BennyBlanco
aka IMurRIVAL69
So now twitter, grok, and SpaceX are all under the same company I guess. Not sure what the point is but I'm sure it's some kind of tax dodge.
-
Hey Guest. Check out your NeoGAF Wrapped 2025 results here!
SpaceX has acquired xAI
- Thread starter BennyBlanco
- Start date
So now twitter, grok, and SpaceX are all under the same company I guess. Not sure what the point is but I'm sure it's some kind of tax dodge.
Chittagong
Gold Member
Simple, making sure bankers will foot the bill for his seemingly non-sensible vanity plays in the future too. Anyone backing the Twitter acquisition will have made a fantastic return. Imagine telling that to someone a few years ago.
falconhoof
Member
EternalSunshine1976
Member
This is the first domino. The next is the SpaceX IPO, then a merger with Tesla...
BennyBlanco
aka IMurRIVAL69
1.25 trillion IPO potentially. Idk what value xAI is bringing to the table though. It seems like small potatoes compared to the other big AI companies.
HoodWinked
Member
the datacenters in space seemed like a joke...
EternalSunshine1976
Member
There isn't much overlap as it relates to SpaceX, xAI and Tesla, so anti-trust likely would not be as much as an obstacle as if they were in any way competitors in the same industries.
Dacvak
No one shall be brought before our LORD David Bowie without the true and secret knowledge of the Photoshop. For in that time, so shall He appear.
Anti-trust doesn't exist anymore lmao
EternalSunshine1976
Member
Arguable it does, and perhaps I'll even give you it is not enforced heavily in the US... but it definitely is in Europe.
Last edited:
EternalSunshine1976
Member
Until it isn't. I mean, what are Starlink's satellites? They are solar powered, satellites with powerful processors communicating with each other through lasers. The proof of concept of megaprocessing AI chips isn't there yet, but the model from Starlink's satellites can be the basis for the datacenters in space...
Last edited:
Bark
Member
TBF, this is Musk selling his shit to himself. Not much trust to break there.
EternalSunshine1976
Member
Related? Perhaps, perhaps...
Nvidia Stock Drops on Report of Paused $100 Billion OpenAI Deal
ChuckeRearmed
Member
Makes sense. Merge with Tesla might be interesting too. Very interesting vertical integration for the future.
SpaceX + xAI + Tesla + Neuralink can become an insane combo.
SpaceX + xAI + Tesla + Neuralink can become an insane combo.
Last edited:
Mengy
wishes it were bannable to say mean things about Marvel
The plan is this:
xAI needs huge compute to train and run its future AI ventures. Land based data centers are getting huge and cumbersome, requiring lots of water and power to operate. Satellite based data centers solve both problems: constant solar power plus radiative cooling in the cold vacuum of space. SpaceX is presently the only rocket company even capable of launching something like a vast solar orbit data/compute satellite network. On a price per compute comparison, a satellite data center is about 1/8th the cost of a comparable land based data center. That's with Starship launching the satellites.
Now, the next step of the plan is very likely Tesla acquiring SpaceX in another merger, brining all three Musk companies together so they can share resources and people. Tesla is planning to build its own chip fab, they are calling it the Tera Factory, to supply chips for its Optimus humanoid robot in the future. Current chip fabs can't make enough chips for the supply Tesla will need for Optimus, thus they need to build their own fab. Tesla AI chips will likely be the same chips going into the xAI satellites, with Tesla supplying those chips to xAI so SpaceX can launch and maintain their satellite network.
That's the general idea.
![L*][*N*K](/data/avatars/m/842/842515.jpg?1756724107){kind=avatar}
ForAcademicPurposes
Gold Member
The shell game continues.
It's to IPO both at the same time before the ai bubble burst.
MisterFalcon
Member
When I look for the source of the claim a space datacenter is 1/8th the cost of a land one I find this: https://research.google/blog/exploring-a-space-based-scalable-ai-infrastructure-system-design/ which mentions a space solar panel is 8 times as efficient but once you take into account the costs of launches it will only get comparable in operating cost in the future with launch costs at $200 per kg. The Falcon Heavy is at $1400 and that 200 is projected to be the Starship efficiency, a vehicle that has exploded on launch more than 50% of the time. I'm sure eventually the Starship will become more reliable but also the $200 is a ridiculously hyped cost that will not be realized.
Mengy
wishes it were bannable to say mean things about Marvel
When I look for the source of the claim a space datacenter is 1/8th the cost of a land one I find this: https://research.google/blog/exploring-a-space-based-scalable-ai-infrastructure-system-design/ which mentions a space solar panel is 8 times as efficient but once you take into account the costs of launches it will only get comparable in operating cost in the future with launch costs at $200 per kg. The Falcon Heavy is at $1400 and that 200 is projected to be the Starship efficiency, a vehicle that has exploded on launch more than 50% of the time. I'm sure eventually the Starship will become more reliable but also the $200 is a ridiculously hyped cost that will not be realized.
I saw a breakdown of the costs online, forget what the site was now, it was talking about operational costs over time though. Building costs were cheaper for SpaceX data satellites too vs land based, but not by that extreme a factor.
The thesis of the article was that no land based data centers would come anywhere near the cost per compute which SpaceX + xAI could.
Last edited:
EternalSunshine1976
Member
All fair points, but Starship blows up because they push to the point of failure in almost everything that they do so they can stress test it to the edge. It allows for much quicker iteration and progress, but does result in a lot more fireworks for certain.When I look for the source of the claim a space datacenter is 1/8th the cost of a land one I find this: https://research.google/blog/exploring-a-space-based-scalable-ai-infrastructure-system-design/ which mentions a space solar panel is 8 times as efficient but once you take into account the costs of launches it will only get comparable in operating cost in the future with launch costs at $200 per kg. The Falcon Heavy is at $1400 and that 200 is projected to be the Starship efficiency, a vehicle that has exploded on launch more than 50% of the time. I'm sure eventually the Starship will become more reliable but also the $200 is a ridiculously hyped cost that will not be realized.
Honestly, their cost projections are likely pie in the sky, but even if they are somewhat in the neighborhood, the long term savings are substantial. Instead of a 1/8 cost, all they need is a 1/2 cost make it ultimately feasible because the incredible speed that they are looking at… I mean they asked for permission to launch 1M satellites… no chance that's what they ultimately do, but it's somewhat of an indication of the ramp they are planning…
If the ramp is only limited by the speed of the manufacturing and launches, then the pace of expansion will completely dwarf any terrestrial data centers.
Last edited:
Chittagong
Gold Member
EverydayBeast
ChatShitGPT Alpha 0.001
$1.25 trillion combined
Trilobit
Absolutely Cozy
Imagine how posting stuff like this on X will in a very small way help humanity land on Mars:
Our descendants will be in awe when we are hailed as heroes by historians.
IntentionalPun
Ask me about my wife's perfect butthole
A lot of this maneuvering has been to appease the other investors in what financially was a disastrous investment in Twitter.
The XAi merger values their shares essentially what they paid for them, and then this merger gives them like a 600% ROI at least on paper. Whereas twitter on it's own dumped most of it's value in a year lol
The XAi merger values their shares essentially what they paid for them, and then this merger gives them like a 600% ROI at least on paper. Whereas twitter on it's own dumped most of it's value in a year lol
Last edited:
EternalSunshine1976
Member
No lies detected. the Twitter purchase went from a disaster to something that gave the investors a huge return... What's most interesting to me is that the synergies between the unmatched user (both regular, news sources and celebrities / personalities) input / database of Twitter / X actually becomes something incredibly valuable for XAi and its training clusters.
You have to admit, Twitter is less valuable as an "ad supported social network" than a collection of both news sources, commentary and debate that serves as a knowledge base for generative AI. Yeah, much of it is trash (i.e., shit posts), but even that is part of humanity. The rest? Pretty valuable stuff that no one has in one place, in one database.
MisterFalcon
Member
Making plans is easy. Just assume launch costs go down by 90 percent in the next 10 years while terrestial data centers stay just as expensive and voila! A massively positive business case to make techbros reach for their wallets to make the Shadowrun future a reality.
And an extra-territorial data cloud also solves a lot of the current problems Musk is having with the CP content being generated. That's got to be worth a few hundred billion.
Do these studies explain how the heat from all that hardware is dealt with in a vacuum?
MidGenRefresh
*Refreshes biennially
Chittagong
Gold Member
Confirmed by FT today
FT said:
Imagine telling someone that investing into buying Twitter for 40B would triple your money in a few years back when the deal was announced
IntentionalPun
Ask me about my wife's perfect butthole
I would not consider this an indication of anything other than Elon's usual pumping up of numbers/expectations which just so happen to make his net worth go way up. You yourself seem to be aware of how bullshit of a number it is.
EternalSunshine1976
Member
SpaceX has been launching satellites for its Starlink system for years, and I believe that number is up there around 9 to 10 thousand and growing. Each of those satellites employs solar, and both passive and active cooling and heating to make sure that the sensitive electronics components are not hit with huge swings in temperature. The cooling is done both by transferring heat to areas that need to have moderate warmth in the internals of the satellites, as well as efficiently radiating the heat away from the satellite often through infrared radiation.
The main form of heat dissipation will be through passive radiators with enormous surface area that emit IR radiation placed below the solar panels and other non-sun facing areas. I'm no material scientist, but there's no doubt that some of the engineers from SpaceX are, and no doubt that they are WORLD CLASS engineers... These same engineers who have achieved previously impossible goals, are working to address the almost pedestrian issue of additional heat dissipation needs for data-center chips.
There is also another real, but man-made obstacle, that often causes these new data centers to be extremely expensive and actually impossible. That is land-use, and resistance from the community (i.e., NIMBY). There is already huge resistance to some of the planned new data-centers... because of ecological effects, energy consumption, water consumption, waste products, etc. Then there is the permitting and red-tape that needs to be maneuvered whenever building a large project. To create a data-center on the scale that SpaceX is conceiving would be entirely impossible on land.
Last edited:
EternalSunshine1976
Member
I'm not a black and white guy, I know that the number is pumped up beyond reality... however, there remains a reality under pinning the proposal, and it is not an insignificant in number. In fact, it is likely a very significant number like something in the hundreds of thousands.
Your skepticism somewhat mirrors the skepticism that people had about Tesla ever making more than 100,000 EVs (they now make nearly 2M), whether SpaceX could create reusable rockets (they have, and use them continually), and whether SpaceX could have the rockets land on moving platforms in the ocean. All these impossibilities fell by the wayside through time. Sure, most of the breakthroughs came later than expected (not the Tesla manufacturing volume btw), but they came nonetheless.
I find it odd that some seem to be hoping for failure, enjoy it when others can't achieve things. I'm an optimist so it's hard for me to understand that mind-set. I hope for success, as it is better for humanity in the long run that these moonshots / aspirational and "crazy" ideas are actually attempted, and many times achieved. People likely thought the Wright brothers were idiots running in a field. Where would the world be without them, not launching rockets into outer space surely?
Last edited:
MisterFalcon
Member
If only there was something other than land on this planet. Maybe even some place with an abundant supply of liquids to cool it with.
There is indeed a lot of pushback on data centers which can be put anywhere but want to be close to the big population and productivity centers to have minimum latency, something that will affect an orbital data cannot give due to basic physics.
It's also a made up thing to compare with. The world is effectively one huge data center with everything connected together like Musk's constellation will not be one giant data center station but a mass of little satellites.
Dirk Benedict
Member
Waiting on the IPO. Unfortunately.. fucking France... is kind of throwing shit into the mix.