M&A: xAI, A SpaceX Company

[spacenut]

Because of the costs, regulations on earth, and power needs, data centers in space make sense.  A solar panel in space can produce 10 times more power than on earth, because of continuous 24/7 power, no cloudy days, etc.  With Starlink operational to communicate with the solar powered data satellites, communication shouldn't be that slow. 

Not only Musk but others are advocating for in space data centers. 

I am retired, but use Google AI as well as Grok.  Wife has used Chat GTP and Grok as well. Grok is more accurate.  Then, we have Amazon Echos all over the house and they are now using AI to talk to you.  So Bezos can get involved launching AI's with New Glenn for Amazon. 

AI's will need a lot of chips, programming, and will need a lot of power that America's grid cannot handle right now.  So, in space they go using solar power.  Lots of new reusable rockets coming on line, as well as SpaceX and Blue Origin.  So it looks like this is the way it will go regardless of what we think. 

[Vultur]

Not to mention that silicon chips need some of the most complex manufacturing processes ever invented. Multiple stages with exotic chemicals, machines and conditions. It's hard enough on Earth!

I actually hope the AI bubble pops before SpaceX gets a chance to IPO. It's a massive distraction.

Wouldn't it be better to hope that AI doesn't experience a bubble, or that the bubble is only a consolidation that doesn't affect the overall market size?

I mean why is AI a distraction? It seems at the very least a very fundamental technology.

If one takes the existence of a bubble as a fact, and I pretty much do, bursting sooner is better, because it means less wealth erased and fewer plans committed to things that won't work out.

A bubble doesn't mean the technology is useless. The Internet was a bubble in the 1990s. US railroads were a bubble bursting in 1873.

"AI" is a very broad term. If "AI" in some sense is an eventual huge success but with a technology totally different from LLMs (and possibly far less energy-hungry) much of the current investment, hardware, etc.

The market leaders, if they were based on technology rather than hype, survive.

Stocks take a couple of years to recover, but the failure of the smaller players makes it easier to become even more dominant.

Like I said before - money already raised is already raised, and revenue will be generally unaffected. It's not like people will stop talking on their phones or using the web.

You worry too much. It's not healthy.

You could be right, a "soft" bubble burst is a plausible scenario.

But the other outcome is also possible, where revenue *does* crash because most of it is circular financing rather than real customer demand, and there simply aren't enough *paying* customers to make ends meet - the scenario where it turns out tons of people will use AI when it's free but not nearly enough will pay high prices for it.

(Unless you mean SpaceX's current Starlink revenue. That I agree is safe. The only risk is anything specific to AI hardware.)

[hplan]

You could be right, a "soft" bubble burst is a plausible scenario.

But the other outcome is also possible, where revenue *does* crash because most of it is circular financing rather than real customer demand, and there simply aren't enough *paying* customers to make ends meet - the scenario where it turns out tons of people will use AI when it's free but not nearly enough will pay high prices for it.

(Unless you mean SpaceX's current Starlink revenue. That I agree is safe. The only risk is anything specific to AI hardware.)

I read so many comments to the effect that there aren't many paying customers for AI yet, so a crash is imminent, that I wonder what I am missing.

OpenAI's revenue is projected to be $15 - 20 billion for 2025. That's very far from "tons of people will use AI when it's free" but "there simply aren't enough *paying* customers." Revenue is also growing dramatically -- 2024 revenue was only $4 billion.

Now $15-20 billion revenue per year is not enough to pay for the $500 billion worth of data centers that OpenAI has announced,  but it represents a significant and quickly growing number of paying customers. It's a level of revenue similar to that of SpaceX, and growing faster.

No doubt there will be some startups closing down, others adjusting their focus. No doubt some investors will lose a lot of money, but others will make a ton. Maybe OpenAI's announced plans are too grandiose and will have to be scaled back. But this is all just the normal give and take of the emergence of a new branch of the economy. Call it a bubble, if you must, but AI is already changing the world in dramatic ways and I don't see evidence that it is slowing down.

[Vultur]

You could be right, a "soft" bubble burst is a plausible scenario.

But the other outcome is also possible, where revenue *does* crash because most of it is circular financing rather than real customer demand, and there simply aren't enough *paying* customers to make ends meet - the scenario where it turns out tons of people will use AI when it's free but not nearly enough will pay high prices for it.

(Unless you mean SpaceX's current Starlink revenue. That I agree is safe. The only risk is anything specific to AI hardware.)

I read so many comments to the effect that there aren't many paying customers for AI yet, so a crash is imminent, that I wonder what I am missing.

OpenAI's revenue is projected to be $15 - 20 billion for 2025. That's very far from "tons of people will use AI when it's free" but "there simply aren't enough *paying* customers."

Essentially, that $15-20B is "not that much" on the completely ridiculous scale of current AI investment.

The argument isn't that there is zero revenue, just not nearly enough. What's the lifetime of the current AI data centers? If you have to replace $800B of datacenters every five years... Then add in the cost of all the extra power generation that wouldn't otherwise be needed (US power demand had been essentially flat for 20 years before the current data center boom...)

Also, $15-20B is a projection, not reality yet. And one I doubt. First half 2025 revenue was supposedly 4.3B, and I've read that ChatGPT user growth has slowed a lot over the last few months, so I'm skeptical they'll actually hit 15-20B.

But say it does. Where do you go from there? Competition is rising. ChatGPT was the first AI to become a household name but Google is pushing Gemini hard. Even if there are enough users to pay for all the data centers (and it's not clear that there are, ChatGPT supposedly has something like 800M weekly users currently ... what's the plausible upper end?) can anyone charge enough to pay for them when competition will eat their lunch?

(And there's other issues too...)

[Vultur]

What I would give to be able to be a fly on the wall for the next 200 years. Neuralink needs to get its Psyche-Upload-To-The-Cloud service launched.

I am very skeptical that uploading minds is even theoretically possible, even with arbitrarily advanced technology. The brain doesn't store information in the same way an electronic computer does; how would you get all the information out without destroying the 3D structure that is key to storing it? I think you'd need something like Star Trek scanners, which probably aren't physically possible.

(Anyway even if possible it would be the Star Trek transporter problem .. it's a copy of you not *you*. It's not immortality just a nonbiological form of reproduction, a sort of mental cloning.)

All I can do is live in hope, lol. I really wouldn't be worried what version of me lives on if it became a possibility.

Eh, I guess this is a philosophical divide. From my perspective a copy of me wouldn't be "me" in any meaningful sense, there would be no continuity of identity. (Ship of Theseus, etc )

But this is off topic.

Back on topic ... I actually don't think this is a "wait 200 years" thing. With the end of Moore's law, I think these sorts of computer based things will either work out in the next 20 years, or never. After that the primary direction of advancement probably won't be computer/IT at all.

[meekGee]

What I would give to be able to be a fly on the wall for the next 200 years. Neuralink needs to get its Psyche-Upload-To-The-Cloud service launched.

I am very skeptical that uploading minds is even theoretically possible, even with arbitrarily advanced technology. The brain doesn't store information in the same way an electronic computer does; how would you get all the information out without destroying the 3D structure that is key to storing it? I think you'd need something like Star Trek scanners, which probably aren't physically possible.

(Anyway even if possible it would be the Star Trek transporter problem .. it's a copy of you not *you*. It's not immortality just a nonbiological form of reproduction, a sort of mental cloning.)

Just like you are just a copy of yesterday's you.  The real yesterday's you died overnight, as happens every night. Yup, that conversation from early childhood - how to define continuity of consciousness... 

Good thing current plans are not dependent on such pseudo-technology.

[meekGee]

Not to mention that silicon chips need some of the most complex manufacturing processes ever invented. Multiple stages with exotic chemicals, machines and conditions. It's hard enough on Earth!

I actually hope the AI bubble pops before SpaceX gets a chance to IPO. It's a massive distraction.

Wouldn't it be better to hope that AI doesn't experience a bubble, or that the bubble is only a consolidation that doesn't affect the overall market size?

I mean why is AI a distraction? It seems at the very least a very fundamental technology.

If one takes the existence of a bubble as a fact, and I pretty much do, bursting sooner is better, because it means less wealth erased and fewer plans committed to things that won't work out.

A bubble doesn't mean the technology is useless. The Internet was a bubble in the 1990s. US railroads were a bubble bursting in 1873.

"AI" is a very broad term. If "AI" in some sense is an eventual huge success but with a technology totally different from LLMs (and possibly far less energy-hungry) much of the current investment, hardware, etc.

The market leaders, if they were based on technology rather than hype, survive.

Stocks take a couple of years to recover, but the failure of the smaller players makes it easier to become even more dominant.

Like I said before - money already raised is already raised, and revenue will be generally unaffected. It's not like people will stop talking on their phones or using the web.

You worry too much. It's not healthy.

You could be right, a "soft" bubble burst is a plausible scenario.

But the other outcome is also possible, where revenue *does* crash because most of it is circular financing rather than real customer demand, and there simply aren't enough *paying* customers to make ends meet - the scenario where it turns out tons of people will use AI when it's free but not nearly enough will pay high prices for it.

(Unless you mean SpaceX's current Starlink revenue. That I agree is safe. The only risk is anything specific to AI hardware.)

That's why I feel SpaceX is safe - it has many levels of revenue on that front:
- AI at the application level (highest potential revenue, highest risk)
- AI at the infrastructure level (this new-fangled orbital compute thrust)
- Orbital communications (virtual guaranteed monopoly, very recession proof)
- Other launch dependent streams such as commercial and military and NASA (safe, super high margins, unique capabilities)

- by association, consumers in the cars and robotics industry

If anyone survives a bubble, hard or soft, it's them

[Vultur]

That makes sense; I'd just rather not see them moving up that ladder of risk.

[meekGee]

That makes sense; I'd just rather not see them moving up that ladder of risk.

No risk, no frisk.

Every single achievement they own right now is a result of risk taking.

Want to avoid risk? That's in the other subforum.

Srsly.  At the top of Musk's resume is "risk management", not "avoidance".

[Twark_Main]

Summarizing, there seem to be two main arguments in favor of space-based AI compute:

    1. Fewer PV and batteries needed vs terrestrial solar. Because AI requires so much power 24/7, this is actually a big deal.

    2. It doesn't deplete the (large but still finite) anthropogenic waste heat rejection capacity of the surface of the Earth. This is the "if you tried to beam the Sun's power to Earth the planet would melt" reason.


#1 is the short-term reason, and it's the reason why Musk says space-based AI will be there cheapest option in 2-3 years.


#2 is the long-term reason, and it doesn't effect current economics because we don't have a Joule Tax yet. This would be like a coal company planning around a Carbon Tax in 1890. It's too early.


And yes, before someone says it, I'm aware that PV-powered chips don't change the Earth's radiant power balance, but it does still "leach" exergy (aka "useful work") from the biosphere and agriculture and other industrial processes. There is a finite (but large) limit on the amount of available exergy that can be siphoned off from the Earth's total exergy budget before those other systems start being starved of useful work, due to unavoidable physics (thermodynamic) constraints.



_{Hopefully we all understand thermodynamics enough that we can skip the back-and-forth about entropy vs exergy vs energy.}

[spacenut]

AI is just getting started.  I don't see an AI bubble burst for several years.  Largest AI companies will probably survive this.

[meekGee]

Summarizing, there seem to be two main arguments in favor of space-based AI compute:

    1. Fewer PV and batteries needed vs terrestrial solar. Because AI requires so much power 24/7, this is actually a big deal.

    2. It doesn't deplete the (large but still finite) anthropogenic waste heat rejection capacity of the surface of the Earth. This is the "if you tried to beam the Sun's power to Earth the planet would melt" reason.


#1 is the short-term reason, and it's the reason why Musk says space-based AI will be there cheapest option in 2-3 years.


#2 is the long-term reason, and it doesn't effect current economics because we don't have a Joule Tax yet. This would be like a coal company planning around a Carbon Tax in 1890. It's too early.


And yes, before someone says it, I'm aware that PV-powered chips don't change the Earth's radiant power balance, but it does still "leach" exergy (aka "useful work") from the biosphere and agriculture and other industrial processes. There is a finite (but large) limit on the amount of available exergy that can be siphoned off from the Earth's total exergy budget before those other systems start being starved of useful work, due to unavoidable physics (thermodynamic) constraints.



_{Hopefully we all understand thermodynamics enough that we can skip the back-and-forth about entropy vs exergy vs energy.}

#2 is true in theory but is sooooo far away that other mechanisms will kick in first.

You're talking about directly influencing the energy balance of the planet, not just messing with greenhouse gasses.

Starship: 1-100 GWatt/yr (reasonable estimate)
Lunar: 0.1-10 TWatt/yr (very handwavy)
Earth energy uptake 120,000 TWatt

So...  If we install 10 TWatt/yr over 10 millennia, we'll have ourselves a competition.

Until then, it's mostly reason #1 and some other related ones.

[Twark_Main]

Summarizing, there seem to be two main arguments in favor of space-based AI compute:

    1. Fewer PV and batteries needed vs terrestrial solar. Because AI requires so much power 24/7, this is actually a big deal.

    2. It doesn't deplete the (large but still finite) anthropogenic waste heat rejection capacity of the surface of the Earth. This is the "if you tried to beam the Sun's power to Earth the planet would melt" reason.


#1 is the short-term reason, and it's the reason why Musk says space-based AI will be there cheapest option in 2-3 years.


#2 is the long-term reason, and it doesn't effect current economics because we don't have a Joule Tax yet. This would be like a coal company planning around a Carbon Tax in 1890. It's too early.


And yes, before someone says it, I'm aware that PV-powered chips don't change the Earth's radiant power balance, but it does still "leach" exergy (aka "useful work") from the biosphere and agriculture and other industrial processes. There is a finite (but large) limit on the amount of available exergy that can be siphoned off from the Earth's total exergy budget before those other systems start being starved of useful work, due to unavoidable physics (thermodynamic) constraints.



_{Hopefully we all understand thermodynamics enough that we can skip the back-and-forth about entropy vs exergy vs energy.}

#2 is true in theory but is sooooo far away that other mechanisms will kick in first.

You're talking about directly influencing the energy balance of the planet, not just messing with greenhouse gasses.

Starship: 1-100 GWatt/yr (reasonable estimate)
Lunar: 0.1-10 TWatt/yr (very handwavy)
Earth energy uptake 120,000 TWatt

...

Until then, it's mostly reason #1 and some other related ones.

Indeed! I thought I made that clear (in fact, pointing out this distinction was my main reason for posting), but it's always good to re-iterate and re-phrase the point.


"#2... would be like a coal company planning around a Carbon Tax in 1890. It's too early."

"(large but still finite) anthropogenic... heat capacity"


We may wish that fossil fuel companies had the foresight to address global warming from the start. Well Tesla is doing exactly what we might wish, but the popular reaction is instead to ridicule the fix as "uneconomical."

Fortunately, reason #1 is enough to make space-based AI economical even today. But you gotta give points for forward thinking... 

So...  If we install 10 TWatt/yr over 10 millennia, we'll have ourselves a competition.

Growth is exponential, more like 2% CAGR.


Thanks. A rare opportunity to use "exponential" in its real math meaning instead of "very big." 

[meekGee]

Summarizing, there seem to be two main arguments in favor of space-based AI compute:

    1. Fewer PV and batteries needed vs terrestrial solar. Because AI requires so much power 24/7, this is actually a big deal.

    2. It doesn't deplete the (large but still finite) anthropogenic waste heat rejection capacity of the surface of the Earth. This is the "if you tried to beam the Sun's power to Earth the planet would melt" reason.


#1 is the short-term reason, and it's the reason why Musk says space-based AI will be there cheapest option in 2-3 years.


#2 is the long-term reason, and it doesn't effect current economics because we don't have a Joule Tax yet. This would be like a coal company planning around a Carbon Tax in 1890. It's too early.


And yes, before someone says it, I'm aware that PV-powered chips don't change the Earth's radiant power balance, but it does still "leach" exergy (aka "useful work") from the biosphere and agriculture and other industrial processes. There is a finite (but large) limit on the amount of available exergy that can be siphoned off from the Earth's total exergy budget before those other systems start being starved of useful work, due to unavoidable physics (thermodynamic) constraints.



_{Hopefully we all understand thermodynamics enough that we can skip the back-and-forth about entropy vs exergy vs energy.}

#2 is true in theory but is sooooo far away that other mechanisms will kick in first.

You're talking about directly influencing the energy balance of the planet, not just messing with greenhouse gasses.

Starship: 1-100 GWatt/yr (reasonable estimate)
Lunar: 0.1-10 TWatt/yr (very handwavy)
Earth energy uptake 120,000 TWatt

...

Until then, it's mostly reason #1 and some other related ones.

Indeed! I thought I made that clear (in fact, pointing out this distinction was my main reason for posting), but it's always good to re-iterate and re-phrase the point.


"#2... would be like a coal company planning around a Carbon Tax in 1890. It's too early."

"(large but still finite) anthropogenic... heat capacity"


We may wish that fossil fuel companies had the foresight to address global warming from the start. Well Tesla is doing exactly what we might wish, but the popular reaction is instead to ridicule the fix as "uneconomical."

Fortunately, reason #1 is enough to make space-based AI economical even today. But you gotta give points for forward thinking... 

So...  If we install 10 TWatt/yr over 10 millennia, we'll have ourselves a competition.

Growth is exponential, more like 2% CAGR.


Thanks. A rare opportunity to use "exponential" in its real math meaning instead of "very big." 

Awright!

---

Meanwhile double checking:

Start with 1 GWatt/yr and grow by 10% every year, getting to 100,000 TWatt is a 170 years.

So yup, about the same time as from the industrial revolution to now.

Though, chat argues that worldwide energy use, since 1800, only grew by a factor of 30, not 100,000,000...

(I pushed back a bit but he's defending that 30x pretty well.)

[thespacecow]

Varda Space employee Andrew McCalip built a cost model for orbital data centers: https://andrewmccalip.com/space-datacenters, you can use sliders to change various assumptions and the model will calculate the cost of orbital data center and compare it to terrestrial data center.

His assessment:

Findings and Implications

Here's the headline result: it's not obviously stupid, and it's not a sure thing. It's actually more reasonable than my intuition thought! If you run the numbers honestly, the physics doesn't immediately kill it, but the economics are savage. It only gets within striking distance under aggressive assumptions, and the list of organizations positioned to even try that is basically one.

That "basically one" point matters. This isn't about talent. It's about integration. If you have to buy launch, buy buses, buy power hardware, buy deployment, and pay margin at every interface, you never get there. The margin stack and the mass tax eat you alive. Vertical integration isn't a nice-to-have. It's the whole ballgame.

[thespacecow]

Comments on the cost model by Dr. Phil Metzger: https://x.com/DrPhiltill/status/2000678045959774521

Solid snapshot analysis but I think it omits the main point. The stated assumptions include, “No adjustments for permitting or regulatory delay.” But that’s the whole reason** to build in space. My analysis also shows it’s not economically viable today, and it is the regulatory delay dynamic that flips it in about 10 years. With the SpaceX architecture of distributed compute built on Starlink, I think Elon’s 3-year estimate is also quite viable, again precisely due to the regulatory delay difference.

**The ultimate reason to build in space is to preserve the environment of Earth, but the way that decision-making gets implemented in our democratic society is through the political-regulatory system. There is no other dynamic to capture this. As damage to the environment increases and voters notice, regulation increases the delay costs, the taxes, and the restrictions on implementing the industry, which creates the economic conditions that result in a different choice. If you leave out this dynamic from a predictive model, you have left out the motive to go to space.

Note that Elon didn’t say “cheapest”. He said “fastest.” We have to include the differences in speed and the costs that accrue from delay to see how space is cheaper.

[mikegi]

Your brain and mine require something like 20 to 25W of power. I don't think that the current models of AI capture what's really going on (by many orders-of-magnitude, even correcting for the magnitude of analog-to-digital conversion). I don't know the correct algorithm, but I do know that this isn't it. The current algorithms *might* be equivalent to the retina, but that's about it.

[Twark_Main]

Summarizing, there seem to be two main arguments in favor of space-based AI compute:

    1. Fewer PV and batteries needed vs terrestrial solar. Because AI requires so much power 24/7, this is actually a big deal.

    2. It doesn't deplete the (large but still finite) anthropogenic waste heat rejection capacity of the surface of the Earth. This is the "if you tried to beam the Sun's power to Earth the planet would melt" reason.


#1 is the short-term reason, and it's the reason why Musk says space-based AI will be there cheapest option in 2-3 years.


#2 is the long-term reason, and it doesn't effect current economics because we don't have a Joule Tax yet. This would be like a coal company planning around a Carbon Tax in 1890. It's too early.


And yes, before someone says it, I'm aware that PV-powered chips don't change the Earth's radiant power balance, but it does still "leach" exergy (aka "useful work") from the biosphere and agriculture and other industrial processes. There is a finite (but large) limit on the amount of available exergy that can be siphoned off from the Earth's total exergy budget before those other systems start being starved of useful work, due to unavoidable physics (thermodynamic) constraints.



_{Hopefully we all understand thermodynamics enough that we can skip the back-and-forth about entropy vs exergy vs energy.}

#2 is true in theory but is sooooo far away that other mechanisms will kick in first.

You're talking about directly influencing the energy balance of the planet, not just messing with greenhouse gasses.

Starship: 1-100 GWatt/yr (reasonable estimate)
Lunar: 0.1-10 TWatt/yr (very handwavy)
Earth energy uptake 120,000 TWatt

...

Until then, it's mostly reason #1 and some other related ones.

Indeed! I thought I made that clear (in fact, pointing out this distinction was my main reason for posting), but it's always good to re-iterate and re-phrase the point.


"#2... would be like a coal company planning around a Carbon Tax in 1890. It's too early."

"(large but still finite) anthropogenic... heat capacity"


We may wish that fossil fuel companies had the foresight to address global warming from the start. Well Tesla is doing exactly what we might wish, but the popular reaction is instead to ridicule the fix as "uneconomical."

Fortunately, reason #1 is enough to make space-based AI economical even today. But you gotta give points for forward thinking... 

So...  If we install 10 TWatt/yr over 10 millennia, we'll have ourselves a competition.

Growth is exponential, more like 2% CAGR.


Thanks. A rare opportunity to use "exponential" in its real math meaning instead of "very big." 

Awright!

---

Meanwhile double checking:
 
Start with 1 GWatt/yr and grow by 10% every year, getting to 100,000 TWatt is a 170 years.

So yup, about the same time as from the industrial revolution to now.

Exactly, it's hardly some unimaginable span of time! Even at a more reasonable (and more historical) 2-3% CAGR, we're down from "ten millennia" to less than one. 

Though, chat argues that worldwide energy use, since 1800, only grew by a factor of 30, not 100,000,000...

(I pushed back a bit but he's defending that 30x pretty well.)

Sure, that's exactly in line with what we should be expecting. That's a CAGR of 1.5%.

[Twark_Main]

Your brain and mine require something like 20 to 25W of power. I don't think that the current models of AI capture what's really going on (by many orders-of-magnitude, even correcting for the magnitude of analog-to-digital conversion). I don't know the correct algorithm, but I do know that this isn't it. The current algorithms *might* be equivalent to the retina, but that's about it.

Landauer's Limit is a long way down, so a lot of those orders-of-magnitude can still come from the hardware side.

Still, this is pretty uncontroversial. I think you'd be hard-pressed to find anyone who thinks today's algorithms are the end. Everyone expects future algorithms will improve efficiency, so it's already "baked into the cake" as it were.

[Twark_Main]

New article on the economics of space-based AI.

https://andrewmccalip.com/space-datacenters

Discussed here https://news.ycombinator.com/item?id=46281288


Notably their numbers assume fossil fuels are used to power the terrestrial datacenter, whereas I presume Musk is running numbers for solar + batteries because it's the only scalable solution. It also uses rather OldSpace-like assumptions for satellite and launch costs.

It does have a cool calculator function though, so with how these things go I assume this will become the "go-to" de facto source of misinformation on the subject...