"AI" is losing money on Terra too. At least here there is a slight chance for it to work out. Data centers in space? Utter lunacy.
We really must be in the end-stage of the bubble if such ideas can gain traction.
Spencer, could you point us into any direction where one of thesr companied or geniuses explain how they plan to solve at least one of the crucial problems, namely heat removal?
Not only that, but datacenter maintenance is way more complex than typica commsl satellites. Is it going to be cheap the just crash a space datacenter if it starts failing? This GPUs running at close 100% all the time are not going to last 10 or more years, who's going to replace them in space?
This is such an insightfull take on how the futre of AI and space exploration are tied together. The connection between solar system resources and who controls them really puts things in a diferent perspective. I've been following the space race for a bit now and never really thought about it from the resource angle like this.
Yet another aspect of the AI bubble debate to be sure. I very much believe this is part of the current administration's Genesis mission, to make progress in space versus China. In recent years China has been moving faster and yet another key area. A lot of the future of the Energy race will be in space. The future of fusion Energy and all that helium-3 on the moon isn't just a tycoon adventure. Venture capitalists looking to pump space technologies will say this is about American dynamism and National Defense.
Whether or not orbital data centers are viable or not. Energy is the real story here. The country that controls the key resources on the moon has the upper hand in colonizing Mars and building out space mining facilities. Harnessing resources in the solar system isn't just hype. It will have future economic value presumably.
All things considered, Musk's Davos speech (interview) wasn't very convincing or at all the same as a confident Elon Musk of former and younger years.
This is a more humble Musk that has been basically beaten in EVs (badly) by China. Beaten in AVs by Waymo and spinning new dreams with his financial leverage and that pleases the rest of the world's financial elite. (who need to keep the AI bubble going for as along as possible).
Listen to the interview at the WEF's event in Davos, January, 2026:
A combined entity of SpaceX, xAI and Tesla might be in the running to become the most valuable company in the world circa 2028. Folding Tesla into the "great merge" would also be one of the only ways to preserve the ponzi scheme of its nonsensical valuation.
Many companies have been working on the R&D of orbital compute and space infrastructure projects for years (in secret) like Blue Origin. The engineering obstacles are minor compared to the energy efficiency. I will remind my audience that the U.S. is irrevocably behind China in Energy Infra and renewable energy investments with a power grid that's very legacy.
While demand for compute is increasing exponentially!
Demand for compute is increasing exponentially but AI compute efficiency seems to scale more linearly (in terms of response to requests). Still not clear to me how AI infrastructure gets sufficiently monetised when the bleeding edge services are effectively commoditised within months.
Can you elaborate on the part about "how AI infrastructure gets sufficiently monetised". The financial elite see it as a race with China that "we must win".
To me this isn’t like the chip race, with a Moore’s Law progression to greater efficiency where cutting edge chips led to justifiable economic advantage to those using them but still could command significant prices over time (and for lower power/binned versions across the whole industry).
People expect “basic” AI to be effectively free, and will continue to do so as it evolves. There will be areas where the cutting edge can command significant premiums (finance, military) but the moment that cutting edge is superseded (very quickly) it loses value exponentially.
Yes you are right, the return on investment story hasn't materialized yet and may never fully manifest.
If you take Meta's latest earnings, they plan to double capex but it's also because their AI Infra investments are showing up in their Ad revenue growth rate. For Cloud hyperscalers it's a bit more tricky because there's far more competition.
The ROI of all of this for ordinary citizens is very low, like I often discuss in my own skepticism.
Because the U.S. energy grid is so limited vs. the rising demand for compute - it's a huge problem for these Tycoons and the shareholders of these huge index funds. So a lot of this posturing is about the wants of the top 20% rather than the real needs of the bottom 80% of this K-shaped economy.
I have read countless of such posts in the past month. That won't stop the capital allotments and Space-tech IPO revolution. Because it's a long-term project tied to the Genesis Mission of the United States.
He’s certainly gone off the deep end in many regards. 😂 He’s a war-time CEO now of the over-promising variety and that’s never a great sign for the ability to manifest your destiny.
However, 🤔 is this a catalyst for a new kind of “American Dynamism” race with China for things like lunar bases, Mars colonlines, space mining and new energy sources? Energy that the U.S. badly needs for compute AI supremacy.
SpaceX generated about $8 billion in profit last year ahead of IPO, sources say: https://finance.yahoo.com/news/exclusive-spacex-generated-8-billion-213458153.html
As always super insightful. Regarding orbital data centers,
I’ve watched this idea move from “sci-fi” to “worth taking seriously” since 2024. Finally, we are having the most waited conversations collectively.
"AI" is losing money on Terra too. At least here there is a slight chance for it to work out. Data centers in space? Utter lunacy.
We really must be in the end-stage of the bubble if such ideas can gain traction.
Spencer, could you point us into any direction where one of thesr companied or geniuses explain how they plan to solve at least one of the crucial problems, namely heat removal?
Not only that, but datacenter maintenance is way more complex than typica commsl satellites. Is it going to be cheap the just crash a space datacenter if it starts failing? This GPUs running at close 100% all the time are not going to last 10 or more years, who's going to replace them in space?
Why, of course the space robots will do the maintenance, produced not far on the Moon :)
This is such an insightfull take on how the futre of AI and space exploration are tied together. The connection between solar system resources and who controls them really puts things in a diferent perspective. I've been following the space race for a bit now and never really thought about it from the resource angle like this.
Amazon bought Kiva Systems warehouse robots 14 years ago.
With these AI tech bros you have to take anything they say with a lot of salt.
Data Centers in space is really just part of the investor hype machine.
This isn’t just another technical problem to be solved by throwing a few billion dollars at it.
Space based electronic infrastructure has been a grand level technological challenge since the days of Skylab and it’s still not really solved.
Anyone looking into the design and construction of the ISS would realize how much of a miracle it is that it operates at all.
The two articles below make very persuasive arguments of why DCs in space aren’t going to work.
Datacenters in space are a terrible, horrible, no good idea.
https://taranis.ie/datacenters-in-space-are-a-terrible-horrible-no-good-idea/
Why Space Datacenters Are a Terrible Idea: A Technical Reality Check
https://devops-geek.net/devops-lab/why-space-datacenters-are-a-terrible-idea-a-technical-reality-check/
Yet another aspect of the AI bubble debate to be sure. I very much believe this is part of the current administration's Genesis mission, to make progress in space versus China. In recent years China has been moving faster and yet another key area. A lot of the future of the Energy race will be in space. The future of fusion Energy and all that helium-3 on the moon isn't just a tycoon adventure. Venture capitalists looking to pump space technologies will say this is about American dynamism and National Defense.
Whether or not orbital data centers are viable or not. Energy is the real story here. The country that controls the key resources on the moon has the upper hand in colonizing Mars and building out space mining facilities. Harnessing resources in the solar system isn't just hype. It will have future economic value presumably.
All things considered, Musk's Davos speech (interview) wasn't very convincing or at all the same as a confident Elon Musk of former and younger years.
This is a more humble Musk that has been basically beaten in EVs (badly) by China. Beaten in AVs by Waymo and spinning new dreams with his financial leverage and that pleases the rest of the world's financial elite. (who need to keep the AI bubble going for as along as possible).
Listen to the interview at the WEF's event in Davos, January, 2026:
https://www.youtube.com/watch?v=IgifEgm1-e0
A combined entity of SpaceX, xAI and Tesla might be in the running to become the most valuable company in the world circa 2028. Folding Tesla into the "great merge" would also be one of the only ways to preserve the ponzi scheme of its nonsensical valuation.
This seems a level-headed assessment of the benefits and challenges of space-based data centres.
TL;DR: don’t get too excited right now.
https://open.substack.com/pub/andercot/p/do-orbital-data-centers-make-sense?r=43jrc&utm_medium=ios&shareImageVariant=overlay
Many companies have been working on the R&D of orbital compute and space infrastructure projects for years (in secret) like Blue Origin. The engineering obstacles are minor compared to the energy efficiency. I will remind my audience that the U.S. is irrevocably behind China in Energy Infra and renewable energy investments with a power grid that's very legacy.
While demand for compute is increasing exponentially!
Demand for compute is increasing exponentially but AI compute efficiency seems to scale more linearly (in terms of response to requests). Still not clear to me how AI infrastructure gets sufficiently monetised when the bleeding edge services are effectively commoditised within months.
Can you elaborate on the part about "how AI infrastructure gets sufficiently monetised". The financial elite see it as a race with China that "we must win".
To me this isn’t like the chip race, with a Moore’s Law progression to greater efficiency where cutting edge chips led to justifiable economic advantage to those using them but still could command significant prices over time (and for lower power/binned versions across the whole industry).
People expect “basic” AI to be effectively free, and will continue to do so as it evolves. There will be areas where the cutting edge can command significant premiums (finance, military) but the moment that cutting edge is superseded (very quickly) it loses value exponentially.
Yes you are right, the return on investment story hasn't materialized yet and may never fully manifest.
If you take Meta's latest earnings, they plan to double capex but it's also because their AI Infra investments are showing up in their Ad revenue growth rate. For Cloud hyperscalers it's a bit more tricky because there's far more competition.
The ROI of all of this for ordinary citizens is very low, like I often discuss in my own skepticism.
Because the U.S. energy grid is so limited vs. the rising demand for compute - it's a huge problem for these Tycoons and the shareholders of these huge index funds. So a lot of this posturing is about the wants of the top 20% rather than the real needs of the bottom 80% of this K-shaped economy.
I have read countless of such posts in the past month. That won't stop the capital allotments and Space-tech IPO revolution. Because it's a long-term project tied to the Genesis Mission of the United States.
https://www.whitehouse.gov/presidential-actions/2025/11/launching-the-genesis-mission/
He’s certainly gone off the deep end in many regards. 😂 He’s a war-time CEO now of the over-promising variety and that’s never a great sign for the ability to manifest your destiny.
However, 🤔 is this a catalyst for a new kind of “American Dynamism” race with China for things like lunar bases, Mars colonlines, space mining and new energy sources? Energy that the U.S. badly needs for compute AI supremacy.