Armada raises $200M+ to build portable AI data centers for the 70% of the world without AI infrastructure

more in the coming days. But we have our first guest of the show, Dan from Armada. Welcome to the stream. How are you doing, Dan? Hey guys, I'm doing great. How are you? I'm good. Welcome to the show. Thanks so much for joining.

Can you kick us off with uh an introduction on Armada and how you're explaining the business these days? Yeah. So, Armada is building the hyperscaler for the edge. Look at a map of the world.

There's about 30% of the world where you have these big hypers scale data centers and I mean 70% of the the world doesn't have AI infrastructure today. And so, we are filling the gaps in infrastructure building that hyperscaler for the edge. Got it.

How much of that is military versus, you know, we hear about the oil and gas industry. They need a data center on site. They're in some remote location versus just energy price arbitrage. Let's go where the energy is cheap. Put a data center there.

Crusoe was doing that early on in a little bit less of a portable package, but how are you thinking about the the market and kind of the most the most tangible opportunities that you can explain the business through? Yeah, it's it's both.

So we're working with critical industries, a lot of energy companies, oil and gas, mining, um manufacturing, logistics, and then in the public sector, we're working with the DoD, uh as well as allies.

And then also doing a lot with emergency response, anything that requires sort of splitsecond decision- making, processing of large amounts of data in more remote locations. That's a real focus for us.

And then what we've seen is that now that the kind of blocker for AI is moving from just the chips to energy and I think Anthropic just came out with a white paper last week where they're talking about we need to unlock 50 gigawatts of energy by 2028 for AI and so uh we actually just launched um a new product which is called Leviathan which is megawatt scale AI data centers that can rapidly uh colllocate with areas where there's stranded land energy.

And you can have a a sort of a plug-in a plug-and-play AI factory in weeks versus years with a traditional data center. And that's important because we've got about six um gigawatts of stranded energy around the country. China's moving really fast. We need to unlock all that energy in order to win. Yeah.

Walk me through kind of the shape and scale of the products. I've seen the containers. I'm curious what what does stranded energy actually look like? I have a bunch of guesses. You know, you know, uh, infrastructure was overbuilt in a specific area.

Whales, if there's beached whales, you go, you extract the whale oil right there and then you can burn it and generate your power for your data center. No, just like that. Just like that. Just like that.

No, I mean there is a ton of uh stranded energy around the country and then also, you know, in allied uh parts of the world like the Middle East. If you if you looked at like the top energy sources, what's top of the list? It's natural gas, right?

So, you have a lot of stranded natural gas in places like, you know, North Dakota, uh, Texas. We're already working with developers that have a lot of that land and stranded energy. And, um, you know, they're looking to utilize it for AI infrastructure.

And if you look at um, you know, the the AI action plan that just came out, I think part of what we're trying to do is say, okay, we have all this drained energy. We're going to need hypers scale data centers, you know, sort of gigawatt scale data centers.

We need to build those really rapidly, but we're also going to need to build distributed uh infrastructure, distributed data centers that can be deployed very quickly to to use that stranded power.

So, uh, when I'm when I hear stranded natural gas, does that mean like some exploration company has gone and and done fracking experiments or research to understand that there's natural gas there, but maybe there's just no pipeline to extract it.

So, you can get it out of the ground, but it's very expensive to move somewhere else that's more valuable, or is it something else that's going on? Like, how can natural gas be stranded? basically.

Yeah, that's definitely, you know, some of the scenarios that, you know, other scenarios are like there are major energy companies around the country that have, you know, sort of flare gas that's just being released today in the atmosphere. And so that they're they're under pressure to do something with that gas.

And so what they're now doing is they're actually using turbines to convert it into electricity so that you can power infrastructure. Some for their own internal use, but then also they're looking at this as, hey, I can create a new revenue stream.

And so it's a mix of different situations, but but those are common ones that we see. Makes sense. Uh let's go back to the actual product, what it looks like. I've seen shipping container renderings. uh what are the different scales and products that you think are uh most important to push forward right now?

Yeah, so again we're we're building the hyperscaler for the edge and what's unique about us is we're doing it full stack. So do the hardware, the software and the AI and the easiest mental model to have is like think about one of the current cloud providers. Think about you know Azure AWS uh GCP what what are they?

They're they're you know really just uh full stack infrastructure companies. They focused on building these large hypers scale data centers in you know areas that had a lot of data a major metro areas but again that's only 30% of the world.

So basically what we're doing is focusing on the remaining 70% but otherwise it's a similar type of approach. We're building full stack. Uh we proide everything to customers as a service in a way that's totally turnkey.

Um, and I'd say that the biggest difference is that we we see a world where it's going to be distributed compute.

So, we're building rapidly deployable mobile infrastructure that can be colllocated with not only energy but data because what we've found is that you've got, you know, some of these energy companies that have oil rigs that are generating 1 to 5 terabytes of data every day.

similar situations with mining companies, you know, similar situations on a a battle ship, right? Where you might have just huge amounts of data being generated from drones, autonomous technology in the middle of the ocean.

Um, now you have connectivity via Starlink or Star Shield and other types of connectivity, but you also need the compute there locally to do any sort of real-time uh data processing at the edge.

Are you seeing more demand for kind of like tokens as a service managed models or just pure infrastructure bare metal let me do whatever I want on it?

It's all so again it kind of comes back to think about like the cloud provider like there's a reason why Amazon web services is called Amazon Web Services right it's it's the is the services that the people care about and so for us we also sell services or we call them solutions where it's like there's a specific uh thing that the customer is trying to do whether it is you know rapidly standing up an AI factory uh in some location location where they have stranded energy or it's a scenario where they have lots of um unstructured data, let's say from drones, and they want to process all that data in real time versus having hours or sometimes, you know, days of delay so that they can do a specific thing.

A good example is uh we're working with the the state of Alaska, their department of transportation, and they're using our entire platform. Um, they're managing all of their connected assets, meaning like Starlink terminals, uh, SkyO drones, SDWAN devices.

They used to have over a day of latency to process the unstructured data from the drones for response to avalanches in remote areas. Wow. And, you know, we're helping them get to near real time to respond to those types of things.

And so, um, that's that's what we're seeing is there's different specific problems that these customers have at the edge. And then we're unique in that we actually solve them full stack. Jordy, I'm curious. How many uh how many uh sort of like systems do you want to have?

Like what what are the sort of like more near-term goals in terms of deploying systems? When I think of when I think of the word Armada, you know, obviously you think of like fleet, you know, a fleet that's stretching as far as the I can see.

like the Tesla power battery pack where I can figure like one and it's a couple grand or I can spend $192 million or something if I buy a ton of them, but it's really just like a lattice. Yeah. Super. Yeah.

And even like h how you're managing uh kind of like how to prioritizing different customers when I'm sure you get calls all day long and somebody's like, "Hey, I need a data center in an Antarctica. " Uh, and you're like, well, like we're more focused on, you know, the northern hemisphere, right? The IMO's next week.

I need a data center. Yeah. Exactly. No, we're we're exactly right. We're focusing on the the largest sort of missional aligned entities in in the world. And what I mean by that is we're working with some of the largest energy players um in the world. We're working also with, you know, the DoD. It's public.

We're working with the US Navy. Uh it's kind of nice that the the company's called Armada. goes hand in hand with that. But um the the the way that we approach it, we really focus on customers that we can see that they will have thousands of what we call connected assets.

So, you know, take take the Navy, have a a a gallion, we call our modular data centers gallion, sticking with the Armada theme, have one on every ship. Mhm. Um you know have one in strategic locations around the world where we have bases or you maybe more of a mission scenario being able to support that.

Um similar situation if you think about like a really large uh energy company like we were in the news with a Ramco and Microsoft where we've already deployed our technology. We're starting with one site, but they have rigs and refineries all over the world, onshore, offshore.

And so the idea is that this is distributed infrastructure for all of their operations.

And what's really interesting is if you open up our operating system, we call it, we call our platform the Armada Edge platform, AE, you see a fleet map and it shows all of your gallions as well as any other connected assets that you have all in a single place.

And so what we want to be is that single partner for the most important organizations in the world, you know, obviously aligned with the the US um for all of their operations.

Can you talk to me about the the history and development of Starlink that feels like a foundational technology that unlocks a lot of what you're doing? Uh how big is Starlink? How useful is it? Where is it going? um where where do you expect the capability unlock points to be?

Should is there a framework where I should be thinking about Starlink in the same roll out as you know 3G, 4G, 5G uh something like that? I imagine that we're we're still in you know the exponential uh rollout phase of that technology. Yeah, absolutely. So Starlink only launched in public beta in November of 2020.

So it's a relatively new technology. Um the first couple of years they were focused on consumer business and then they started selling to enterprise and then also working with the government uh with Star Shield.

So you're really only talking about a few years and they're now in well over 140 countries all over the world. And what's been really interesting, we've been working with um Starlink from the beginning of Armada.

have been working with them since the inception of the company and we have a great partnership, but the technology is continuing to get better and better and better. When we first started working with them, a lot of people were using it in like pockets. Take like a really large energy company.

They might use it as as backup form of connectivity. Y um but they wouldn't use it as a primary. What's ended up happening is that everything that Starlink does is getting better really, really fast, including, you know, new generations of the birds, the satellites up in space, uh, which are getting better performance.

Um, there's, you know, new types of terminals. There's even, uh, you know, a new kind of like priority lane, they call it multiu where you can get better performance, a little more expensive.

So what's happened is a lot of these um organizations they started to use it as backup and now they're actually starting to use it as a primary source of connectivity.

And what we expect is over the next 5 years it's going to cover the entire globe and you're going to have you know LEO connectivity that's as good as having fiber in a lot of these locations. And so we're we're building for that world today and sort of completing the rest of the compute stack for for that world.

in in terms of fiber, I mean that seems like speed, but is is there a world where when you're you when you're in the data center space like you need the pipe to be extremely big and and we're we're hoping that you know the roll out of Starlink will enable bigger and bigger pipes. Is that the correct like mental model?

Yeah. What's what's interesting is that a lot of times um people will not want to send all the data back. So they they only want to send the meta metadata back. So what's very kind of like synergistic between what Armada does and what Starlink does is we can work with them.

The the performance, you know, will continue to get better and better uh from a latency standpoint. We will handle all of the, you know, local data processing and then we'll use Starlink to send the metadata back. Right. Yeah. And um there's really interesting things also as we're continuing to evolve what we do.

So we started out with um inference that was the primary focus was like okay you take an AI model that's been trained in the cloud you run it at the edge on our infrastructure.

Now with this Leviathan launch that we recently did, we've kind of completed the picture for a AI factory that can be deployed in weeks um versus you know years with a traditional data center and it's still mobile. So you can relocate it as needed.

And what is interesting there is that you can do very cool things with like federated learning.

People are going to want to take these models, sort of fine-tune them for different scenarios that and run them at the edge, but they're going to want to leverage the improvements in the model across all of their different sites.

And so the way that we're evolving, the way that Starlink's evolving, we think we're going to continue to unlock a lot of new high value use cases that way. Yeah. Can you walk me through some of the some of the like enterprise like infrastructure that you bring to a customer that's not out of the box with Starlink?

Because I know that SpaceX is a fantastic consumer product, but there's a certain level when you get into the enterprise workloads that you need another partner in the stack to actually deliver capability. Yeah, we're we're seeing that customers want to use multiple connected assets.

We call all these things connected assets and and it's you know LEO and geo satellite connectivity. So they they sometimes will use Starlink but they'll use Starlink plus like a Hughes or a via SATA geo satellite connectivity. They'll use SDWAN devices like Cradle Point and Peplink.

They'll use drones like we have partnership with Skyo on the drones. Oh wow. And so what these customers want is they want somebody to help bring it all together to solve their problems at the edge. And that's that's what we do.

Um and so a lot of times when we'll get involved um you know working on on opportunities with Starlink, they are already working with Starlink but then there's something else that they want to do in addition to that. Uh and we help we help that.

And then the other thing there's um you know different things that customers want just in terms of um optimizing for different scenarios based on not just performance but for cost. And so we help with a lot of those types of use cases. Yeah. Um what what's the scale of the business these days?

So we've been growing really fast. So we we've only been around for a little over two and a half years. Wow. We spent the first year in stealth. uh raised now well over $200 million from bunch of great investors like Congratulations.

Uh one of the the really uh exciting things that happened last year is that Microsoft ended up uh in making an investment in the company. They invested in this round and then I met with Satcha Nadella. Yeah.

Yeah, he got excited about what we're doing and so uh he ended up talking about us on stage at Ignite, their big annual customer conference and we're going to market really globally now with uh Microsoft and and and SpaceX among other partners. Um and we're working with some of the largest companies in the world.

I mentioned to Ramco, there's a bunch of other large oil and gas companies. We're working with some of the largest mining companies. Um, and then you know we're working with the Navy that's public and then with state.

So we're kind of just continuing to focus on ones where like we were talking about we can see fleets of connected assets like thousands and tens of thousands of connected assets being deployed all over the world and then you know scaling to become their single partner for the edge. Yeah.

Do you think uh the relationship with Microsoft looks more like in the future I go to Azure and I need to do a workload and leveraging Stranded Energy just gets me a better price or do you think it's more like a channel partner where a really big enterprise client goes to uh Microsoft and uh they can they can uh route to you as a specific solution for a very uh like you know I don't know like like off the off thebeaten track use case.

Yeah. I think what c we're already seeing is that customers want to optimize more and more workloads between the cloud and the edge and I I think that's going to continue like the cloud is always going to be there.

Um but more workloads are going to also be run at the edge and so the more those things work hand inand the better it is for the customer. Yeah. And so how how it's working today is most of our customers are already working with Microsoft. They have really large, you know, agreements. They have Azure credits.

They can actually buy all of our products under their agreements with Microsoft and using those credits. That's interesting. And then Microsoft sellers also get some credits. So it kind of all works together. Everybody Yeah.

Um, but what's nice for the the the customer is there's different capabilities that they already are using from Microsoft that have never been available in these more remote sites and we're making them available uh in these mo more remote sites and that can range from you know the latest AI model that's been trained in Azure that somebody wants to run locally at the edge y to you know things that are more basic like an example is in the gallions you can have what's called Azure stack which is a piece of hardware that Microsoft produces.

Oh, interesting. And a lot of applications that these companies like to use are designed to run and they already run on that, but that's never been available at the edge. So, we had a partnership that we announced with Microsoft and also Hallebertton. Yeah.

Where there's a a series of applications for like production automation that these large energy companies are already using in their more connected sites. Yep. We're just extending that to the more remote sites.

And I think that's so you don't have to completely replatform and go down to okay, I got to make sure it runs on Linux or it runs on CUDA. Like you can assume that you have access to all the nicities of Azure but at the edge. That's right. Exactly. That's awesome. Congratulations.

Uh this is a fascinating business and I really appreciate you taking the time uh hopping on. We love talking your travel schedule must be absolutely insane. I'm on a plane last few days. Yeah.

Next time you're out at some Stranded Energy, call in from the field from your phone, presuming you have Starlink or or at least cell service, and we'd love to see what it looks like to see some some natural gas plant flaring for for the last time before it becomes tokens. Well, let's do it.

I'll I'll put my Starlink Mini in the backpack and fantastic. Make it happen. Amazing. Thank you so much for hopping on team on all the progress. We will talk to you soon. Have a good one. We'll talk soon. Thanks again. Up next, we have Samir from the Colin and Samir YouTube channel.

Of course, second time on the show, we ran into him at the Coinbase base launch event. Very briefly chatted in the stairwell. We were moving past each other. Very excited to welcome him to the stream. Samir, how you doing? Looking professional. Looking extremely professional. I don't know if we have your audio yet.

It might be on our team side. Let's check. We we we are we are not getting audio, but um it doesn't matter because you look fantastic. Yeah, the the the I'm clearly hitting the gym. The Yeah. Ben, is this on our side or or he's muted? Okay, I think you're muted.

Let's try and throw that on and we'll bring him in and uh and continue the the Glazenator 3000. The white balance on that camera, phenomenal. The exposure to die for. The lighting very soft, very neutral. Sorry. Let's see. Uh can you hear us Samir? Welcome to the stream. Nothing yet. Let's work on that.

Um we will bring I mean what what a what a what a moment you know even even creators at the absolute top of the top of their field the apex content creators still still learning still grinding. Yes. Uh let's uh let's do one timeline post and then we will move back into Samir.

Uh this uh interesting data point to uh calculate the grind score of a company. Have you heard of this metric before? So uh FEGMA is in the top 15 companies in terms of grind score. Grind score is the percentage of employees that have a positive business outlook divided by the work life balance.

So you're working really really hard but you are extremely optimistic about the company and uh topping the charts are applied intuition who we talked to queser ununice we talked to him at Hillen Valley uh the boring company of course Elon company you're going to be working very hard but if you believe that you're going to drill tunnels all over the world you're probably pretty optimistic I had a uh I had a a roommate when I first moved to LA that worked a buddy of mine that worked at the Boring Company and he I thought I was working hard, but he would would often uh leave before I was up and come back after I went to bed.

It's it's an extreme extreme company. I've heard of stories of investors who were trying to get allocation in the company and uh a prerequisite was basically that you go do a full month of full-time work to to, you know, advance the company.

It's like, oh, oh, you think you're going to value you think you're going to add value? Yeah, come out here for a month. just be in the office. I'd love to see RAMP up there and uh Vanta and ideally, you know, we can get all of our our sponsors into the top 15. Shield AI. Uh yeah, it's a good crew.

Uh also, uh Goateier, I'm pronouncing it correctly finally six months into reading his posts on the stream. Uh I just vibe coded this. It works flawlessly. And it's a picture of a plane with an engine where the plane think this plane should exist. Would this fly? It might actually work.