Dr. Doctor said:

---

The idea of data centers in space is dumb. Like really dumb.

One of the main issues is heat rejection aka keeping it cool.

Heat transfer in space is order of magnitude harder in space than earth. And it's hard on earth.
~egon

---

The JWST gets pretty damned cold using compressed helium in a closed loop. Just a thought.

Fightin_Aggie said:

---

Is there a required hold period for the ipo purchase?

---

Here is where Musk's ultimate market manipulation is happening.

When Musk bought Twitter he borrowed the cash. He then folded X (Twitter) into xAI without any cash paying back the lendors, just stock. He then folded xAI into Space X, again no cash but with the permission of the original lenders who received stock in SpaceX..... WITH NO LIMITS ON SELLING!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!


So these original guys can sell whenever the price rises.


But it gets better....

Musk made a deal with NASDAQ to list SpaceX on Nasdaq with only a 15 day seasoning period. Normally a company has to wait a year to be listed, Musk got 15 days.

So the stock will be listed on Day 15, which requires all Nasdaq index funds to buy it.



But it gets better...

Musk has limited public sales of Space X stock to around 5% of the shares. Most everything else is locked in.


Its going to be CRAZY!!!!!! Public holders will see upward pressure from the forced purchases by market index funds happen around day 15 but that will compete against the free selling by the people who funded X (Twitter), xAI, and Space X.


And it just gets better...

Space X only had around 19 billion in sales last year, and they lost 5 billion... but they are saying the value is 1.5 trillion!!!!!!! because they will build AI data centers in space. Are you kidding me? How are you going to power these? How are you going to cool these? How are you going to get all that stuff up there? It is a joke!


I guess history will show the best investment ever made happened here. Not when Musk built Space X, but when he invested 275 million into Trump. That 275 million is going to make him 2 trillion. And guess who will ultimately pay it? Hint, it is not Mexico.

I expect the stock will rise off the bell like a missile. It will take a few months to start dropping, but this is such a cash grab almost worthless company that it will drop. I'm staying away forever. This whole thing is fraudulent prison stuff.

What happens if/when spacex and tsla merge?

Quote:

---

Here is where Musk's ultimate market manipulation is happening.

When Musk bought Twitter he borrowed the cash. He then folded X (Twitter) into xAI without any cash paying back the lendors, just stock. He then folded xAI into Space X, again no cash but with the permission of the original lenders who received stock in SpaceX..... WITH NO LIMITS ON SELLING!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!


So these original guys can sell whenever the price rises.

---

Private credit lenders that fund LBOs typically don't want stock, as it's not their game. Also, even if they did in the form of warrants, nothing about that comes close to market manipulation

Quote:

---

But it gets better....

Musk made a deal with NASDAQ to list SpaceX on Nasdaq with only a 15 day seasoning period. Normally a company has to wait a year to be listed, Musk got 15 days.

So the stock will be listed on Day 15, which requires all Nasdaq index funds to buy it.

---

At the time of an IPO, the stock of the underlying company trades freely on the exchange. The one year wait is a specific requirement for certain SPAC transactions or reverse mergers. That does not apply here. On the index inclusion after 15 days, that was a rule change for any company that meets the threshold of being in the top 40 Nasdaq-listed companies by market cap.

Quote:

---

But it gets better...

Musk has limited public sales of Space X stock to around 5% of the shares. Most everything else is locked in.

---

IPOs or secondary offerings are only for a limited number of shares with the proceeds going to the company to fund operations.

Quote:

---

Its going to be CRAZY!!!!!! Public holders will see upward pressure from the forced purchases by market index funds happen around day 15 but that will compete against the free selling by the people who funded X (Twitter), xAI, and Space X.

---

This IPO actually has a very nonstandard lockup provision that will allow selling by insiders, directors, officers, etc. before the standard 180 day lockup period has lapsed.

Quote:

---

And it just gets better...

Space X only had around 19 billion in sales last year, and they lost 5 billion... but they are saying the value is 1.5 trillion!!!!!!! because they will build AI data centers in space. Are you kidding me? How are you going to power these? How are you going to cool these? How are you going to get all that stuff up there? It is a joke!


I guess history will show the best investment ever made happened here. Not when Musk built Space X, but when he invested 275 million into Trump. That 275 million is going to make him 2 trillion. And guess who will ultimately pay it? Hint, it is not Mexico.

I expect the stock will rise off the bell like a missile. It will take a few months to start dropping, but this is such a cash grab almost worthless company that it will drop. I'm staying away forever. This whole thing is fraudulent prison stuff.

---

The valuation is hard to justify, but nothing about any of this is fraudulent. There is also a reason that SpaceX is incorporated in Texas. Corporate oversight in the state is nowhere near the level of Delaware.

All of this looks pretty brilliant tbh. Elon deserves to be a trillionaire.

TxSquarebody said:

---

Dr. Doctor said:

---

The idea of data centers in space is dumb. Like really dumb.

One of the main issues is heat rejection aka keeping it cool.

Heat transfer in space is order of magnitude harder in space than earth. And it's hard on earth.
~egon

---

The JWST gets pretty damned cold using compressed helium in a closed loop. Just a thought.

---

Sitting behind 5 solar shields and using an active pump to keep it that cold.

You realize that in space (Low Earth Orbit, LEO) you can go from 120C to -160C?



And radiative cooling (what is going on in the thing on the right in the image) is worse than convective cooling (air fin fans or forced air cooling).

The ISS is moving through the heating/cooling cycle every 90 minutes. Moving up in altitude can increase the cycle time (slower heat/cool cycle), but you are trading latency for stability.

Clearly we can do it, as we have the assets in space. But when you look at the numbers that have been mentioned (GWs of computing power) along with the 'experience' of being in space and what all can go wrong, there is massive risk for, what seems, to be minimal return vs. building on Earth.

The power needed for cooling will be more than what it would potentially be on Earth, along with the power for computing and data transfer. Added in the need to minimize weight (since you are paying a premium to get it UP into space), along with all the extra work needed to PUT something in space (engineering reviews for launch forces for example), how do you lower costs? And then how do you repair anything up there? If you are talking making a private manned space station, that's, to me, a different animal than a 'data center in SPAAAACE! (cue Tim Curry from Red Alert)'

~egon

Will over time approach the cost of terrestrial construction while absolutely obliterating it from a speed of deployment perspective.

AI is going to be won by quantity of compute rather than quality of model. As models improve they are going to exceed the accuracy necessary for the vast, vast, vast majority of queries made and the differentiator is going to be who can answer the most queries in a given amount of time. Whoever has orbital compute will win that competition, because their deployments are assembly line rather than bespoke deployments.

Costs can (and likely will) come down. Earth has finite amounts of available land and water, and humans can't live without the latter. Given the desire to expand utilization of AI and the massive amount of cooling capacity required, figuring out space is preferable to dying of thirst.

hph6203 said:

---

Will over time approach the cost of terrestrial construction while absolutely obliterating it from a speed of deployment perspective.

AI is going to be won by quantity of compute rather than quality of model. As models improve they are going to exceed the accuracy necessary for the vast, vast, vast majority of queries made and the differentiator is going to be who can answer the most queries in a given amount of time. Whoever has orbital compute will win that competition, because their deployments are assembly line rather than bespoke deployments.

---

How do you figure that? How does the earth construction costs get 'obliterated' from space?

I have one place to make chips: Earth (Taiwan or AZ, take your pick). Once I've 'made' the computer, I then have an option. I can put it in a DC on Earth or put it on a satellite and ship to space. Unless you have a fab shop in space, you have one weak link for the deployment.

The next step is what are you putting in space vs. building out infrastructure.

A rack (basic google search) is 4k lbs (2 tons). Using Falcon 9, that's 9 racks per launch (launching JUST the computer rack). We'll make the VERY BAD assumption that the rack contains all the support stuff for the server (radio, batteries (not a bad assumption), power cables, etc.) and we are going to deploy them as starlink satellites.

BTW, the current thought is that each rack will be 30-60 kW of power. So the solar panels needed for EACH RACK is 13,500 ft^2 (ISS Panels), which is half the ISS solar panels. Again, each solar panel on the ISS weighs 5,300 lbs, so we'd need 2 per rack; a net of 10,600 lbs or the equivalent of 2.5 racks.

OH, let's not forget the heat radiator as well. Heat disapation is the same as power demand; so need 60 kW of heat removal. Again the ISS has a 14kW unit at 1,633 lb; scaled up 60 kW is now ~7,000 lbs. (another 2 racks).

Right now, PER RACK, we are at 4 + 10.6 + 7 = 21,600 lbs. Falcon 9 (recoverable) is 38,600 lbs. SO every single launch of a F9 is essentially 1.75 DC racks. A single F9 launch = $74MM

So....

If we're going to build a 1 GW datacenter (In SPAAAACEE! (sorry, I loved Red Alert and still do)), that's 1,000,000 kW of computer power OR 16,666.6 (repeating) racks/systems, rounding to 16,670.

Say we get costs down to $50MM per launch, that's $833.5 BILLION in LAUNCH COSTS ALONE. Reminder, we haven't SPENT A DIME on building this, because we assume (badly) that the computer/rack costs are the same between going in space and building on earth. You can make (badly) the assumption that the solar panels/heat radiator and all the associated bits/pieces are the same costs as building a DC on Earth (concrete, land, wires, water, etc.)

But now I have an OPEX of launches of $833.5 BILLION vs power/water/labor costs on Earth. Coupled with the idea that I can FIX something on Earth vs. putting that much in space means it will all be 'thrown away' once it gets to a point that it no longer functions. While we trash computers all day on Earth, we can technically fix things and/or repair them rather than trashing a 21,600 lb 'system' wholesale when something goes wrong.

Hell, the better idea (if you want to do distributed DC) is to put them all over the place in small buildings that are near needs. If you put 4 racks (240 kW) in a building when building it out, you now have a mini-DC in your building and surrounding area that can help with lag/speed issues. And when I say locally, I mean like put 1 in every downtown building (Houston, DFW, Austin, etc.) and now you have businesses locally covered. Besides upgrading power, you have what you need there already. Without the added costs of going to space. Or adding a new building wherever.

~egon

hph6203 said:

---

Will over time approach the cost of terrestrial construction while absolutely obliterating it from a speed of deployment perspective.

---

Re-read what I wrote. Re-read your first sentence again. Ask yourself "Am I addressing their claim?" If you answer yes, then proceed to first step in this paragraph and repeat until you answer no.


You did a lot of calculations for a launch vehicle that they are absolutely, positively, not going to launch orbital compute on and ignoring adaptable variables that can be adjusted to limit cost.


Watch this:



Read: This

Debunking the Cooling Constraint for Space Data Centers ❄️🛰️

We tackled one of the most persistent objections to orbital compute: “How do you cool data centers in space?”

Using first-principles and Starlink V3 scaling (20kW > 100kW) as a reference, we show the cooling problem… pic.twitter.com/wSSAF6RA46

— Vlad Saigau (@VladSaigau) December 17, 2025

You are under the false impression that the limit on terrestrial data centers is the cost to deploy them rather than the speed of acceptance by society. NIMBYism is a powerful effect that can raise costs dramatically.

I mean, I was just kinda half joking about the data centers in space bit. But I guess it's a real thing?

Hmm....

BigPete3281 said:

---

I mean, I was just kinda half joking about the data centers in space bit. But I guess it's a real thing?

---

Very real. Sometime Elon first mentioned I think months ago.

The plan is to put them in sun synchronous orbit along the Earth's day/night terminator do they will forever be in sunlight on one side.

My suspicion would be that the back side is the solar array could be the radiatorsv since they would forever be in darkness

SpaceX was valued at their last secondary sale at $800 billion. Their characterization of the degree to which it's overvalued by retail compared to institutional is mistaken or dishonest. The insiders have been paper rich for a very, very long time. The issue with the SpaceX IPO, OpenAI IPO and eventual Anthropic IPO is caused by their private status until they hit $1T valuations. There's only so much money they need to raise at IPO and issuing $200-400 billion to create the typical public float percent at IPO is unnecessary. That creates a mismatch between I want x company at any price and the availability of the stock in the market.

Have to fix the reason why they all decided to stay private until $1T+ valuations. It prevented a lot of Americans from making a lot of money off of their rapid rise in value.

There will be compute clusters in space. It's odd that this upsets people. There are many, many advantages. Also odd that people refuse to realize them.

VLEO (very low earth orbit) coming soon with atomic oxygen resistant materials and air breathing electric propulsion (i.e. unlimited fuel). That will be the sweet spot for data centers in space.

https://vaxonspace.com/technology

With this IPO eminent, would the better investment be this or Tesla given the rumors of the merger post IPO?