Euwyn Poon: Why Elon Musk is Right About Space AI | The Further, Faster Podcast

Euwyn Poon (0:00)

Once you see the opportunity, and you realize you've seen it ahead of 99.999% people in the world, you definitely just want to plunge ahead and get on that thing and move humanity forward, and also move yourself forward and continue to learn and grow.

SPEAKER_2 (0:24)

This content is provided for informational purposes only and does not constitute legal, business, tax, or investment advice. It does not constitute an invitation or recommendation to purchase or subscribe for, and should not be relied upon when making any investment decision in any Antler Fund or any other investment instrument.

Jeff Becker (0:43)

Welcome back to The Further, Faster Podcast. I'm your host, Jeff Becker, and general partner here at Antler in New York City. Today on the show, we have Euwyn Poon, a Cornell-trained engineer who started his first company at YC back in 2010, and then went on to build Spin Scooters, which grew from to 1,800 employees in 18 months and sold to Ford for $100 million.

Now he's back with his third company, Orbital, where he's putting data centers in space. He recently closed a $5 million pre-seed with Andreessen, Basiset, and Antler. And I'm here to talk to him today about what it takes to put these machines up in orbit. Welcome to the show.

Euwyn Poon (1:20)

Awesome. Thanks for having me.

Jeff Becker (1:21)

Love having you, man. I'm excited to talk to you about this. You and I met probably six or seven years ago. You just built an exited spin.

Before we get into that, though, I mean, the question that I have on my mind and everyone has on their mind, Elon says within 30 months, that's the cheapest place to run AI is going to be in space. And a lot of people disagree with that. And so I'm just curious for your take. I'd love to hear your story about why we need to put data centers in space.

Euwyn Poon (1:48)

Yeah, no, he's right, absolutely. And I think that if you look at the trend lines, it tells the story. You see AI demand just going absolutely vertical. I mean, now we're approaching the age of inference and agents, and not only will humans be consuming tokens, but agents themselves. So we see that already.

And we see, I think the projections around 200 gigawatts of demand from the chip makers, only 100 gigawatts of new capacity from the grid. So there's going to be even through 2030, 100 gigawatt shortfall, not to mention beyond that. And the other big trend that we see is the costs of launch. We're going from 50,000 per kilogram in the space shuttle era, down to 5,000 a kilogram in this Falcon 9 era, down to potentially 500 or less in the Starship era, which is rapidly approaching. So those two trend lines paint a picture, and when you see that dramatic drop in cost, it points to an interesting opportunity. On the other hand, the sun is an incredible source of energy, potentially obviously the most important one in our universe.

Every second the sun produces more output than humanity has ever consumed. We look like a tiny speck. And to actually advance civilization, we just simply need to harness more of that. And we have pretty much maxed out grid capacity on Earth. Solar on Earth is challenging. We have day-night cycles. And it only makes perfect sense to bring solar into space, where you have five times more energy to be captured. You don't have a day-night cycle.

And it's something that I saw, and you can't unsee the obviousness of it and the potential scalability of it. And it's a great way to advance civilization.

Jeff Becker (3:45)

Yeah. There are, I think, known problems. The regulatory landscape, the ability to build a team that can understand how and what to build.

But there are some unsolved challenges, too. The thermodynamics of it and some of the things that I think people that think Elon would describe as fantasy, you know, that they think are kind of impossible challenges to solve. I'm curious how you think about the things that are possible and solvable and the things that are not.

And just your take kind of as you approach this in the next decade. Yeah.

Euwyn Poon (4:17)

When I first heard the idea, I mean, and I think the most people hear about the idea as a data center in space. The issue is that that's a bit of a misnomer. It's not a literal data center in space. And early concepts of that were kind of fantastical. And it's something that didn't actually make sense to me. I think this idea became more real, and this made sense for me to pursue when I realized that a data center in space didn't mean a giant megastructure, which is unfeasible for and unpractical for a lot of different reasons.