[Interview+] SpaceX's AI Data Centres Might Actually Be A Good Idea. Here's Why

I'm really glad to see all the pushback on this interview. Your guests are generally very knowledgeable and evidence-based, this one fell far short of your usual standards. Waving off the radiative cooling issue by pointing out that Starlink satellites don't have a cooling problem, therefore any data center won't be too much harder to deal with seems like wishful thinking. The amount of processing happening aboard a Starlink sat seems wildly different than what a maxed-out bank of processors would be doing to train AIs? Please consider bringing on someone to do a counterpoint interview.

I think the appropriate place for this interview is as a Patreon exclusive and labeled as a chat between friends, this is not science communication. He egregiously compares the fuel of starship with a Boeing plane, then when you brought up that it’s going high into the atmosphere he handwaves it as “just water”, I have no post high school education and even I know water is a greenhouse gas. This guest does not come close to meeting the standards of the guests you bring on.

Starlink alone is already enough to cause Kessler syndrome. Now add millions of tiny satellites on top of that. What could possibly go wrong?

Abandoning his plan to set up housekeeping on Mars is a good idea since it was bound to fail (at least in Musk's lifetime). Too many technical problems and even too expensive for him. I don't see how data centers on the moon would be cost effective, but I am not an economist so maybe it is.

So-called AI will be extremely good at coding, after all it is software writing more software, simple for a system that does dumb things very fast. As to eating the world economy, it will never happen because the real world is far more complex than computer software. If we stop calling it Artificial Intelligence, and start calling what it is, Statistical Pattern Recognition Software, then it can be used for what it is actually good at. LLM's are a stupid idea because they are trained on internet scrapings.

This argument in favor of space-based data centers relies on the assumption that demand for AI computing power will continue to grow indefinitely. The same could be said of anything: If demand for sneakers grows indefinitely, eventually it will make economic sense to put sneaker factories in space.

Fraser, I truly appreciate this interview. It is important to hear this take. However, I followed Casey for a while, but he is a boot licker. By that I mean, not only does he not reconize Musk's "Roman Salute" as an issue, he is also on board with the ethno-birth issues. So... forget all that... let's ignore all that.. Casey has never said one single thing negative about Musk, ever.. or maybe I did not notice. To me, he seems a bit biased. Now, if you want a real contrarian, please get Dr. Zubrin in an interview. FFS, Ellie in Space got him: https://www.youtube.com/watch?v=tYSMSLkIDvY edit: you just replied to me in the live Q&A about this... but... if you can have fash Casey on... why not have Dr. Zubrin on, even if you think you know his previous shtick.

You've done it. You've successfully rage baited me into making an account and commenting. Long time listener, love your RSS feeds. I'm sorry, but this is easily my least favorite interview. Your guest was factually wrong, confidently, multiple times. It makes me think he might be an AI. Exhibit A: saying that Elon is always right. Factually false, we can look at any number of predictions he's made or technologies he's promised and while some have come to pass, many have not. Exhibit B: Rockets at scale are no worse than aircraft because "it's just carbon dioxide and water." I used to think the same, but it was your own articles that convinced me otherwise. High altitude water and CO2 is much worse than low altitude https://www.universetoday.com/articles/if-we-launch-ten-times-more-rockets-well-start-damaging-the-ozone-layer. Exhibit C (more of an opinion): Saying we'll solve pollution from reentry by catching and bringing back spacecraft is insanely naive. Much like how we don't use solar farms and batteries despite them being profitable (even without subsidies, update your internal factbook if you find this shocking), I would bet good money that no (for profit) company is going to return their satellites instead of burn them up without being forced to. All I got from this interview was some guy selling more AI maximalist grift. I honestly wouldn't be surprised to hear that SpaceX is acquiring or has acquired his company with the amount of boot licking he does. It's a shame because Dr Handmer probably has some insightful knowledge, but as a non-expert, I can't tell baby from bathwater and just have to throw the whole thing out. Fraser does push back a bit he's clearly not used to needing to be the expert for his guests.

The arrogance of this obviously very smart man is astounding. He dismisses the entire desert ecosystem with no sense of curiosity about how it really works. Many good points but perhaps he should read Shade by Sam Bloch and ponder what AC has done to us. Of course sonic booms out on islands would be a problem. Has he never heard of whales and other forms of life that aren't human?

I am trying very hard to be open-minded about this, but what I'm hearing is "Elon Musk said it, I believe it, that settles it" and "money number go up for tech bros is good for people who are about to lose their jobs", and "throwing money at it always solves the problem". But yeah, maybe that's on me. What I DID hear that was an eye opening perspective was one throwaway sentence: acreage spent on fuel corn converted to acreage covered with solar panels. That's a topic worth looking into that's more likely to succeed than making fuel from literal Earth atmosphere.

There is not any solar and wind energy plant that is not subsidized. This is not your area of expertise, Fraser. No one is interested in tiny AI satellites. Economics dictate that the satellites be big.

And I am sorry, on Earth, solar power is more expensive. Everyone country relying on solar and wind has relatively high energy prices. Germany and Denmark are first and second in power rates and have been for the last 20 years.

Your guest is hopeless. SpaceX does not have a technology moat nor will it have a cost advantage long term. Everyone is figuring out reusability. Also, we don't know SpaceX profitability. We know revenues. Remember, your guest actually thinks he can make money extracting CO2 from the air and making it into a fuel. Now that is far fetched.

Probably wrong for two reasons. AI is closer to a technology fad and bubble than a revolutionary technology. The weaknesses of large language models are well known. Secondly, how do you dissipate heat given that a vacuum is a natural insolator?