Why space data centers do not math out for him
From Why Pure Software Is Uninvestable: The SpaceX & YC Way to Build Hardware on Deeptech Decoded, with Jamie Gull.
Transcript
it's you know, it's it's not totally true, but like the thought of like working for somebody else and like not being in c total control of my destiny and like is is kind of hard to swallow. And it just wasn't interesting. Like I a lot a lot of people, you know, Pioneer Fund is a very part-time thing. Like they have a unique model where the venture partners are, you know, involved at the end of each batch, essentially at a quite a bit, but you know, in between less so. but like a lot, I had a lot of folks be like, why don't just go join? Like with your experience, like you should be able to go join as a like not super junior person at like a really good VC fund. 'Cause they would love to have your experience on you know, helping them.
Mm.
It's just like no, like
I crossed the threshold.
Yeah. Yeah.
I'm curious about Star Cloud because they were on your radar I bet, right? Since they launched.
Who what can be?
Star Cloud?
yeah.
I'm wondering how was your evaluation back then and did you have a change of heart through time?
They were not I don't think they were in my batch if if that was
No, no, they were not the company that building data centers in space is StarCraft.
Yeah, yeah.
Yeah, they were in 2024 summer batch.
Okay, yeah, I thought it was a lot more recent. yeah, like my thinking on space based data centers is evolving, but like I don't think it maths out for them and there's a bunch of reasons why at a high
Mm.
level it's It it's basically like their launch cost that they're using to do their math is like optimistic at best, if not delusional. SpaceX charges an external customer a very different price than they charge themselves internally. And it's market driven. They're the only provider at scale that can do it. So are you gonna get charged cost plus seven percent or something? No. You're getting charged whatever the market bears. So even if Starship like starts launching thousands of times per year, which
Mm.
is what it requires to get anywhere near the cost people are using to estimate these, the question is like, are there gonna be enough competitors doing the same thing where like the market price comes down for an outsider? So like there's like this big underlying assumption that like in my opinion is just false, which is a huge driver. And the other thing is like doing it with a large single data center where you have enormous solar rays and enormous radiator required has not been done. And like
