Josh Gilbert (00:00)

You know, we've been able to help in a very short space of time with a very lean and mean startup team, some of the world's biggest pension funds, of the world's, two of the world's biggest financial data platforms and indices, you know, countless large consulting firms, asset managers, investors across the world. We want to do more and more of that. And we want to continue to transform industries, but an even bigger scale. That's why we're in this game. That's what the trophy is.

Chris Mailander (00:30)
Several years ago, I was listening to Elon Musk talk with Dan Carlin and he was explaining the every great power, every great shift in geopolitical power was preceded by some fundamental shift in technology at the time. Then last week, Alex Karp, the CEO of Palantir spoke at a defense conference and was talking about that we are currently in the midst of a great competition.

And it is the use of artificial intelligence to create strategic advantage that how this would shift geopolitical power. This positioning of technology to create power is something that's fascinating. And we're in the midst of it right now that is accelerated by a couple of additional factors, one of which is that we have a new administration that's coming in that has promised to change the decision making processes of the US federal government and that trickles down into the very fabric of how the United States presents itself. It is a competition with China, Europe is falling away, Saudi Arabia is trying to get it in the game. This is the debate and we have this dynamic environment right now where artificial intelligence is accelerating the pace of change. We have a dynamic environment politically and economically which allows things to change.

What does that mean? It means that this is about competition. The way that the major players are teeing it up is in that framework. It is about who sees best, who perceives best, who has the ability to react fastest, and that creates the advantage. There is a lot more to this story that'll play out over the coming three to four years. And for that, I'm really thrilled to have today two experts in this particular field to help us walk through this battlescape. Understand how the market has developed to this point in time, where we are today and how we see it unfolding and how those players may shift around over the course of this competition over the course of the next three years. Joining me today is Josh Gilbert, who is the co-founder of Sust Global, who is based in London, today is living in Austria, as well as Gopal Erinjippurath, who is in San Francisco.

Together they are co-founders of Sust Global, which is a geospatial AI company and have been working diligently over the last five years to take really complex data sets and put them together in both a structured and unstructured way, interleave artificial intelligence into a market offering that has been realized or being utilized by a number of your leading financial institutions. And so they're gonna help me understand how this battlescape unfolds. Welcome Josh and Gopal.

Josh Gilbert (03:17)
Thanks for having us, Chris.

Chris Mailander (03:21)
So this battlescape is something that's, as I mentioned, is terribly interesting to me. And I want to set this up, this conversation up by understanding who the major players are in a way that what I see here is the armies gathering on the step for the great game, which was that, competition between the United Kingdom and Russia back in the 1800s for central Asia. It played out that great game. Another instance of it was in the cold war as United States and the Soviet Union and their allies and affiliates and coalitions competed for power influence over a period of time.

I feel like we have this shifting in the great game and it's underway right now and I want to understand it better. So the way to do that is to talk about some of the major players that are out there right now. And then we'll talk about some of the secondary players, tertiary players and how this battlescape may evolve and shift over the coming several years. One of the first I mentioned in the introduction is Alex Karp and Palantir. Palantir comes from this, has this extraordinarily interesting background coming out of the military industrial base and intelligence community. It has this bravado, braggadocio styling associated with it, but also comes out of a bit of darkness because of its work in the intelligence community. Its valuation is phenomenal. It's right now at $166 billion or so. And looks expensive when you look at it from a traditional PE valuation perspective. And yet the market is extremely excited about what they bring to the table. Palantir and Alex Carp very much position that company as enabling decision makers at government, at intelligence agencies, at corporations on how they can make better decisions. How to change the decision making process so that it creates competitive advantage for their clients. I think that's an extraordinarily important variable within this unfolding landscape. You all have studied Palantir and know it quite well. Tell me how you see it and their capabilities going forward.

Josh Gilbert (05:30)
Thanks, Chris. I guess maybe I'll kick off discussing a little bit on the way I see the commercial side of the business and maybe taking a step back what it takes to win.

I think then I can throw it over to Gopal and I would love to get his take on some of the technical aspects with Palantir. I certainly think that they're an interesting one because they seem like they've really broken into the public consciousness, but they've been around a long time and interesting even when Gopal and I founded Sust Global four or five years ago. We had a little sticky note and it was one of the sticky notes we put up on the whiteboard as you do in San Francisco as you're building a new company.

And it was Palantir for good, right? And Palantir does a lot of good. think they work with the NHS in the UK. They work a lot with police enforcement agencies, do a lot of important stuff there. What they've really been able to do well is find proprietary data processing approaches, but also find a spot that's really, really close to the customer. I think you have many of the companies that are collecting data or processing data are often relying on the channel to get the data to their customers. And I think there's a few that can do that. So the Fang or whatever the new version of the Fang is now, the Metas, the Amazons, the Microsofts. So Apple, if you look at the cloud providers, they process a lot of data. They have the big marketplaces. And then they say, hey, everybody come and do your thing through our channels.

I think the really powerful thing with Palantir is they said, well, we're going to build the OS as well as the customer relationships in some of these areas. So I think, in terms of what it takes to win and what I think they've done well, they capture a lot of data, but they process a lot of data. So they can actually process data at a very big scale, but they're deeply, deeply embedded in the customer workflows. And maybe initially their share price was depressed because many people saw them as consulting dressed as SaaS, and they said, hey, look, how does this model scale? How do you get out of the 1x multiple on consulting businesses that's often there? But over time, we've seen that this is really a data-powered business that is just really, really close to the customer and actually builds the operating system for the customer. So I think it's really smart, and I think it shows maybe them harnessing a lot of the technological advances we've seen in terms of having tremendous data infrastructure, but also a cost-effective data infrastructure that ultimately gives them this asymmetric advantage where they've been so close to these customers for so long, they've worked with lot of these really big customers for so long, that they're perfectly placed, I think, to harness a lot of the tech things that have happened in the last couple of years with AI, with data processing, with even geospatial, which is, of course, the domain that... that we come from.

But yeah, I just think ultimately they've done a really good job of understanding multimodal data sets, some of the different internal data sets of the client, as well as all these other data layers and creating a mesh or what they call an ontology that ultimately is consultative upfront, but massively, massively scalable in terms of the revenues on the back end.

Chris Mailander (08:57)
Another one of the major players in this particular space is Nvidia, which also has this phenomenal valuation, $3.4 trillion right now. And in some ways, is the chosen one within the AI space because they allocate out the chipsets that are used for the tremendous computing power which is required for AI-based computing. And yet are doing really interesting things on the software side that I think that we should talk about a little bit to understand more so beyond what they're doing on the hardware side.

And then an interesting announcement this week as well, which allows them to actually be able to do AI computing at the edge. So one of the things that we should talk about as part of this is some of the major players, the Googles, the Amazons of the worlds, the open AIs want to consume that data, centralize it, and they become the control layer for how a client uses that. Whereas what Nvidia is doing with some of the newer that they're putting out there would actually allow AI computing at the edge and retaining the data sets on a distributed basis. Very different strategy. And when we look at this as a competition, when we look at the battle, could unseat power as part of what's unfolding. Talk to me a little bit about that, the different strategies of these different games, of the models that are being used, the technology architectures, and how that plays out to create advantage.

Josh Gilbert (10:20)
Many of our customers would see this type of capability as super relevant. Having an internal data lake and being able to process on the edge where you don't have these proprietary or off kind of approaches where you have to send the data back. And actually, ironically, a lot of these consulting businesses, which I likely to kind of be toast in the long run, in my frank opinion, they're actually really thriving by being able to build these on-prem LLMs for some of these large companies where they don't have to then go off to the open AIs or the Microsofts and they say, well, where on earth is our data going? So I think that's an interesting trend there. And of course, I guess maybe most of the focus is on these big cloud AI companies that are obviously running away with much of the winnings today on the battlefield. I think maybe one of the interesting things I see is they are undoubtedly going to be the runaway successes in the future. OpenAI maybe has lost some ground to Google in recent days in terms of new updates from Gemini. I know Claude is outperforming many of these OpenAI things, but they've got maybe the first mover advantage in the communication there. So think those guys are going to keep on crushing it. The way that I would look at this is maybe, you have these runaway winners that are no brainers. You then have the companies like Palantir and even Anduriil in the defense space who have been building these kind of data informed approaches over a very long time. These are not new companies. been around for over a decade, and they've been able to really get close to the customer and start to win some really, really big contracts purely on merit. So I think there's obvious winners. There's maybe non-consensus winners that are breaking through.

The different way that I would look at that question as well, though, who's going to lose? Who has the most to lose in this battle? Because I think it's very easy to pick the big obvious winners. And I don't have a crystal ball. I'm sure there's going to be many others that kind of develop some cool new technologies. I think the lens that I would use to view maybe who's losing or who can lose in this battle is not just about data capture. It's about data processing. There's a lot of satellite companies that have proprietary data sources. They can see a lot of things and certainly there's more and more of them going up every day. Just as an anecdote there, a fellow CEO who runs a launch startup, they're launching stuff into space, they got a $6,000 per kilogram price quote from SpaceX and a $60,000 per kilogram quote from the European Space Agency.

So that's a 10x, obviously, reduction in cost. The fact is SpaceX can probably subsidize their own launch to basically approaching zero when you look at some of these new things. I think just having the observation data isn't maybe the winner we thought it would definitely be in the long run. And I think there's a lot of companies that may be spacked a couple of years ago that are still working out how to really drive revenues in a significant way. So I do think that processing the data and getting close to the customer and really solving customer problems is a massively important part of the process.

Also being able to understand all these different data sources. Maybe it's an Excel spreadsheet, maybe it's the customer's internal data in their own data lake, maybe it's these Earth observation satellites, all of these other data sources. Having the smarts to process all of those things is really important. And we can get into some of the specifics on that later. But I think those that capture the data are at risk when they're not going downstream, I think just tech capacity in terms of processing the data at wild scale. There's still kind of people trying to work out how to do that in terms of lowering the costs and being cost effective in the way they process the data. And yeah, I think ultimately it's going to be the teams that can harness some of these new technologies, understand all of these complex relationships and get really close to the customer. I think maybe if I did have to look into the crystal ball there, I think we'll probably have some intelligent consolidation of some of these businesses as well, where you do have some of these runaway winners in the public markets who've got, and they're starting to throw off some free cashflow. And I think there's some really, really interesting stories on the data processing side that will boost some of these companies a really, really profound way over the next one, two, three, four years, as there's kind of unique expertise in processing all of this data. Gopal, I'll throw it over to him now, but I'll steal his quote before I do.

You know, we don't have a data starvation problem. We have a data indigestion problem. You there's so much data, but how you use it and turn it into insights beyond, you know, kind of maybe sometimes what seems like a magician's trick with some of the LLMs where you go, wow, that's cool. But whose problems are you solving?

Chris Mailander (15:29)
Yeah. And I think that's a really important piece of this is to understand and Gopal, I want to get your opinion on this, which is, you know, one of the challenges is that many times the data sets look like a black box and how that digestion process works is we don't understand the cross correlations that are at work underneath it. A, B, you're consuming potentially very different kinds of data sets. And so how you get that to reconcile, integrate well, rationalize, and then be able to actually act on it in a way that avoids bias or inadequacies in the decision-making process downstream associated with it. So take me through that because what Josh is teeing up is super interesting in terms of, you know, and I think it's the right way to look at it is like who wins and who loses. And so one of the first things that he's identified is those who are creating data alone, but not getting further downstream in order to provide analytical insights on that data, probably are going to get left out or be ancillary players. A another group, another factor in who wins and who loses that Josh has identified that I think we ought to drill into a little bit more is these technologies are allowing the firms that enable it to get closer to the customer and thereby disintermediating a whole series of players potentially that otherwise are providing interpretations and risk analysis and reporting. And what does this all mean? And knitting it together, the consolidation, there's a consolidation of that workstream that potentially eliminates a whole series of players. And so this piece in the middle about how you consume the data and get it closer to the end user and their decision making processes is fascinating to me. So walk me through that, walk me through a couple of the challenges along the way, educate me on what the data sets are that you're consuming with SustGlobal as a geospatial AI company and how to make that actionable. What's that take?

Gopal Erinjippurath (17:25)
Firstly, thanks for having us here. And in the context of all the companies you mentioned, one thesis I am building is that there's no one winner. I think there'll be multiple winners because there are multiple paths to unlocking enterprise value based on the currently disparate data sources that live, or private data sources that, private data sets that live within internal on-prem, as well as internal warehouses, data warehouses across businesses. So in that context, I feel the unique, the thing that parented it very uniquely in the last first 10 years of its existence is having a deep appreciation of the deep value locked up within enterprise and within the defense and intelligence sectors, which is unharnessed in the absence of the right data infrastructure that plays on top of that data. So over the last decade, what they have built up is the ability to link data sets and ability to turn them into valuable information and analytics, not by serving up a standalone, not by serving up a standardized platform, but enabling humans to go into gated environments, like the defense and intelligence services of allied nations, and map the infrastructure into the ontology of how those organizations think. And through that, they've unlocked the combination of machine driven as well as human driven decision making. And that's kind of what's now manifested in what they're packaging as the AI platform, the AIP, what they've done with Gotham, what they're doing with Apollo. Each of those are scaffoldings around data transformation, data processing, and AI primitives that work within the gated environments of organizations that they serve. And now with the new partnerships that they are signing, there is a deeper linkage between what they do and what the LLMs from an Anthropic or an OpenAI do, what they do and what the sensors from an Anduril would produce. So that deeper linkage unlocks even more enterprise value because they've already mapped it to the ontology and the action landscape within a business and within the enterprises they serve. I would say the other companies that you mentioned are also very interesting. These are you know, Palantir, NVIDIA, Snowflake. These are all very inspiring examples of very new businesses that have been formed from existing legacy infrastructure that has been built up. NVIDIA, for example, is now the essential fuel for AI advancements. So over a decade of focused execution, they trained up and created a rich collection of software libraries and a vibrant community that is very familiar with those libraries that unlock the full power of GPUs for AI training. So regardless of their legacy being in gaming hardware, none of the other gaming hardware companies have had a chance like come up to speed with Nvidia, primarily because of this, the unlock across software libraries and the unlock across communities that are deeply familiar with that technology that enables you to program GPUs, leverage their full compute capability for performant AI model training. And in that sense, they are now competing with like the bigger players like the hyperscalers, the Amazons, the Microsofts, the Googles of the world.

Primarily because if the compute-native workflows are GPU-centric, then that could happen through their own infrastructure and to their own data centers. So the introduction of the DGX cloud for cloud native workflows and the Jetson GPUs for the more edge focused workflows is one way they're working to build out their presence beyond just being a chip company turning into more of a compute company. Now, translating to what you were describing with the questions you had around geospatial, one way we've been thinking about geospatial data and thinking about data sets in the enterprise is the abundance of data sets, as well as data observations that are getting collected across organizations is tremendous.

We're seeing an increasing volume of light data. And you're also seeing an increasing volume of applications that could potentially leverage this data if they had the right infrastructure. We look, one simple example is that we look a lot more at map-based content today than we as a community have looked at in the last 10 years. Case in point, we recently had the US elections, one of the biggest exercises in democratic franchise. It glued many of the folks to screens, to analysis at census track, county, zip code, as well as state level. that's just for the outcome of a democratic decision-making process. Across the enterprise, you're seeing similar kinds of analysis for so many different modalities where the environment meets human activity. And those applications are all targets for leveraging the abundance of geospatial data. But what is missing is the middleware. The middleware that aggregates, runs inference, and links geospatial data into the business operations, the business decision-making workflows across the enterprise. So that's one of the things we are seeing as a growing need. And our activities have been very centered around making harnessing the value of that middleware and building on the essential components to make that a rich, vibrant set of capabilities that unlock new value within the organizations that we serve.

Chris Mailander (24:10)
So, walk me through this landscape also in this context from this perspective, which is that you have the large language models, you have large vision models, you have vision models, you have in your space, geospatial AI. Create some segmentation for me so I can understand this from an enterprise AI perspective and understand also where is the state of development of geospatial AI. Give me a sense of where we are today and where it goes.

Gopal Erinjippurath (24:36)
To a great extent, to understand the potential of geospatial AI today and in the future, you kind of need to understand the challenges and opportunities of geospatial data. Most LLMs like ChatGPT are built on transformer-driven architectures. And in order to fuel them with data, you create these large corpora of textual content. Some examples are the Common Crawl database and the Colossal Clean Crawled Corpus (C4).  All LLMs use them in some form because they have aggregated textual data at web scale and brought them into a formulation that can be or into a shape that can be fed into LLMs. I would say in geospatial, that is still in its early days. There are a few early efforts where there is the aggregation of geospatial data that can be then fed into large models that understand not just the textual content, but large models that can understand visual content and understands spatial context. So I believe that that's kind of like where the challenges and the opportunities start.

So maybe it makes sense to define what is geospatial AI. So broadly, geospatial AI refers to the category of AI applications, modeling techniques, and analytical approaches linked to geospatial data, i.e. data that is present with spatial context. And when you apply AI techniques to this form of data with spatial context, you have the ability to analyze, understand, and run inference on geospatial data. So that's our framing around geospatial AI. At the heart of all these geospatial AI techniques are networks that encode the spatial context which is the secret sauce, the real value unlock, encode the spatial context into information rich, lower dimensional, numeric representations or numeric vectors. And that's the equivalent in the linguistic world or the LLM world, which is called embeddings to visual embeddings. And that's in the geospatial land. Those visual embeddings run from vision models, turn into spatial embeddings where there is a spatial attribute and spatial understanding to the numeric representations. So we walk through the challenges with geospatial data. I would say the big opportunity is you have data sets coming from satellites, like Josh described, drones that are dominant, sensors and human activity, sensors that are tracking mobility across communities, across regions, across countries, and harmonizing that into a form that can be fed into these large vision models that can process range and reintroduce spatial data. So that's one of the things that is the first step in that unlock. The second step is enabling the training and building the machine level understanding of human level spatial context. And that's largely what is packed into vision models that run for geospatial data. And then the last, the third component, which I believe is kind of the most linked to the Palantir world, is say you train the model on public data on some commercial data sets. How do you make it work in the operating environment of a business?

For that, you've got to understand the business. So linking the spatial data and the spatial inference as attributes into objects, into human action, into object linkages across the enterprise is where that value gets unlocked. So those three components, building the data transformations, enabling the running of inference coupled with the ability to link that into business operations is kind of where geospatially I can have a massive impact as this middle value sitting between data and apps.

Chris Mailander (29:24)
Josh, so one of the things that Gopal just mentioned is that the state of the geospatial AI market probably is behind some of the other subsegments within the larger space, within the larger umbrella. And yet as he walks me through what you are doing to transform satellite data, drone data, other sorts of spatial representations into the ontology, the methodology that can be integrated with other sets of data, that strikes me as tremendously valuable, right? And when I tee this up and talk about in the same way that Elon Musk or that Alex Karp is talking about this battlescape and this competition that is out there, what we're talking about is one of these potential areas of new technological development and techniques, which has significant value to shift the battlefield. I mean, when we talked about Palantir, you mentioned 10 years ago, where were they? Now they're, you know, if you look at the rankings of defense contractors, they have completely transformed that competitive space. Now they're going even bigger in order to compete with other technology companies across the board and disintermediate non-technology companies that are providing a lot of the smart work, the consulting, the advisory type work, et cetera, associated with it. My question for you is this, which is, when we're talking about geospatial AI, it sounds like a market which is in an earlier stage of development, is it the sort of market competency capability which has the potential to truly transform some of larger players are doing once they're able to integrate it into their overall ontology?

Josh Gilbert (31:04)
I think that the really exciting thing to me when I look at the potential of geospatial AI is the way that it can bring all of these previously these data sets that were previously seen as externalities to people's models, to people's processes, or even just people's worldviews, or kind of frames of reference, or the ways that we build our own mental models of the world. There are so many industries that are touched by the real world, but the real world is not baked into the assumptions that many of these folks make in their day-to-day. And certainly we've seen that with some of our customers bringing, you know, a lot of the time for us it's climactic changes, understanding how over the next five, 10, 15 years, the world is going to change and ultimately how that will impact on areas of the business that, you know, they wouldn't necessarily have bought these ideas into their frame of reference.

My studies I trained as an economist, know, and I was always interested in the idea of economists love to talk about these ideas in the economy as if it's physics, know, if it's Newtonian physics, you know, they talk about general equilibrium, they talk about all of these rules. And they're all none of them are rules, you know, none of them are real Newtonian physics, you know, the apple doesn't fall and land on your head. These are all human made mental constructs.

And think the really exciting thing here is bringing real world events into these kind of mental models and kind of abstractions that many of these folks have. So I think that's the true transformative power of geospatial inference and geospatial AI is to be able to look at all of these very, very disparate data sets. You you have observations of the world as it is today. That can be a satellite looking down from above at wildfires. It can be demographic changes. It can be the movements of people, urbanization, deforestation, all of these things that are happening. And then you can have models of how we think these things may change in the future. You can then have tabular data, so the Excel spreadsheets and all of these data sets that exist. And then, you know, the proprietary data that customers have, financial models that they have, and all of these different domains that are going to be massively transformed by the power of simply expanding the aperture of some of these enterprises. You know, where we're saying, hey, we can be less and less reductive in our assumptions because we are able to have more and more confidence in bringing these multimodal and incredibly complex data sets into our regular processes. A financial analyst sitting in a Wall Street investment firm is never really going to be... they have no interest in becoming a geospatial AI wizard, getting into the transformer level and doing things. When we have interest, they certainly don't have the time. These people are working 80-hour weeks, 100-hour weeks. They're getting absolutely crushed. They would love to bring in new data sets that give them an edge.

It's just a different world. know, it's just not something that's possible today.

Chris Mailander (34:48)
What you're saying is that you if I were to think about it in terms of layers, bringing in geospatial data, satellite drone, whatever it might be, visual data, together with financial data, with climatic data, with socio-economic data and, and other sorts of proprietary data that might be held by a particular client and being able to create the cross correlations or the inferences between those, which actually has meaningful value and integrates into the decision making of those and list all the way up to the CEO of, and I want you to talk a little bit about some of your experiences as a technology company with the London Stock Exchange, the New York Stock Exchange, with a wind farm in the North Sea, with one of the largest pension funds that's in the world.

Talk to me about how they're using these new inferences as part of their workflows.

Josh Gilbert (35:41)
Most of these teams have very, very specific outcomes that they are trying to achieve, which are a million miles away from geospatial data problems, or maybe on the surface appear a million miles away from geospatial data problems. They are trying to understand how the cost of capital will change when they're looking at infrastructure investments. They're looking at their balance. They're looking at risk in a much broader sense. But one of the exciting things that we've seen is being able to index, harmonize, and ultimately transform these multimodal datasets into their world. So if we take the example of some of these large exchanges where they have a load of existing customers looking at fixed income analytics products that they serve to their customers. So one such example is CMBS, Commercial Mortgage-Backed Securities, where this is very well-established industry. is an industry where yield is obviously incredibly important. It's not like equities where, or investing where, your upside is uncapped and your downside is uncapped. We're talking the difference between a 5 % and a 15 % yield year on year over a 10-year period is pretty substantial. And in no way am I an expert in the in and outs of CMBS modeling and the prepayment rates and the probability of default that these folks do.  But one of the really, really interesting things is the impact that real world events have had and the correlation that we can see in the data where we take these pools of securitized mortgages and you look at the underlying mortgages in the CUSIP and you say, well, here's the actual geographic distribution of these securitized financial products, which are an abstraction. They're an abstraction, but at the root, there's real stuff. There's real real estate that really matters there. And then understanding, well, how does the real world impact on the value of these assets historically? And then how will it continue to impact on these assets in the future? So it's a really interesting one of the technical team at Sust Global, being able to, number one, index all of these different types of data. How do you map, for example, a physics-based climate model of the future to commercial, MBS, real estate holdings within a CUSIP. You have to get the CUSIP, you have to get the underlying assets. It's a non-trivial thing. You then have to basically create a like for like in these different data sets. You have to be able to index them in a way that the financial world meets the real world in a tangible way. So there's a massive amount of smarts that goes on in the background in terms of the product team at Sust Global and the technology team at Sust Global to harmonize these different data layers to make them simply be able to be standardized in terms of having a unit of comparison. And then you bring in some of the financial models and you bring in some of the proprietary smarts on the customer side. And all of a sudden you have this beautiful fusion of the real world and the financial world and ultimately unlocking insights where you can say, know, the data in terms of this specific example is something like a 3% change in the CPR, the pre-payment rate in, you know, high flood risk versus low flood risk individual properties that are then grouped up to securities. That 3% is pretty substantial over a 10-year period. Again, it depends upon the market dynamics, but we're talking like that could be on a $10 million investment over 10 years, that's a 500k delta. That's a pretty significant half a percent in terms of the yield. Just in this one instance, that's a really, really significant change. A crucial piece of information that end customers can get through expanding their aperture in terms of the types of data they can see, but still in a process and a form that is just native to them. They're not going out and modeling this stuff. It's just a number on a spreadsheet. there's real science and real world modeling that goes on in the background to transform that into insights.

Chris Mailander (39:48)
It strikes me that what you just described about the way that you deal with these diverse data sets and being able to do the indexing so that you can do cross indexing and cross correlations, the inferences that have meaning strikes a lot about what you said about with Palantir earlier on, which is creation of that ontology, which is customer specific to the data sets that they're utilizing. You create the ontology as part of this middleware, but then also are able to get closer to the client and integrate it into their workflows in a way that has meaning and value, but fits into their ordinary course of how they make decisions about CMBS. Is that fair?

Josh Gilbert (40:24)
Yeah, yeah. But I would add it's not just the CMBS example. The beautiful thing about the product that the team have built. And I think the thing that I'm most proud of is exactly the idea of a customer ontology. This isn't a Sust Global ontology. Each customer has their own ontology in a very Palantir-like way. And this is an incredibly complex set of geospatial processes that need to happen, which are all, not all, but hugely automated. And many, many moving parts of complexity in this indexing harmonization an auto link transformation of the data using proprietary AI smarts on the geospatial AI side, which Gopal mentioned earlier, really, really complex things to do. But then bringing that into the customer's world means it's not just CMBS. know, it's the chief risk officer at a hundred billion dollar plus AUM pension fund who's saying, hey, we need to understand emergent risks to our 3000 private assets across the world. And just being able to drill down to the asset level to be able to say, hey, each one of these locations has certain risk profiles. But also, what about the admin boundaries or the geospatial polygons that you can then kind of scale up or scale down that aperture to? So you can say, this oil and gas production plant in the Pacific Northwest or on the west coast of Canada, which is the case study we've got on the website, actually, wildfires are going to massively impact on the not just the production, but the transportation and the logistics and the supply chain, days of business interruption as well as actual real production risk. But at the same time, maybe they have assets that are public utilities in the US, and maybe they have assets abroad or they're looking at on or offshore wind. All of these different paradigms and being able to model the complexity of the real world and transform that into these views of risk where, again, these teams look at risk in a very different way to maybe some of our Earth system scientists or climate scientists at Sust Global or a CMBS investor in the previous example.

But each one of these cases has a massive amount of intelligent processing and inference in the background, which then enables us to get really up close with the customer and say, well, what are your actual problems? What are the transformational capabilities that we can bring to your world? And then it's a unique ontology for them, which creates a massively scalable product on the back end. I'll finish it. The example of the exchange that we mentioned earlier, so with London Stock Exchange, this ontology is not just in the CMBS offering, in the fixed income analytics offering at Yield book, which I described a minute ago. The same data and the same climate ontology is enabling the workspace product with commodities, metals and mining supply chains to understand how changes in the real world there are going to impact on the value of various commodities and various companies in the future. It's also enabling a report that came out recently, the Net Zero Atlas for COP29, which is where they're looking at 50 cities in the G20 and understanding, well, how is the geospatial view of risk across a city in different zip codes and different post codes and different admin boundaries going to change over time as well? So it's really about understanding, meeting the customer where they're at through a geospatial AI enabled process, which is massively complex.

Cris Mailander (43:57)
Yeah, interesting. Gopal. It strikes me, as Josh is talking this through, with being able to take a number of these diverse data sets, being able to create that ontology and the indexing that goes associated along with that. The amount of inferences is increasing exponentially as part of that process. Talk to me about that as a, you know, there's probably two sides of the coin here or several sides of the coin, which is one is through exponential increasing of the data sets and the inferences that can be made, you're increasing analytical insights. So that creates more value. Whoever's able to use that creates more competitive strength, has advantage in the game that's being played out. The other side of the coin is the computing power is requirements are significant, the electrical components are significant, how you architect this seems to be extremely important so that you can do the scaling, but do it on a cost effective basis. Right. Is that fair?

Gopal Erinjippurath (45:00)
I would say for like a minute, if you think of inference as AI driven, statistically driven and decision driven, then it unlocks the potential for us to do it efficiently. So just to doubt into the last question that you were discussing with Josh. Across the spaces we operate in, we're beginning to see an early majority of teams that appreciate the fact that there are these three dimensionalities of change. So there is change from the human dimension, which is population dynamics, income dynamics, employment dynamics around human activity.

There's the dimension of environmental change, which is climate, which is environmental signals, which is around land related changes. And then the third dimension is the dimension of change around the decision making. So the way an investor operated 10 years ago has probably evolved in the last 10 years to something different today.

So being able to capture those three dimensions of change, the human activity, the environmental activity coupled with the decision-making process, when we map ontological primitives to those three dimensions, it enables us to evaluate how to leverage compute in the most efficient way. So if you have vision of connecting it back to vision and AI. If you have vision models that are trained on environmental, climate, and human activity, and the thing that's changing is decision-making methodologies, then you don't need to retrain the model. You can use the same modeling outcomes, but frame it to statistical layers on top of the modeling methodology outcomes, on top of the AI inference. In cases where there is slower moving data, in the case of climate, you have climate data, that doesn't evolve the same way as weather. So we're dealing with climactic data and your inferences on climactic data. You could run the inference once, store it in an intelligent way for fast retrieval, efficient retrieval, and then connect that into the same statistical processes. So by mapping what an organization is trying to do through a combination of these three dimensions of change, the environment, the human activity is coupled with the decision making. We can then map out what is the most efficient way to leverage the compute rather than having to run complex, expensive inference each time.

Chris Mailander (47:53)
Right. That's an interesting concept also of fast moving data and slow moving data and being able to parse that respectively so that you're using the computational power efficiently. I don't need to draw on this as frequently as much. How do I integrate that into the overall process?

Gopal Erinjippurath (48:09)
Right. And this is a very, very exciting space because if you think about how data has been collected in the past hundred years, human activity related data has been collected through census. Now, most countries do a census once every 10 years at best. And that data once aggregated is pretty static. Like for 10 years, it doesn't change. So in between that, so if you're looking for alpha in your decision making process, and you want to link to faster moving data sets, you want to link to human activity at a higher cadence than census derivations, you need to map alternate sources into your input. So if that is the desire at the decision-making level, then you want to map and bring in the right data sources. And that's where you have the arbitrage. You have the unique opportunity to play with the lever, which is the delta in information you can generate depending on the cost that you're willing to incur for acquiring that information.

So case in point, there are alternate sources of human activity, like nightlights is one, traffic is another, density is another, credit card information is another. So all those data sources have some kind of spatial attributes that can be unearthed to give you human activity at higher spatial fidelity at higher temporal cadence than a census can.

So the moment you get into that paradigm, it's more expensive data because you're seeking differential information. But that differential information can be what unlocks new value in the enterprise in terms of the decision that someone's trying to make using the geospatial data.

Chris Mailander (49:58)
Interesting. So Josh, what does this market look like in three years time? We've got a sense now of the state of its current development, what you're working on in order to advance that case, the experiences that you've had, et cetera. What does it look like three years out as this trajectory continues to unfold?

Josh Gilbert (50:16)
Maybe to understand where it'll be three years out, I want to start by going 10 years back. This idea of a geospatial inference engine and a way to understand the complexity of the real world, climate changing, movements of people, urbanization. The exciting thing is it's almost like understanding the butterfly effect and these complex systems that can get codified and quantified in a way that was previously impossible.So, in the way that's previously impossible, the way that you can make the impossible things possible is by looking in the past instead of trying to predict the future. So 10 years ago, a very pertinent and relevant news story today is Syria and also the Arab Spring and understanding the events that happened in Egypt and Syria, I think, in 2013. The idea that there was this uprising, and many researchers have looked at this, but you can start to see all of the complex underlying trends that kind of accumulated in this big event. There were droughts and food shortages for three to four years in the local area in the buildup, which then drove a massive amount of urbanization. So had a load of people that were living outside of the cities that came into the city. So there was an influx of different cultures, different perspectives, as well as a food shortage. There was also in the same year in 2013, there was a wheat shortage, again, due to heat waves in other areas of the world. So, there was then I believe it was bread subsidies that were then introduced for the poor people, many of them recently urbanized, you know, immigrating into the city centers, which then kind of increased the tension until it reached boiling point. So there's all of these. And then, of course, there are the kind of geopolitical and kind of political factors that were also accumulated. it was one of very first kind of social media kind of moments on record. And so there's all of these different complex inputs that historians can then look and say, well, this was an interesting aspect, and that was an interesting aspect. The really powerful thing with geospatial inference when we look forwards to the next three years is being able to understand some of these inferential relationships between data points that weren't possible before. Much of our work at Sust Global over the past three or four years has been focused on understanding climactic change, building high resolution, higher performance models of how climate conditions will change based upon different scenarios over a five, 10, 15, 20, 30 year time horizon.

And that obviously in recent years, many teams cropped up around the ESG and TCFD, which is a type of regulatory reporting. it's simply not the huge opportunity that we see and I see over the next three years. I think regulatory push is never as compelling as market pull. And I think that the true power of a geospatial inference engine, which can understand some of these climactic conditions and other spatial data sets and demographic change and all these other areas where you can start to see inference between these disparate multimodal data sets and start to understand patterns, is actually something that should sit on the desk of every serious hedge fund trader to be able to understand how these complex patterns happen. And of course, it would be a tool in the toolbox. It's not going to be the oracle that they rely on for everything.

But all of these different players in the world, should be governments should be looking at this data to understand how geospatial AI and geospatial inference can enable them to have new insights in understanding what is, quite frankly, a pretty scary geopolitical dynamic in the world today.

It's certainly an exciting time to be alive. I look at these technology advances. Our company couldn't exist without the advances in cloud compute in the emergence of satellite observational data sets at scale and the cost going down and the compute going up and this raw horsepower that can then power some of these intelligent geospatial processes. But it's just the most important thing in the world today. And we've seen again with the focus on defense as an area and national security as an area and energy security as an area, that understanding how the real world is changing is going to impact on every single enterprise and maybe in the 90s and in the 2000s, it was the idea that Google, do no wrong, Apple that was building these products and then maybe even Facebook slash Meta, they were in the kind of social media world and in the data world and there was this huge expansion of value. I strongly believe that not just in the next three years, there's going to be tremendous value in the next three years, but the next five, 10, 15 years, it's going to be this approach of understanding the real world, understanding things that are happening across the world and how they intersect and the inference between those things. That's just a simply transformational business opportunity. And we're talking a multi-trillion dollar industry because this technology will touch on every single business across the world. And I see it as a perfect complement to many of the companies leading in LLMs, leading in AI, leading in cloud compute today, leading in these data platforms. You look at Snowflake marketplace, you look at Google Marketplace, all of these providers have a huge amount of spatial data already in their platforms. But it's the inference between these data sets and the geospatial AI engine that can power insights that's going to be truly transformational.

Chris Mailander (56:07)
Yeah, it's fascinating. When you talk about opening up trends, dollars of new market opportunities by impacting those decisions in a new way, I wasn't familiar with that story. And I immediately find myself kind of reassessing like the Arab Spring that you talked about in Egypt and the preceding conditions being a change in the physical environment, which is food shortages, which led to subsidies and then political unrest and what unfolds from there, as opposed to thinking about political memes and the use of social media at the time was the paradigm of the conversation. And yet you're surfacing a whole series of physical data, physics-related information that really is probably the underlying impetus for stirring a lot of that momentum up. And I think this is also interesting as well, which is transforming from regulatory compliance, that these are things that we need to study and look at from a regulatory compliance perspective to instead take the information and purposely put it into the workflows of those who are out making investment decisions and operational decisions and changing how people flow and move and use their resources and exist in the world. Making it available as part of their decision-making processes day in and day out will be far more influential to how we exist and live and grow peacefully here on Earth, it strikes me. It seems like a phenomenal opportunity. So when we talk about geospatial data, there are tremendous repositories out there. You mentioned that the major providers have libraries, legions of libraries of data that is set there. And it's really about creating the inferences associated with it. It strikes me that the players I would be most concerned about are ones that are sitting on a huge asset base, a satellite or data companies that have a huge asset base of imagery or data, and they sell it, purvey it to others, but they're not able to do anything with it. That strikes me as one. The others are folks that are offering pipes or platforms that enable a customer to do something, but they're not close enough in creating a customized ontology or a customized workflow that knits into an end user's use case.

That's a lot of different companies that are out there that potentially could miss it, could miss this game. And my biggest fear is that, you I've been in the technology realm for 30 years and the pace of change has always been fast. And that was at the late 1990s and early 2000s with ERP systems coming out. And the great game was really who would win and who would lose amongst ERPs being introduced into that realm. It changed then again in the 2010s and 2020s with new technologies coming out. We are now at a pace of change that rapidly exceeds anything that I've seen over the last 30 years. And I think that the player set will also change quite rapidly. Give me a little bit more depth about what it takes to win in this new environment that's changing so quickly.

Josh Gilbert (59:27)
I mean, the one interesting thing with this moment is it is by far the most incredible moment of change across industries. And that's very scary, but there's also a lot of opportunity in that. You know, I remember as an anecdote, this was in the year 2000, or maybe even 2001, it was pre September, pre September 11. I was a kid, I was, I don't know, like 13. I remember literally thinking to myself, nothing ever really happens. You know, I was just like, obviously, when you're an 11 year old kid, you're going to school every day, not a lot really happens. Your world is small. But I think since that year, there has just been constantly change and I think more rapid change than ever before. There's more risks, there's more opportunities. I really do think that many of these ERP companies, many of these geospatial 1.0 or GIS companies, some of the incumbent data companies, there is a lot of risk. This huge transformation in the way that we process data, the efficiency in processing data, the insights that we can derive through inference within the data, there's definitely a risk. I think moving into that, getting capabilities organically or inorganically in the data processing and multimodal processing area is super critical. So, satellite companies moving downstream into analytics and having a credible platform that can transform their data with other different types of data because just having a powerful single source of proprietary data is an edge. I'm not saying that it's not. And there are some companies that will continue to thrive.

But to really capture transformative value in the market, you've got to compose that data with other data sets. So think the processing part is important and very much gettable. This is not a war that's been won yet. There are a lot of companies here that can make plays, either with the existing capabilities they have or in partnerships, strategic engagement, investments, acquisitions. They can begin to get value in those areas. But then secondly, it's what Gopal described earlier, it's building these ontologies for customers, understanding how this data can transform each company's world. Because it's not just risk. There's opportunity as well. Another example of the work that Sust Global do is in wind energy infrastructure investing. And understanding how wind patterns are going to change, understanding how that will impact on power generation curves, different types of wind turbines and different production capacities, different costs of capital for all of these different companies, understanding how well in some areas wind speeds are going to go down by 5%, in other areas wind speeds are going to go up by 5%. And we're talking in the next five or 10 years. That 5% change can lead to a 20% annual impact in terms of the production of one of these sites. So obviously, have 20% up, 20% down. That's a 40% differential between the winners and losers in that specific market. And of course, that's going to be tremendously impactful on the IRR of these companies. It's another area where most of these folks, they're super quantitative, they're super locked in in terms of understanding the quantitative world, but they need access to these new data sources. They need to find a way to be able to process and harmonize this data, and they need to have a way to have this data integrated into their workflows. And I actually think that's a tremendous opportunity for many of these ERP providers, many of these cloud data marketplace providers, even the other companies that we've mentioned earlier in this conversation, to be able to apply this data and apply these ontologies in huge, huge markets, which are going to continue to grow, to be able to ride the wave and bring what I see is going to be the same level of transformative change in geospatial AI as we've seen in large language models and the AI boom that everybody would know over the last two years.

Chris Mailander (01:03:40)
It's interesting that you, when you were 13, you thought things never change and now you're in an environment and working where things change extraordinarily fast and in dramatic ways. And I agree with you that I think it's a very dynamic environment and exciting to be around. And there's going to be people who perceive this with fear because it is tremendous change and their stakeholder interest as well as just uncertainty and human reactions. But it creates significant opportunity for those who can navigate the change and compete.

Josh Gilbert (01:04:13)
By the way, Chris, the compounding impact of making those changes now is the really important thing as well, I think, for these businesses. Every moment you get away from T0 in this kind of transformative moment. And it is, you know, the compute power, the AI power, the costs and the efficiency drivers that we've seen and we're going to continue to see each year that these companies delay. The compounding effect of moving now is going to be tremendous over the next three, five, ten years.

Chris Mailander (01:04:45)
So it's now is that linear journey. It's compounding. It's exponential. And if you miss the window, you're out.

Gopal Erinjippurath (01:04:51)
Yeah. And Chris, in the great game of business, pervasive across capitalism, the thing I feel would be different for all these organizations who are collecting data is the ones who win are the ones who have the pulse on the activities and the things people are trying to do with that. So, my personal belief is however transformative and proprietary a data source is, it is transient. So, the only time it can preserve staying power is when they link to activity and then those same businesses are able to either through partnerships or by doing it on their own or through other forms of engagement like Palantir, which is like directly embedding yourself into organizational activity and mapping it through humans and then bringing in the machines.

Doing those things are the only way to stay at the top of your game because otherwise you quickly get disconnected because the landscape is evolving so quickly. So what was a pattern of usage two years ago might be different from a pattern of usage today. It comes down to the tools we use every day. So the way we use tools for search aggregation and simplifying information today is different from what we did five years ago. And that piece of transformation is going to continue. And if I can, I wanted to just share a anecdote or a quick thought around the three years out vision. See, oftentimes being a technologist and having worked in tech for like 20 years, my mind raises to what is science fiction today that can be real in the next three to five years.

And what seems very much science fiction from like two years ago that is already getting real is what people are doing with existing data sets, linking them through linguistic as well as language interfaces. So case in point across the enterprise, the movement Excel and copilot come together you're having people who are not Excel experts do things with data that resides in Excel. So if we map that paradigm out to our world and the spaces we operate in, I see the beginnings of agentic workflows where geospatial data is implicit. You're no longer saying, you know what, this is geospatial data. This is in raster vector format. It is implicit. And through enough usage of new intelligent workflows, you establish the patterns and buy the data together. And then it's no longer an engineer or data scientist or a machine learning engineer from Sust Global going into the $100 billion pension fund or going into the London Stock Exchange. That gets encapsulated in software where the agent does the thinking and the linking and building the understanding from the data sets it has access to. So that's the brave new world in which being at the forefront, like what we are doing with Geospatial AI, connecting them into workflows which are inherent and native within organizations takes a leap forward. Because at that point, there's no human in the loop, it's the agent in the loop. And at that point, you're not programming the workflow, just a copilot is linking into Excel. You have the geospatial AI agent linking into all the geospatial data sets which are proprietary into the multimodal data sets, be it tabular, be it unstructured data or semi-structured data that's within the enterprises on-prem or within the bounded context of the enterprises data warehouse linked with public data sets and unearth the meaningful insights to pre-programmed workflows or on the fly programmatically generated workflows.

Chris Mailander (01:09:21)
I'm interested, entrepreneurs always fascinate me because they leave structured worlds. They take on tremendous amount of risk. They're in pursuit of a vision. There's something that they want to accomplish or something that they want to build. Many of them express all of those dynamics, which sound very admirable. And yet it's a fear inducing journey. We talked about fear and how that affects things. I'd love to know about your journey since you started up Sust Global four or five years ago and the journey that you've been on and you both have a very interesting heritage and how you came together. Tell me about that journey.

Josh Gilbert (01:10:00)
We both were attracted to the hard problems and exciting opportunities at the intersection of kind of everything we'd done before. So I was at the CleanTech Group, which is a research and consulting company in the climate space. So I was working with large financial institutions, governments, NGOs, doing research on emergent technologies. And it led me to work at a geospatial AI company called Orbital Insight in Mountain View. did 12 months with them, working with the CEO and the chief marketing officer to help launch a new product and was constantly saying to them, look, the intersection of changes in the real world. And by the way, Orbital Insight were pioneering things like the footfall data that we mentioned earlier. So looking at data in terms of footfall and saying, can we get a signal ahead of quarterly earnings releases in terms of if Sears has sold more spades than they did last quarter, or whatever it might be. But looking at the real world in a much bigger way, looking at the climate, looking at the built environment, looking at the real world environment there. Gopal was working on many of these problems at Planet Labs. And I'll let him give a little bit more of his view and his experiences there. But we met on a panel, and it was on something about AI in an era of data abundance, I think, was the name of the session. And we met and I went over to meet with Gopal. We started talking, we started writing some research pieces together. And interestingly, the foundational piece that we wrote five years ago holds very true today. It's kind of approaching geospatial 2.0, this unlocking billions across industries at scale. That was the title of the piece. It's out there still.

I don't know, Gopal, if you've looked at it recently, I was pretty pleased with the piece. know, I got a lot of traction in that industry. And yeah, I think ultimately we both saw this opportunity. And I guess from the entrepreneurial side, you know, I realize the one thing I've learned is there's that reptilian part in our brains that tells us if we fail, we die. You know, like if we were out hunting woolly mammoths and we fail, we die.

That's a very useful part of our brain in those circumstances. But in this world, it's like we have the opportunity to do impactful things. And the more you can get comfortable with the idea that it's not a failure in that extreme sense. And we've been tremendously good and tremendously fortunate over the past five years as a company to drive success. It's getting over that fear of failing because you think that it has some tremendously awful outcomes and you actually then are pulled towards this future. As Gopal said, what science fiction today that's not going to be in three years time? And the best thing about being at Sust Global, I think at the heart of my value system, which drives me to do it is we're in the process of bringing that science fiction into life, removing the fiction from science fiction.

Chris Mailander (01:13:12)
Awesome.

Gopal, you grew up in Oman, which is a fascinating place as well. And then school in India and then also at Southern Cal. Tell me about the journey. Tell me what motivates you as an entrepreneur and pick up where Josh left off when you guys met five years ago and sort of building this.

Gopal Erinjippurath (01:13:29)
So, yeah, I've been very fortunate to live in those three places that you mentioned, Chris, early in my life. Southern California, Kerala and Muscat in Oman, rich countries with heritage as well as a lot of beautiful palm trees. So, the thing I did before the five-year journey where I met Josh was working with very high-volume, data-dense environments in multimedia. That's what I spent the early years of my career doing. And then got into AI and computer vision and the application of these methodologies, this thinking, new thinking to environmental data sets a little over 10 years ago. And when I was at Planet, I got to stand up the team, the product and the business that transformed, going back to what we just discussed earlier, transformed from being a data company to serving the analytics and meeting customers and the operating spaces where they're at in their levels of maturity. And that was a very enlightening, exciting journey for me, building those multi-million dollar businesses, plant analytics. And the thing that I learned in that environment was that there is a tipping point happening at the moment. The amount of data collection is unprecedented. The volume of data accumulation is unprecedented. And the only way to unearth that data is through to mean aware automation and intelligence. And that's kind of largely what we brought into the spaces with geospatial AI at Sust Global, where we operate. In that process, one of the things I have learned is many of the times, and the thing that is common across businesses today is everyone feels they're under-resourced. You ask a big company, a Fortune 500 company, you ask a mid-size company that is growing pre-public, or you ask a startup, all of them are going to tell you that they are under-resourced. They don't have enough people. But the genesis and the unravel in entrepreneurship is when you shift from that lack of resources into being resourceful, being able to acquire the right talent with the right balance and be resourceful with that talent. And the way we have been able to do that is by what I like to call the extreme Pareto optimal. Where can we go into spaces? What can we be doing where five to 10% of our additional effort can unlock 90 to 95% of business value in the space with the problem set that the customer is facing at the moment.

And that requires a slightly different mindset shift compared to traditional SaaS because traditional SaaS is about having the right product and having millions of users with a standardized workflow, like a pattern. And the reality is, if you're in an early market and you're serving an early adopter, that pattern doesn't exist. So there is no one perfect product. There are solutions that sell. And that's where it's possible to find the extreme Pareto optimal solution by being resourceful and connecting into domain knowledge that exists within a team. So that's kind of been our journey in terms of if I think about the pivotal dimensions that we've been able to unravel, it is about finding the right problem spaces which are ripe enough for the disruption and the innovation that we can bring in, identifying that, just talking about AI doesn't solve the problem. You kind of need the scaffolding around it, the glue around the model to make it worthwhile and operational within a customer environment, and being extremely judicious with the limited resources and being resourceful in terms of tapping into the partner ecosystem, into the customer base, into our existing networks to capture the value.

Chris Mailander (01:18:04)
It's one of the things that gets me excited about this period that we're in. And part of the reason that I've always enjoyed working in the technology space is because you're creating asymmetric advantage that with good technology, new technologies, cost efficiency, and clever solutions and good thinking, you can outpace your competition is David versus Goliath type stuff. And that's what creates excitement and value. It's how players that if we look at any one of these market segments that we've talked about today, 10 years ago or five years ago and the scaffolding of the players in those competitive sets. If we were looking at the battlescape as it existed then versus today versus what it will be, vast amounts of change. Sometimes we don't appreciate how much change you can create. And then we look back and then do the decoding to try and understand how and why. But I really appreciate this conversation today, which allows us to look at where we are right now, what the factors are for winning, the factors for losing, who's at risk of being left out of the great game, who's going to consolidate power and continue to pass through that window and then have the exponential increases associated with it.

It's really a dynamic time and exciting for each of you. What is your trophy? What is it three years from now that you want?

Josh Gilbert (01:19:26)
Our trophy, I think, is that we've developed something truly unique in the geospatial AI domain, which allows us to consume really complex data sets, cross correlated across tabular data, spatial data, real world modeling data sets. Our trophy is affiliated with one of the players that's going to win in the war that you've described. The industry is moving really, really fast. I think there's many, many exciting opportunities to partner, but kind of each time you do that, you're going deeper with that specific player. We think we have something truly special. I think that geospatial AI, the way I always describe it, is necessary but not sufficient to transformatively impact the world in the way that we want to. You know, the trophy for us is simply to create as much impact with the work we do every day. We have built something incredible. But we want to get that into the hands of these companies with massive distribution, who can really move the needle. We've been able to help in a very short space of time with a very lean and mean startup team, some of the world's biggest pension funds, two of the world's biggest financial data platforms and indices, countless large consulting firms, asset managers, investors across the world. We want to do more and more of that. And we want to continue to transform industries, but an even bigger scale. That's why we're in this game. That's what the trophy is.

Chris Mailander (01:21:02)
Well, an exciting time to be alive. Good luck to both of you.