Aaron Ginn's Hydra Host raises $100M Series A to become the software OS for global sovereign AI data centers

just everyone's sharing, you know, homegrown reels in that they're making in their own homes.

Yeah, why not? Imagine

that seems more healthy than I don't know. It seems like it solves the problem that they're going after if they think it's a problem they've been solving. Anyway, we have Aaron Gin from Hydra Host. He's the CEO and co-founder. You know him. Aaron, how you doing? Been too long. What's up?

Yeah, it's been a while, man. I missed David Cameron, so that that's what that's what I take away from that.

Yeah.

Uh Oh, really? Victorians are a mess, but Yeah.

Yeah. Yeah. Have you been grappling with the phones, fertility thing, the social media thing? Like, where do you stand on all that? I mean one most of the longitudinal study they actually can't normalize or control for most of the stuff versus like already predisposed behavior. Sure.

Uh and then like you already were existingly depressed and accelerates the depression similar like alcohol and other things like that.

Uh but but but the second is that this is going to go the way of the 18th amendment

because what Jordy said which is like but the world is flatter than it's ever been before just like the model stuff with anthropic.

Mhm. It's like it's like okay so we make a big deal about this but then six months later China already has it so like what's the point of this

and and and I think that that's like I I I like the intent but the intent should be like empowering parents to like educate their children and and empower them to control even a curfew feels more like even a curfew feels more enforceable or like enforced screen time which is like you get 30 minutes a day on social media

like that feels like easier for for the platforms to actually enforce then or the country to enforce than like a total ban.

Yeah. But but this this just goes the normal trajectory of how technology gets adopted. There's always a late stage of regulation and and for like GPUs, you know, like the the reason why the UK government is so aggressive on sovereign AI is because they want to do similar things the to like what this is. And and I I think that that trend is a little bit unescapable because in the UK you have the right to do that like they are a sovereign country. You can regulate how you want to regulate. Uh and then the latter component is that if if you think about the kind of how this whole space is flushed out like what has enabled our business uh and why we just uh you know closed a very large series A at $und00 million. $100 million. Yeah. Yeah. I know. Right. Right. So appreciate that. Uh, I I have a gong in mind whenever y'all had your uh successful application. I rang my little I ran my little gong for y'all. So, uh, yeah, of course. And the, uh,

did you did you prime it? Cuz we've been getting a lot of critiques online. People said you didn't prime the gong.

And I just thought you come in with the hardest possible hit.

No, I'll show you. I'll show you. You have to do five priming strikes and then the real strike. So, you go. Now it's

Now it's primed.

Okay.

And then now the reel hit like this.

That's how you prime the gong before you strike it. This is the way you do it. We learned this from the commenters who said we weren't warming it up.

Hey, I'm I am pro appropriation of my culture. So like like this this is this is always a pro for the

You brought you brought us hats. You brought us hats that I didn't even I didn't feel comfortable wearing because it they just

I thought he was talking about appropriating GPU whisperer culture whispering to the GPUs.

The the uh I mean literally my dad gave me this on my desk so like this holds my phone. Love it. So $100 million uh what is the distribution of that? I mean are you going to be buying a lot of GPUs? You're working with Nvidia. Is this like one data center multiple data centers? Are you part are you building like what restate the thesis of like where you fit in the bottleneck?

Yeah, the uh so no we're not using any of the money to buy GPUs.

Uh we are an asset like company. Yeah.

Uh I believe we hold the title as the only NeoCloud in Nvidia's orbit that is a software company.

So we we manage GPUs. So we provide an operating system for the data center

to where any data center in the world can become a neo cloud.

Sure. uh we're already in almost two dozen countries close to 60 data centers and we have uh let's say several billions uh in contracts that have been signed

uh for us to continue this project. So so that's why sovereign is a big uh customer of ours is you got to think that you know is is a US company really going to go try to build a data center and supply in England or are they just going to rent from somebody in England? Then you then, you know, add the layers of now we're on social media, but this is clearly going to be applying to GPUs.

Then you can see how this all kind of all flows out where data residency in nation states is going to be a permanent fixture going forward. And and then that's not me approving that or wanting that. Like it's just I just think that that's just where we are right now. And and then you then add the continuous layers to that of like well what is AI going to work on? healthcare, education, defense already all those are nationalized industries in these other countries. So in summary, public cloud is dead and Nvidia in there all the other accelerators are basically are applying a ton of free cash flow to that same argument. And so we're the software that enables all that to work together. How much of your thesis is like you want to stay Nvidia only, you want to compound the learnings around the Nvidia ecosystem, there's value like we talked to another uh a neocloud last week that's AMD superfocused, but then you talk to other folks and they're like we want to be completely agnostic. We want to be, you know, one big blob of fluid compute. There's multiple thesis uh that are playing out and people are sort of testing every idea. Um but how do you see it like what is your strategy?

Yeah. Yeah. So our our software works with with any OEM or or ODM. It's intentionally designed that way. But of course like the capacity in our data centers reflects exactly what the market demand is

and the video is just the best. And and yes I I do think that there is a place for uh not necessarily the hyperscaler chips. I think that that market is pretty narrow. I think it's probably five customers that actually would ever buy that. Um, but the rest of the uh kind of alternatives, yeah, they're they're all going to be designed for very specific use cases. But if you're in a market where power is going to continue to be constrained, data center be you're going to pick the most like you're basically always going to pick the Porsche Cayenne Turbo

over and over and over again. You want to put your kids in it. You want to put some stuff in the back and you want to go fast. You want to go you want to go fast or or or Mercedes uh uh station wagon, right? the AMG station wagon, right?

Yeah. Classic.

So that so that's Yeah. So that's the Nvidia play, right? Is that

I'm the best. I'm expensive, but I can do everything you wanted me to do. And AS6 and TPUs, etc. They're very They're like, I'm getting a motorcycle. I'm getting a SCOD. Y

and the market will eventually get there. But in a world cross cabin Baker just told us in the show earlier. I asked him about um it looks like the trend lines are widening between the United States closed source model capabilities and Chinese open- source model capabilities. Uh there's a bunch of charts that show one line is growing linearly. The other is growing exponentially over time. That could be really good for America. It could mean a lot of different things. They could catch up. Who knows? But he was saying that uh the the big reason for that is that uh Blackwell is a great chip. they've had the ability to buy it and they've turned their back on Nvidia. Even though Jensen and Nvidia went to Washington and figured out the correct framework to allow for exports, those exports have not been received on the other side because of the push to build an indigenous supply chain. How strong is that narrative for you? What would you add to that? Uh I think on one of my last opeds and appearance of the show I told all that was going to happen

that that China was going to basically reverse ban.

Okay. Uh because and that's why I wasn't really that concerned about the initial kind of B.

Yeah. Because that that wasn't actually the goal. It's again like y'all both know my disposition. I love this country and I'm happy my family came here.

But there's a kind of an underappreciation about how actually Chinese people think and what their goals are as society. So the I agree with Gavin's kind of you listen to him on all in this other podcast like Yeah. Yeah. He's very aligned with kind of our thesis about global comput.

Sure.

The the the one thing that I would add as a caveat like the question is like does it matter and and that's what we don't really know is that that if if we're in the market as customers and I think probably all three of us has at least some German car.

Sure.

Then other models of German cars are very interesting to us. But if China doesn't care because it buys Toyotas, it buys Volkswagens, it buys whatever like lowerend stuff, then it's just a different way of understanding the preference theory of another country. And and currently I think the safest assumption that we should have as Americans is that it doesn't matter to them

and because it could be 80% and that's all it needs to be. Uh and if you look at the history of CCT policies, everything from environment to manufacturing to buying stuff from TIU, it seems like the consistent strategy is that 80% is good enough.

And and the the one thing that we do have to be very careful about which Sax has talked about pretty religiously is that they are winning on the adoption side.

Yeah. and and every kind of arc of innovation uh that has happened in in the world like previously AI America has won the several stage of the in uh like industrial revolution to other phases of the internet because we are first to adopt and China clearly is the first to adopt on robotics on drones on AI integration and and that's where maybe they figured out it doesn't matter like like they can solve it via other mechanisms like dirty power inexpensive chips and they can kind go their own their own trajectory. So, so like you know I'm I'm with the administration on the changes that they've all implemented. Uh I would say that the uh Freedberg's kind of last rant last week was definitely echoing like my view on things which is like the the government always thinks they can do something about something when they're out they can't and then they spend a bunch of energy and at the end of the day doesn't work and then they still claim victory

like the war on poverty is a great example

and and and so I feel like that's where we're going to end with AI where in the end it kind of like all this effort really didn't matter and it caused you know entrepreneurs to slow down and then eventually the government backs off and realizes that we have to uh to quote David Saxs, you know, let America cook.

Yeah. Is distillation a bigger problem than uh chip exports?

Uh I I think most of that's overblown and it's kind of hard to prove.

Yeah. I think I I think it's actually quite hard to prove and and and I I also believe there is this kind of

even if the model will just tell you straight up that

it's Yeah.

You mean Yeah. I like that. Like I I I undoubted like I'm not to to be clear it it totally does happen. The question is like how much does it matter right like like the and and I think a lot of it is oriented towards the fact of like underappreciating the fact that Chinese engineers are actually amazing at at this br and I think it's kind of all coded with that that like oh if they're copying us somehow they can't do it themselves and they have proven over and over and over again they can

and then you then go into the fact of obviously coding is now universally accepted as a wide adoption or like you know the first CSA foothold for AI Well, that's because it was also the most widely used by the customer base and it has a you know a firm yes or no answer. So now you see China which is using kind of grade B models currently but it accelerating in its adoption

like and it you know doesn't care about the what things that we care about in terms of everything from formality to safety and things like that. So, of course, they're going to be able to find use cases and things that we didn't understand. And and and that that's that's something we as Americans have so far deeply underappreciated and and you can see that in the poll numbers where data centers are more unpopular than nuclear energy.

Yeah.

And and that's a quite scary place to be. uh which is why the kind of I think in arc of the data center business is actually going to mimic the oil and gas business where they the reason why fracking became this great energy revolution

and why we became the number one exporter of oil and gas in the world is because oil and gas companies pay off owners of land.

Yeah.

So that that's just what

too I mean like you have a small piece of land that they're fracking on like you get a check in the mail.

Um and that's starting to happen. I I I there's a

Sounds like Brad is working on something.

Oh, with the Trump accounts, but also I mean

No, not even He was sort of teasing like we have something around like

Yeah,

cuz we were just saying why would you want data center in your backyard if you can get if you can use infrints from, you know, somewhere in another state. Makes sense.

Um Tyler just wanted to say hi. It's been a minute.

Oh yeah,

there we go. There's the hat.

He kept the hat.

Yeah. Oh, of course we kept Of course we kept the your gift. We would never we would have we would never uh it's it's here always in the office.

It's on our hat rack with a space helmet, hard helmet,

cowboy hats, unicorn heads,

welcome.

Oh, that's so great.

Anyway, thanks so much for taking the time.

Yeah, congrats on the milestone. Great to see you.

Yeah, thank you. It's been too long. Let's do this again soon. We got to chop it up on all sorts of things. This is always a great time.

Have a good one.

Yeah, absolutely. Yeah. And congrats to y'all, too. God bless y'all. Y'all definitely out of the few entrepreneurs