British astronaut Tim Peake envisions an era where data centers in space become a reality, and he’s a man on a mission as he partnered with Axiom Space to make orbital AI farms a viable option.
(We have covered Axiom Space several times on TechRadar Pro. When they launched a prototype with Red Hat back in March, crossed path with AWS at Re:Invent in 2022 and when they teamed up with Nokia to add 4G LTE capabilities to spacesuits.)
I met him last Friday at the Goodwood Festival of Speed, where he has been an ambassador for FOS’s Future Lab, an immersive STEM-focused exhibition covering robotics, mobility, and space, for the past eight years.
Beyond his involvement with Axiom Space, I was particularly keen to hear his thoughts about AI. “I use ChatGPT probably as much as other people do”, he told me, “What I find is if I'm asking it a question on something I know quite a bit about, but I'd just like to see what ChatGPT does, it's fascinating to see the structure and the logic and the thought processes involved.”
He is however, optimistic. “It is perhaps the most powerful tool that we've ever developed, who knows?”, he candidly added, “And maybe it's a bit like having a calculator or a watch that does 1,000 functions. At the moment, we're playing with the first three or four of those functions, not really understanding the other 997 functions that it could do. We haven't yet opened the instruction manual and really dug into it that much”.
But it is far from perfect and Tim Peake came up with this great example to illustrate the sort of mistakes ChatGPT - and indeed any other AI tool - can make. “You ask ChatGPT today what instrument did Tim Peake play on the International Space Station? It will come back and tell you I played a saxophone. I didn't play a saxophone in space. You know, I played a guitar. [French ESA astronaut] Thomas Pesquet played a saxophone. I know that as a human because I spoke to Thomas.”
Bytes in space
The reliability - and quality - of responses produced by generative AI was not the only concern for the Future Lab ambassador. Data centers, he said, are “heating vast amounts of water, as we use the water for cooling. We're simply transferring that energy into our water, which needs to reject that heat somehow. So we're again contributing to climate change problems. And it's costing a lot of money.”
Data centers in low earth orbit (LEO) make sense: thermal rejection in space is less of a problem than on Earth (but still a problem as there’s no conduction or convection happening) and the Sun can provide limitless, cheap, clean energy supply all year round (but you will still need solar panels).
The game changer is the commoditization of space travel, made possible through private firms such as Space X and Blue Origin.
They have slashed the cost of sending reasonably heavy payloads in orbit. So much so that Space X, for example, has an online calculator where anyone can transparently book flights, like you’d do for your holiday. At the time of writing, sending 100Kg in LEO by November 2025, costs $650,000.
He revealed that Axiom Space plans to launch two orbital data center nodes later this year and expects orbital data centers to reach cost parity with their earthly brethren within a decade. Axiom Space is not the only one shooting for the stars.
Another US-based company. Lonestar, is exploring the idea of putting a data center on the moon no less. It suffered a significant setback in March 2025 when its Freedom Data Center payload failed to land after Athena, the spacecraft carrying it, toppled over. Lonestar’s website hasn’t been updated since but the company aims in the long run to have a “series of ever increasingly capable multi-petabyte data storage spacecraft [sic]”.
Let’s talk about power
There’s something else, more than space travel and AI farms that Tim Peake is absolutely passionate about. If he were to invest £1 million in an emerging technology, it would be in fusion, as in nuclear fusion. “I think it's our greatest challenge.” he said, “I think it will be game-changing to live in a future where we don't even question the fact that everybody has cheap, limitless, clean energy. Imagine that. And they'll look back to our time and thinking, "Yeah, back in the days when we were burning fossil fuels." And that, you know, this is a small piece of history which is about to go extinct. And excitingly, I think it's going to go extinct fairly soon.”.
Don’t get me wrong, he added, “space travel is exciting. There's a lot going to happen. And it's kind of an incremental process. But I don't think there's nothing revolutionary in the medium term. Whereas with energy, in the medium term, I think we'll be revolutionary.”
In the near future, a not-so-insignificant portion of the world’s electrical power production (up to 14% of the total US output according to a 2024 report from research firm Semianalysis) will feed hyperscalers and AI. And that’s something that puzzles Tim, the environmentalist.
“I can't quite see the business model as to who's running these systems. The fact that they (OpenAI) are quite happy with, you know, billions and billions of people around the world anytime, able to demand so much energy from the systems that they're providing for free.”
And it will get worse once Agentic AI becomes mainstream. “The pace of learning and the quantity of data that you can analyze is going to increase exponentially. But at the expense of Earth's resources.”
PS: The entire 30-minute recording of my conversation with Tim Peake was transcribed by Google AI Studio and formatted by Google Gemini.
You must confirm your public display name before commenting
Please logout and then login again, you will then be prompted to enter your display name.