Robinhood Co-Founder's New Space Startup

Yeah, so I started the company with a concept of building a power grid in space. So what does this mean? The idea here is that low Earth orbit is this unique environment for building consumer electronics applications because you can fly one of the cheapest forms of power production, solar, and also the environment is one where oftentimes consumer-grade electronics work. So you can actually operate relatively low-cost devices in this environment. And the idea here is that this is one of the new untapped sources of power for humanity. We started out with this mission to actually beam power down to the ground for military bases or places where there's no existing power grid. And we're now spending the majority of our focus on using this concept for powering orbital data centers, which, you know, as we've all witnessed, there's a lot of talk about recently.

Yeah. So the concept is actually to build orbital AI data center satellites. So this exact concept where you take the chips and you put them directly on the satellites in space.

And yeah.

Correct. Correct.

Yeah. So, we joke around internally that it was a pivot, but we never lifted our pivot foot. To back to a basketball analogy, so you're not traveling. No, no travels. It was we never lifted our pivot foot. The thought process is actually very closely related between where we started and where the majority of the company's efforts are. Whereas the original concept we're trying to beam power down to places where there's no existing power grid, the sort of like the basic concept or the rationale for switching to orbital data centers is if you can actually use the power directly in orbit where you're generating it, you don't have to incur the transmission losses because transmitting power from space to Earth, you'll lose a significant amount of the power on the journey. And it's useful in places where there's really no alternative. But for places for applications like AI, you need as much power as you can get. And based on everything that we've been working on, the space environment is actually one that's uniquely well suited for this application. And we can talk about that a lot more.

Yeah, so we have two missions that are going to space this year. The first one is going to be demonstrating the power beaming application. Our team is actively working on doing things like internally testing GPU performance in a space environment. So simulating that, simulating all the challenges with what it's going to look like in space and sort of going down the path of designing what that future system looks like. So, we'll have a lot more to share on this, but we're targeting sometime in the first half of '27 to have GPUs in orbit that will be the building blocks for building a much broader deployment of this technology.

I think we'll start to see this be a more significant sort of application of space in the next 5 years. Hopefully a little bit sooner, but I think that's roughly the range that I have in my mind. And I think that's actually...

I think it'll start being useful for meaningful compute applications within the next 5 years. Yeah. I think the overall pie is going to grow so substantially in that time. But we actually think that, you know, within the next five to 10 years or so, ideally a little bit sooner, the advantages of doing this in space, I would say from our perspective one of the big advantages are once you get the technology up and running and you sort of work through all the technical challenges and making it do what it's supposed to, the advantage is going to be actually the ability to bring new compute online much more rapidly versus the terrestrial option. So the process by which the chips are manufactured is industrial manufacturing and on the ground the process by which they're brought online is much more driven by real estate and permitting timelines. The dream of this idea is that you could actually take the process of powering the chips and largely turn that into an industrial manufacturing problem. So, you know...