that whole world. I would have been an astronaut, but I'm too tall and probably not smart enough and not fit enough. Like every reason, not, but it would have but it would have been a one otherwise if I wasn't even remotely kidding. Totally. It would have been one. It would have been one. I had it nailed, except for every reason that I didn't. But um I think it's I think that economics are really seems tricky. There's a lot of the physical things like like heat rejection and everything everyone talks about. The piece for me is ultimately it's just like actually what we've done down here on Earth, too. It's like theoretically, any of these assets or projects or ideas could be good individually, but when you do them at scale and we do them on top of each other, things get crazy. So like the Kessler syndrome. Define that. Tell me. So I believe it's named after a person who worked at NASA. And again, I'll this is like I'll caveat this is my rough memory of uh reading on about this a couple years ago. I think it was someone who worked in NASA in the last few decades. Um, and base of the concept, it's a bit like that movie with uh Sandra Bullock, where uh a space uh satellite or something explodes, and then every hour there's a kind of like spinning uh debris that's breaking up her space station. And so that's sort of basically what happens is that you start to have satellites and infrastructure in space break up, they hit other pieces, and you sort of get a low Earth orbit Vitamix spinning above us. And then then maybe people get to make businesses out of like space trash trash collection. So it seems like that would be an interesting problem to collectively solve. And we don't solve those collective problems well on Earth, so I'd be a little skeptical of doing well in orbit. But um, I mean, interesting conceptually. I think there's a lot, I'll actually put this way, I'll frame it differently. There's a lot of land left on Earth. Like our whole point is that we have bit of other sphere, is we have eight plus billion people, and it'll be more before we're done with all this uh to provide for, and we have to do it in the most efficient way possible. So, how do we like deploy infrastructure so that we can kind of have our cake and eat it too? There's a lot of land left on Earth that is good locations for data centers and various forms of energy production, and there are trade-offs in all of it. And how do we like reveal that and then let markets and policymakers and whomever decide how best then?

SPEAKER_02 20:16

So the name other sphere isn't about finding other spheres like Mars. It's about seeing our sphere differently.

SPEAKER_00 20:24

Exactly. It's actually trying, it was, yeah, it was the concept is like we have like biosphere and cryosphere and anthroposphere and sort of geosphere. So we have all these sort of spheres of both scientific and human understanding, or like physical realities and human realities and environmental realities and so on. And so other sphere, the idea is like how do you the connective tissue in between? Like, how do you think about the interactions between all of them? And our, I think, human biggest footprint on all that is when we uh like infrastructure, energy is how we most disrupt or interact with or complement or even benefit some of those sort of natural spheres, call it. And so it's a balancing act, basically.

SPEAKER_02 21:00

What do your clients come to you for? Like maybe another way of saying it is like who specifically are clients and and like when they say, hey, go call Robert, what are they solving for?

SPEAKER_00 21:15

It was different, different things over time, honestly. So when we originally launched, the concept was uh like around Breakthrough talked a lot about this idea, like the green premium. So the green premium is sort of the cost differential between say existing steel production and like a lower carbon version. And usually it's a bit more expensive to make a lower carbon commodity because commodity markets are very good at internalizing the right technology to lower costs, but they have a really hard time passing through any higher costs. So they're the idea is a green premium is like where can you sort of compete best? So for us originally, our clients were developers and financiers, and they were where can I build infrastructure or fund infrastructure that it will be highest performance and it can scale best? And so they were trying to deploy capital around sort of, you know, climate, new energy technology. Thesis and do it in the most intelligent way possible and do it based on fundamentals. Because I think everybody wants to have policy at their back, like policy so tailwind and someone at their back, but it's tricky to rely on it, especially over the long, like 20, 30 year or more life of an asset. And then that shifted. I think data centers has become much that as a class, uh, asset class, and the teams working on it, it's much more about how do I get things built quickly? Where do I find energy? Where do I find locations where all of the sort of uh realities and sort of fundamentals on the ground will let me get my GPUs online sooner so I can get my tokens out the door fastest and make my money as quick as possible. So low, like early, early days, it was like green premium, low carbon. Now it's more about speed to market. And then I think that's evolving a little bit in two vectors, which is the community piece you mentioned is like, oh, wait a second. Actually, these are highly visible assets now, or effectively, I mean, data centers are sort of refineries. You're turning energy into information as opposed to like jet fuel. So now I'm building a refinery and I'm finding out communities don't like a refinery next door. So they're having to deal with that reality. And then I think, particularly from a financial perspective, like financier perspective, there's so much froth in the market around AI demand at a root level, at an application level, and then stepping up to then AI demand for infrastructure and compute and data centers. They're, I think, trying to put money into these markets. So there's many, many multi-billion dollar funds announced focus on AI infrastructure and so on. But a lot of it's, I think, struggling to get deployed. It's like, where can I put this money that's safe? Like, will these data centers be needed? Will this land be needed? And it all kind of chains back to what Dane was talking about at the outset there. It's like, well, does how much how much fable model compute do we need and who's going to use it and for how long? And all of that cascades through into like how many data centers do we really need? Um, how many gigawatts and terawatts of data center capacity does Texas really require? Uh TBD.

SPEAKER_02 23:49

So are we in a do you think we're in a bubble right now? Like, are you able to, you know, you maybe put your economist hat on? Where are we in this cycle?

SPEAKER_00 23:58

I would I would be, I don't know how we couldn't be in a bubble. Put it that way. Like if I had my chevron hat on, um, and I don't anymore, just to be clear. I have no, like nothing touching that world at all. But like what I was doing there in terms of demand, like an energy demand, in particular energy pricing. So a lot of it comes back to demand. And I think the problem is like really no one knows what the root demand is for these models, and that change like a whole bunch of factors play into that, right? It's like the high cost frontier models versus sort of the open uh weight, open source, cheaper ones. What are people gonna go for? What are they willing to pay for in the higher end ones, how much they use consistently, what is the end user demand? So, like for software engineering, I think AI has proven to be helpful. And now you can like really do a lot of coding, and we're like internally figuring that out constantly. Like how much coding can, how much can we rely on agents for coding internally? Fine. But then you try to bring AI into other real world applications, like say in energy, and probabilistic answers don't work well when you have like high explosives at your facility and you have a very deterministic system. And and then say it's true for finance, right? It's like I made my model, I made my spreadsheet, I thought this is how the company was gonna perform. And I put by like I put my stamp on that. So can AI help me build that model better and faster? Probably, but at the end of the day, it still has to go into a system that has certain like requirements or uncertainty. And AI is fun, like LMs anyway, are fundamentally a roulette wheel, and they're just getting better and better at spinning them and aiming them. And so I don't know where the long-term demand lands. I think so the bubble piece then gets to like we don't own demand for sure. There's a lot of heavily, heavily funded companies that are basically trying to all own the market in parallel. So I don't know how that doesn't sort of lead to some degree of overcapacity. Um, you have the sovereign piece on top of it too, where then comp countries are saying, well, I need to also win this. Yeah, and so you have a lot of, and then you just have the hype, and then honestly, the little bit of dread works too. Like anyone like a an overpaid brain, like we all are, how long will I be needed for? What I what are what is my role on the far side of this? So there's sort of this pervasive thought around this that I think leads to then a lot of interest and bubblishness. And so you see SpaceX and other things sort of coming out and where those the market seems to land. I don't know how we couldn't. And then the and the last thing I'll say is also we're still in the very early days. Like we're in the subsidized Uber ride phase of AI at the moment. And so, how much are people willing to pay when you're actually paying the real bill as opposed to the VC sponsored bits? And so it's like Uber in the early days, but Uber also didn't have to build the roads, and so I don't know.

SPEAKER_02 26:18

Oh my god, what a great point. Um, look, there's probably so much we could do here. I I I do want to like understand so one of the things that we spot are new financialized investable markets, and compute is becoming an asset class. There's derivatives being built around compute, there's all kinds of options and pre-purchase agreements and even new accounting treatment coming online. You know, to me, it sounds like this is the new oil, for lack of a better term. But at the same time, like, you know, there's also skepticism around how big this gets, and maybe this is a little bit like crypto or like Uber at the time. What do you think about the companies being built around compute as an asset class?

SPEAKER_00 27:05

I think it's a real there. I think so. When I left the in between ACLA, the environmental data startup and breakthrough, I had a little project on the side for when I helped one of the blockchain networks set up their decarbonization program. And so that was my one and only little dip into the crypto world. Right. I think crypto and AI, like I never totally understood the utility of crypto. I can see it. If you squint hard enough, the need for it in certain applications 100%. AI, it is like we use it daily. So I can feel the value at any given moment. And so I think there's a totally different reality of like there, there in terms of the value of AI. But I do think, yeah, that someone said this to me while I didn't think this one up, but like AI and LLMs and sort of the you know, anthropics and open AIs of the world tend to end up like airlines, like sort of commoditized providers of a commodity product that are really fighting hard on cost and like little micro features and little differentiation and so on. And I think we're already there. And then I think what's interesting is what you're describing in terms of like the um commodification of computes and then financial models and businesses setting up around that is a lot like oil and gas trading. It's like there's lots of people that are there to help the arbitrage be fast and smooth and commodification be like extreme effectively. So that's where I don't understand where like especially some of the big uh model providers go in the long term, is that you can't, you'll only, it's actually like what I was saying about breakthrough and the green premium. They'll only be able to charge what the market is willing to bear. And if you're competing with open source, open weight sort of models and so on. Like you have the route, Sam Altman did a post a long time ago now, a long time ago, like a year ago, but it was basically like intelligence gets reduced to the cost of energy. And I think that's true. There's certain input costs, you'll have the capital costs of the GPUs and the buildings, you'll have energy. And then beyond that, there's gonna be just really thin little margins for people to provide the basic models. And then you can sort of stack the applications on top, but even those get very easy to build. So I think there's a long-term commodification of intelligence in general that leaves the value to you know, who has the money, who has the insurance, who has the land, who has the energy. Uh, like I think about us for uh for us as a business. I don't know how much the world needs going forward in terms of insight businesses or how, yeah, like the margins for that, the willingness when then all of a sudden you have like we we used Fable for a while before it was pulled back off the shelf. And it was, it felt like a meaningful upgrade from OS. Yeah. And and I think if that's the sign of where things go, then Yeah, we have some really interesting times ahead, I think. It's awesome. In exciting ways, in the as the curse goes.

SPEAKER_02 29:28

Well, it's definitely maybe the single biggest thing investors are asking questions about. It's not the biggest part of their portfolio, it's not the biggest part of their exposure, their risk budget, but it is this transformative engine that is kind of taking hold of everything. Everything they own is now wondering how much are their energy costs going up. And you see, you know, the hyperscalers going behind the meter, building their own energy facilities to power, to power these data centers. But I've been talking too much. I want to give my co-host a chance to wax poetic a little bit here, Robert. Dane, any deep thoughts?

SPEAKER_01 30:05

More one quick question for Robert, and that is related to complexity. A lot of the folks that Ashby and I interact with the most strongly in the academic front are technologists and data folks within large financial institutions, primarily public pensions and soft and wealth funds. And listen to what you're talking about, what other sphere is doing. There's a lot of echo in terms of the problems that these organizations face, these types of operators face, in handling complexity when it comes to integrations of so many different forms of data, so many different levels of quality of data, and being able to compress that all into insight. And you're talking about rasterization of data, et cetera. Um, I'm just kind of curious, for their benefit, as well as my own edification, how so what if if you have any sort of core learnings or thoughts or approaches that Othersphere uses to handling that degree of just gnarliness of problem when it comes to working with so many different angles of data? Is there a prioritization schema you tend to lean on? Are there any sort of frameworks that you utilize, um, or just any real painful learnings that you would love to share?

SPEAKER_00 31:03

Yeah, happy to. I love, I mean, fundamentally we got 3R, we're like a complexity reducer. We we make some of our own data sets, we derive some of our own data sets from what we bring in, but generally what we're doing is bringing a lot of information together. And so there's a lot of technical learnings there and even legal and like licensing and how do you like work well with providers. And again, and that's also the you know, like data as a moat, I think it's trickier and trickier over time as well. I think one of the key things I took away when I when we first started Other Sphere, the idea alignment was a big underlying problem I would have loved to target, which is saying, okay, the world can't agree on like where to build a data center. Okay, well, if we put all the information in front of the regulators and the developer and the financiers and the communities and so on, we could at least raise the bar and they can have a better conversation as opposed to one that's driven by misinformation. And I think this to the degree that's still possible and it's worth trying to do. But then the reality is that the like whose data you use is half the problem. So one of the things we ran into in the first year when we're building was that we'd meet teams and we had built this big SaaS tool to let you search the whole world and be like, well, this is great. And we actually don't have 90% of this data, so we just leave that in there. But we have our own 10% for this, that, or the other type of asset. You've got your refinery data, you've got your sort of uh ammonia plant data. We want to bring in our own. Like, oh, okay, that makes a lot of sense. Um, but it's tricky for us internally, just from a space of like, how much do you customize for one user as opposed to try to provide as a common platform to everybody? So I think there's an ideal version of the world where you provide a lot of common data and then people make, they make, they change, they decide their priorities on top of that common view of reality. The unfortunate truth is that the reality itself is what a lot of firms, well, like I subscribe to provider X, no, I subscribe to provider Y. My oil price forecast looks like up, yours looks like flat. And so you end up with different versions of the reality to then build into your models. And so that was that's been a constant struggle, honestly. Like, how much do we provide a common view with the best data we can put together? We do a lot to vet every data set and every data provider, or do we end up being better off strapped to like some big data providing firm who they have all of it? And so we're now sort of an application on top of a big data provider's sort of existing internal data sets, or do we go in-house somewhere and say, okay, we are we are going to bring together, you subscribe to 500 different things. We'll bring them all together into one platform. It's not truth, it's truth based on the subscription pattern you happen to have. And so it's tricky.

SPEAKER_01 33:28

Oh, yeah. I mean, this this rhymes with a lot of the daily headaches and heartaches that uh CTOs and COs are dealing with in major public pensions. Um, so quick question. Ashby and I actually are both geographers by training. So I'm very curious. You had mentioned um there's a whole lot of land left on Earth uh that is potentially usable for data centers or other other energy-intensive infrastructure. Any sort of quick little thoughts on where it might be underutilized in terms of an opportunity? That would be a cool answer to a Jeopardy question. So are there far-flung geographies that no one's thinking enough about?

SPEAKER_00 34:04

It's so it's yeah, I so honestly I would love that. If that was my only job, I would that would be I would love that um to just answer that question. So it really depends. So on the asset type. So for example, when we first started out, hydrogen production is tricky because hydrogen is hard to transport. But there's a lot of companies who are being pretty creative, say, I'm going to produce hydrogen in, say, like southern Chile, turn it into a derivative fuel, and then ship it to Europe. So all of a sudden you're basically turning like Chile or Namibia, who have good wind and solar, respectively, um, into the but it's kind of like captive. You have an energy resource that has no end user, all of a sudden you can send that end user uh the sort of ammonia or whatever, this uh the jet fuel. And then data centers are an interesting version of that too, because you can in theory say, especially for like AI training data centers, who you're not worried about latency. And I again, I don't know how many of those really need over the long term, but like I was originally born in uh, well, I was born in and originally from Alberta and Canada. So a lot of landlocked energy, oil and gas, among other things. And all of a sudden, oh, I can put a data center right on top of the gas field and not have to worry about shipping it all the way down to the Gulf Coast or whatever. And so you can mop up stranded energy with data centers in a very interesting way if you're not worried about latency. So then what gets interesting though, so we have a score for every asset that we create. So we have a data center score, and then we have team companies will do their own. And so the answer to your question is tricky in the sense of I can tell you where I think the extra land is. But then someone will come along and say, well, yeah, I see what you're doing here, but I would wait water stress differently, or I would wait distance to fiber differently, or like I'm doing a training data center, so I don't care about latency. So don't even show me that at all. Um and we've done some really cool work in terms of like the reachable users from any given point on earth. So we did a blog post a while ago, like Tibet. Near Lhasa is a great spot to build a data center if you want to reach the most people in the world most quickly, because you're right in between India and China. Um, though there's some geopolitical and other and energy realities, it might mean it's a trickier spot to fund something. And so it comes down to preference and thesis. Um, and so, but you can look at, I again have more time for this type of thing. Uh, like Google announced a project with Intersect Energy a while ago, I forgot I think it was in Texas, I want to say, and it was going, it's a data center with a lot of solar attached to it. And so what I wanted to do, it'd be interesting to do like a little post and to say, okay, take the characteristics of that land and then extrapolate out across the rest of the world and say what other pieces of land have those same characteristics. So have good solar, the right fundamentals for data center, so on, so on, so on. And then you start to reveal where those opportunities are. And then it gets really interesting when you overlap them and say, hey, that land's good for data center, but also solar and also SATH, and also this and also that. And then you get to an interesting land valuation engine from that. The other thing that I think happens is if the data center bubble, if there is one, arguably, uh pops, then all of a sudden you're gonna have a lot of people who spend a lot of money getting powered land, quote unquote, ready. And all of a sudden, maybe there's like people who have power-intensive assets who can take that and benefit from the lack of need from data centers for a while. So a lot of what breakthrough fund was a lot of electrochemical processes. So create like sustainable aviation fuel for power. Now all of a sudden you don't need to compete with Google in order to do that. You've got land available and someone's kind of built something up, and there's no one really to put the data center there, and they'll settle for a SAF project. Fascinating. That was a long answer to a short question. I'm sorry.

SPEAKER_02 37:17

No, I loved it. I am more looking at the time and wondering if I need to ask you my final question since we're coming up to the top of the hour. My final question is is built on everything you just said, which is if you are wildly successful with this company, what does the world look like 10 years from now? How have you helped all these players um transform themselves? How have you helped governments reallocate infrastructure? Just give us a vision.

SPEAKER_00 37:45

The dream version of this is what is sort of alluded to before. It's you know, we have eight, nine, ten, whatever we have to end up, sort of billion people who want the material prosperity that, you know, I'm in Canada, you guys are in California, like that we have, but we want to provide that without breaking everything in the process. So there's a again, it goes back to preference, but there's a theoretical way to deploy infrastructure most efficiently so we can provide material prosperity uh at the lowest sort of environmental and human cost possible. And that's got full of trade-offs and subjective viewpoints. So that's the challenge with this. But if we can basically help optimize the deployment of infrastructure around the world, that's our goal, and basically in service of human prosperity, uh, what I think that means practically is we have um more energy, like capital, new, more capital going to new energy technologies. We have um better performing assets because we're placed in locations of lower risk and better fundamentals. So our value is very commercial and tangible, but it leads to the mission if we get that value right.

SPEAKER_02 38:43

I love it. Well, and true uh data center needs and energy needs. Dane Rook's power just went out. So we just lost it.

SPEAKER_00 38:52

It wasn't my long-winded answer that scared him off.

SPEAKER_02 38:54

Not at all. His power just went out. He he lives across over the ocean, and uh this has happened before, my friend. So um, Robert Murphy, I'm just so excited about the work you're doing. The the initial project I came to Stanford to work on was rethinking how we finance infrastructure, how we do public-private partnerships. And I was the executive director of the Global Project Center, and a huge part of what we did was actually digital transformation and data centers. And it's really exciting to see all the work you're doing to kind of advance all of that localization theory and uh citing theory that you know was very immature when I was you know kicking off my my academic career. So um that's it. We're we're grateful. Dane, I'll just say thank you for Dane since he's uh he's in the dark. Um that's the Technologize podcast with uh Robert Murphy. Um check out other sphere, and we will be back with more um episodes and founders uh ushering in the robot invasion of the investment industry. Thank you.

unknown 39:59

Thank you, guys.