SpaceX's IPO filing discloses a $1.25 billion monthly compute deal with Anthropic for Colossus clusters through May 2029
Capacity ramps in May 2026 for AI training including Grok 5.
Heh
SpaceXAI will provide @AnthropicAI with access to Colossus 1, one of the world’s largest and fastest-deployed AI supercomputers, to provide additional capacity for Claude → http://x.ai/news/anthropic-compute-partnership
@rpoo @fouriergalois Congats, Ross
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
@pitdesi Until 2029? On both Colossus? So it's over, SpaceXAI won't train anything big for a while? Even though they just said they'll buy Cursor?? I'm just overall so confused!
Anthropic is paying $1.25B a month to SpaceX for compute
@_sholtodouglas Claude in space wauw
More compute -> straight to you
@daniel_271828 I think you have no idea what’s going on man. what this shows is that everyone is more in thrall to technocapital optimization ghosts than to any hint of something resembling human values or politics. good for acceleration and bad for you
Happy to see this positivity between Elon and Anthropic! This sort of collaborative, win-win dynamic seems more likely to usher in a good future than adversarial or winner-take-all dynamics
@rpoo Congrats, Ross!
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
This week made something clear: you shouldn't take what most tech ceos are saying publicly seriously!
From “Anthropic is Misanthropic” to “Claude is good for humanity and was impressed.” Most ironic outcome is most likely.
@chamath 🎯
Called it!
@Mononofu @SpaceX 🚀
Very excited to be partnering with @elonmusk @SpaceX! Visionary engineering + Claude is going to be awesome, scaling is continuing for a long time!
The GB300 is the best AI computer
Two frontier labs. One accelerated computing platform. Congrats to @SpaceX and @AnthropicAI on the new compute partnership, powered by 220,000+ NVIDIA GPUs inside Colossus 1. The future of AI runs on NVIDIA.
Same here.
By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed.
Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good.
After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
In the next few days we'll be ramping up Claude inference on Colossus. Grateful to be partnering with SpaceX here. We are going to need to move a lot of atoms in order to keep up with AI demand, and there's nobody better at quickly moving atoms (on or off planet Earth)
@_sholtodouglas 🚀
https://x.ai/news/anthropic-compute-partnership Incredibly excited to partner with SpaceXAI on Colossus 1 - and looking forward to scaling to many terawatts in orbit :)
As the recently expanded partnership with @AnthropicAI demonstrates, @SpaceX is offering AI compute as a service at significant scale.
We are in discussions with other companies to do the same.
Over time, especially with orbital data centers, we expect to serve AI at extremely high scale.
Compute finds a way to goto good models.
“This gives us access to more than 300 megawatts of new capacity (over 220,000 NVIDIA GPUs) within the month.”
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
Presumably Blackwells, thats a lot.
Compute finds a way to goto good models. “This gives us access to more than 300 megawatts of new capacity (over 220,000 NVIDIA GPUs) within the month.”
So basically grok training was utilising 11% (I'm told) of compute availability, so as a face saving move they sell off compute... What was SpaceX aquiring xAI even for? A massive value pump? Value extraction for X and Xai equity holders by shafting SpaceX investors?
SpaceXAI will provide @AnthropicAI with access to Colossus 1, one of the world’s largest and fastest-deployed AI supercomputers, to provide additional capacity for Claude → http://x.ai/news/anthropic-compute-partnership
More compute -> straight to you
Effective today, we are: 1) Doubling Claude Code’s 5-hour rate limits for Pro, Max, and Team plans; 2) Removing the peak hours limit reduction on Claude Code for Pro and Max plans; and 3) Substantially raising our API rate limits for Opus models.
Incredibly excited to partner with SpaceXAI on Colossus 1 - and looking forward to scaling to many terawatts in orbit :)
Never has the 🚀 emoji felt more apt.
In the next few days we'll be ramping up Claude inference on Colossus. Grateful to be partnering with SpaceX here. We are going to need to move a lot of atoms in order to keep up with AI demand, and there's nobody better at quickly moving atoms (on or off planet Earth)
I can see why Anthropic felt they have no choice but partner with Elon, but I wouldn't have been so enthusiastic about it.
Never has the 🚀 emoji felt more apt.
Hot take on Elon’s surprise decision to rent 30 megawatts of compute to Anthropic:
1. It’s a tacit concession that xAI is not all that close to AGI (despite what he suggested last year).
2. It’s more evidence that pure scaling doesn’t get you to AGI.
3. If xAI turns out to be just another cloud company, in an increasingly crowded market, SpaceX investors very likely overpaid.
4. The very fact that xAI *has* an enough of oversupply to sell some to a big competitors doesn’t bode well for hyperscalers. We may very quickly find ourselves in a world where supply outstrips demands and prices drop precipitously.
5. Customers win but …
6. Companies like Amazon, Google, Meta, Microsoft and Oracle could all take a hit, especially given what they have done to their cash flows to build up their cloud compute resources.
*apologies for typo: 300 not 30 megawatts
Hot take on Elon’s surprise decision to rent 30 megawatts of compute to Anthropic: 1. It’s a tacit concession that xAI is not all that close to AGI (despite what he suggested last year). 2. It’s more evidence that pure scaling doesn’t get you to AGI. 3. If xAI turns out to be just another cloud company, in an increasingly crowded market, SpaceX investors very likely overpaid. 4. The very fact that xAI *has* an enough of oversupply to sell some to a big competitors doesn’t bode well for hyperscalers. We may very quickly find ourselves in a world where supply outstrips demands and prices drop precipitously. 5. Customers win but … 6. Companies like Amazon, Google, Meta, Microsoft and Oracle could all take a hit, especially given what they have done to their cash flows to build up their cloud compute resources.
If point #4 is true, things are gonna get really wild.
Hot take on Elon’s surprise decision to rent 30 megawatts of compute to Anthropic: 1. It’s a tacit concession that xAI is not all that close to AGI (despite what he suggested last year). 2. It’s more evidence that pure scaling doesn’t get you to AGI. 3. If xAI turns out to be just another cloud company, in an increasingly crowded market, SpaceX investors very likely overpaid. 4. The very fact that xAI *has* an enough of oversupply to sell some to a big competitors doesn’t bode well for hyperscalers. We may very quickly find ourselves in a world where supply outstrips demands and prices drop precipitously. 5. Customers win but … 6. Companies like Amazon, Google, Meta, Microsoft and Oracle could all take a hit, especially given what they have done to their cash flows to build up their cloud compute resources.
This likely costs about ~$500m a month, ~$6bn a year to rent blended
Which is about the run rate net loss of xai end of q1
Anthropics revenue run rate was $9bn end of 2025, $30bn a month ago
I usually avoid commenting too much on industry deals, but this one is fascinating. Certainly seems like a blow to the idea that Grok will remain a frontier model.
@rpoo congrats ross!
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
Predicting future is hard (for everyone)!
I'd be great to have a maintained tracker to evaluate: specificity vs sensitivity of predictions, especially from leaders 😇
OG post:

Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
@rpoo congrats ross! major steal for anthropic
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
They've done it
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
“Compute as a service.”
Wait I thought @SpaceX was a rocket company.
But last night I put up a video from @theresidency’s San Francisco startup demo day.
First company on the video is putting NVIDIA datacenters up in space.
“How you getting them up there?” I asked its founder.
“SpaceX.”
“How you connecting them to humans?”
“SpaceX.”
Wait a second. You know how Uber is a transportation as a service company but owns no vehicles?
SpaceX could rent out compute in space without owning any datacenters.
Highly profitable.
And, of course, Elon will have his own too.
As the recently expanded partnership with @AnthropicAI demonstrates, @SpaceX is offering AI compute as a service at significant scale. We are in discussions with other companies to do the same. Over time, especially with orbital data centers, we expect to serve AI at extremely high scale.
The video:
San Francisco is a series of startup factories. It is demo day season. There are several demo days the next week. Will be at @fdotinc on Friday. Tonight @theresidency held theirs with 10 new startups that just spent a month working on their companies. You might recognize the last one. It is @blevlabs who made the AI that I use to make https://alignednews.com/ai (I was first to discover him about a year ago. Yet another company launched in my home. The factories, including YCombinator, @hf0, and a few others, help founders accelerate their companies and set them up to present to investors. Most of the people in the audience are investors. I am blessed to get invited to record from the front row. Each company gets about 90 seconds to give their pitch.
Turns out you can just buy more compute
SpaceXAI will provide @AnthropicAI with access to Colossus 1, one of the world’s largest and fastest-deployed AI supercomputers, to provide additional capacity for Claude → http://x.ai/news/anthropic-compute-partnership
🎯 But also congrats to both and very exciting that there’s more compute used for generating more value in the world!
From “Anthropic is Misanthropic” to “Claude is good for humanity and was impressed.” Most ironic outcome is most likely.
@kyliebytes
SpaceX, the neocloud?
SpaceX, the neocloud?

We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
TechCrunch stole my joke 😭

SpaceX, the neocloud?
Very excited to be partnering with @elonmusk @SpaceX! Visionary engineering + Claude is going to be awesome, scaling is continuing for a long time!
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
@rpoo congrats to me, welcome
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
@rpoo @kipperrii Really excited to work with you, Ross!
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
We’re expanding our partnership with @SpaceX, and will be scaling up on GB200 capacity in Colossus 2 throughout June.
Appreciate @elonmusk and the team helping us find good homes for the Claudes.
In the next few days we'll be ramping up Claude inference on Colossus. Grateful to be partnering with SpaceX here. We are going to need to move a lot of atoms in order to keep up with AI demand, and there's nobody better at quickly moving atoms (on or off planet Earth)
We’re expanding our partnership with SpaceX, and will be scaling up on GB200 capacity in Colossus 2 throughout June.
Appreciate @elonmusk and the team helping us find good homes for the Claudes.
In the next few days we'll be ramping up Claude inference on Colossus. Grateful to be partnering with SpaceX here. We are going to need to move a lot of atoms in order to keep up with AI demand, and there's nobody better at quickly moving atoms (on or off planet Earth)
In the next few days we'll be ramping up Claude inference on Colossus.
Grateful to be partnering with SpaceX here. We are going to need to move a lot of atoms in order to keep up with AI demand, and there's nobody better at quickly moving atoms (on or off planet Earth)
In the next few days we'll be ramping up Claude inference on Colossus. Grateful to be partnering with SpaceX here. We are going to need to move a lot of atoms in order to keep up with AI demand, and there's nobody better at quickly moving atoms (on or off planet Earth)
Let the tokens flow
Effective today, we are: 1) Doubling Claude Code’s 5-hour rate limits for Pro, Max, and Team plans; 2) Removing the peak hours limit reduction on Claude Code for Pro and Max plans; and 3) Substantially raising our API rate limits for Opus models.
But, but… I thought they were morally depraved purveyors of Woke AI
(Jk; capital in a functioning market will be allocated to its highest and best use, but I do encourage you to remember all the people with supposedly principled opposition to ant who look foolish now)
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
@AnjneyMidha Perfect example of capitalism being pro-social
Mission alignment is a beautiful thing
@deanwball https://x.com/teortaxesTex/status/2052080900280557749 elon is exceptionally cringe here
not even 3 months ago Elon was rabidly chudposting at Amanda Askell now, all water under the bridge! (water goes into Colossus to be annihilated by Anthropic's inference demand) Not evil at all! Highly impressive people!
Elon's rule for life: Money Talks, Bullshit Walks
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
not even 3 months ago Elon was rabidly chudposting at Amanda Askell now, all water under the bridge! (water goes into Colossus to be annihilated by Anthropic's inference demand) Not evil at all! Highly impressive people!

Elon's rule for life: Money Talks, Bullshit Walks
> SpaceXAI Might be the first time I see this abominable name used in the wild
@pitdesi Until 2029? On both Colossus? So it's over, SpaceXAI won't train anything big for a while? Even though they just said they'll buy Cursor?? I'm just overall so confused!
Claude in Space fearsome
SpaceXAI and @AnthropicAI have also expressed interest in partnering to develop multiple gigawatts of orbital AI compute capacity
In many ways America is a perversion of Rome, a LARP of that great and terrible empire. Romans had a principle to NEVER reward the great with a permission slip to be vile; Americans pride themselves on elevating gifted scoundrels. Elon is ur-American in this. And it saddens me.
not even 3 months ago Elon was rabidly chudposting at Amanda Askell now, all water under the bridge! (water goes into Colossus to be annihilated by Anthropic's inference demand) Not evil at all! Highly impressive people!
Anthropic is getting 220,000 Nvidia GPUs from SpaceX and has committed to spending $200 Billion on Google TPUs.
Claude Code’s 5-hour rate limits are doubling.
Dario is finally getting aggressive about chips from every possible source to fight OpenAI Codex’s growth.

From “Anthropic is Misanthropic” to “Claude is good for humanity and was impressed.”
Most ironic outcome is most likely.

Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
More chips, more Claude
Effective today, we are: 1) Doubling Claude Code’s 5-hour rate limits for Pro, Max, and Team plans; 2) Removing the peak hours limit reduction on Claude Code for Pro and Max plans; and 3) Substantially raising our API rate limits for Opus models.
Utilization matters. Compute is limited. Power is getting scarce.
SpaceXAI will provide @AnthropicAI with access to Colossus 1, one of the world’s largest and fastest-deployed AI supercomputers, to provide additional capacity for Claude → http://x.ai/news/anthropic-compute-partnership
Claude may not be going to the moon per se, but a trip to halo-orbit Earth-Moon L1 seems likely in the cards for the 2030s…
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
SpaceXAI and @AnthropicAI have also expressed interest in partnering to develop multiple gigawatts of orbital AI compute capacity
turns out Ricardo’s Law of Comparative Advantage *is* relevant to AI economics, just not in the place people thought
In the next few days we'll be ramping up Claude inference on Colossus. Grateful to be partnering with SpaceX here. We are going to need to move a lot of atoms in order to keep up with AI demand, and there's nobody better at quickly moving atoms (on or off planet Earth)
@ohabryka @daniel_271828 A hypothetical world where an incorruptible IAEA-like institution ends the race dynamic would sure be preferable, but since that no longer seems feasible, imo Elon and Anthropic teaming up to race faster against others does seem good on current margins.
@daniel_271828 I do think it’s a good sign of an easing of race dynamics, though Elon and Anthropic teaming up to race faster against others is bad.
@scaling01 I wanted to post this stuff, lol, it's all in the prospectus, every time I tried to read it something else happened. Big plans for the data satellites!
From the SpaceX IPO filing: Anthropic is paying SpaceX $1.25B a month untill May 2029 for the COLOSSUS 1/2 compute
Confirmed by Anthropic.
Confirmed by Anthropic.

Anthropic and SpaceX have signed a compute sharing agreement to give Anthropic access to Colossus for training.
Confirmed by Anthropic.
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
Claude is going to space!
SpaceXAI and @AnthropicAI have also expressed interest in partnering to develop multiple gigawatts of orbital AI compute capacity
In the next few days we'll be ramping up Claude inference on Colossus. Grateful to be partnering with SpaceX here. We are going to need to move a lot of atoms in order to keep up with AI demand, and there's nobody better at quickly moving atoms (on or off planet Earth)
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
https://x.ai/news/anthropic-compute-partnership Incredibly excited to partner with SpaceXAI on Colossus 1 - and looking forward to scaling to many terawatts in orbit :)
Elon has made his bet on who he thinks is going to win this race. But there's another Kingmaker who could also share his power: Mark Zuckerberg.

Elon: 'By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed.
Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good.'
Elon has made his bet on who he thinks is going to win this race. But there's another Kingmaker who could also share his power: Mark Zuckerberg.
xAI is being dissolved.
@SawyerMerritt xAI will be dissolved as a separate company, so it will just be SpaceXAI, the AI products from SpaceX
Do you feel like things are speeding up?
We’re expanding our partnership with @SpaceX, and will be scaling up on GB200 capacity in Colossus 2 throughout June. Appreciate @elonmusk and the team helping us find good homes for the Claudes.
Anthropic is paying $1.25B a month to SpaceX for compute
@zephyr_z9
Elon is in discussions with other companies besides Anthropic to sell compute as a service.
Elon is in discussions with other companies besides Anthropic to sell compute as a service.
As the recently expanded partnership with @AnthropicAI demonstrates, @SpaceX is offering AI compute as a service at significant scale. We are in discussions with other companies to do the same. Over time, especially with orbital data centers, we expect to serve AI at extremely high scale.

Elon is in discussions with other companies besides Anthropic to sell compute as a service.
Mission alignment is a beautiful thing
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
@rpoo 👀
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
@rpoo Congrats @rpoo!
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
SpaceX as a vertically integrated AI compute company makes an insane amount of sense.
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
@Mononofu @elonmusk @SpaceX Congrats! Good outcome.
Very excited to be partnering with @elonmusk @SpaceX! Visionary engineering + Claude is going to be awesome, scaling is continuing for a long time!
@rpoo Congrats!
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
merge and assist? in this economy?
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
👀👀
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
@elonmusk @seanmcdonaldxyz @nottombrown 👀👀
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
It was always clear that Anthropic would get compute if they needed it. Demand finds supply.

We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
@elonmusk @nottombrown Great news
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
Not so "mis"Anthropic any more it seems.
SpaceXAI will provide @AnthropicAI with access to Colossus 1, one of the world’s largest and fastest-deployed AI supercomputers, to provide additional capacity for Claude → http://x.ai/news/anthropic-compute-partnership
@rpoo @_sholtodouglas Congrats!
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
some interesting numbers from the SpaceX IPO filing:
musk stock conditions: > 302.1M performance-based Class B shares > vesting requires both market cap milestones ($1.065T to $6.565T) and SpaceX completing non-Earth data centers capable of 100 terawatts of compute per year
anthropic compute deal: > $1.25B/month through May 2029 for capacity across Colossus + Colossus II, with reduced fees during May/June 2026 ramp. either party can terminate on 90 days notice.
colossus: > Colossus 1: ~100k H100s, ~130 MW, online in 122 days > Colossus II first cluster: ~110k GB200s, ~210 MW, online in 91 days > Colossus II second cluster: ~110k GB300s, ~220 MW, online in 64 days > next expansion: at least 220k additional GB300s, 400+ MW
AI capex: > 2023: $463M > 2024: $5.6B > 2025: $12.7B > Q1 2026: $7.7B
AI revenue/loss: > 2025 revenue: $3.2B, operating loss: $6.4B > Q1 2026 revenue: $818M, operating loss: $2.5B
compute lease: > SpaceX/xAI renting a lot of compute hardware from Valor, run by SpaceX board member Antonio Gracias. total payments over time: ~$20.2B
infra: > April 2026 turbine deal: ~$2.0B for mobile gas turbines to power data centers > additional turbine commitments: $925M through 2029 > servers + networking on balance sheet: $23.85B > data center infrastructure: $2.965B > construction-in-progress: $14.045B > Terafab target with Tesla/Intel: one terawatt of compute hardware per year
SpaceX's IPO prospectus (S-1 filing) is now officially public! You can read the full document here: https://www.sec.gov/Archives/edgar/data/1181412/000162828026036936/spaceexplorationtechnologi.htm
> SpaceX completing non-Earth data centers capable of 100 terawatts of compute per year
that's totally insane btw
some interesting numbers from the SpaceX IPO filing: musk stock conditions: > 302.1M performance-based Class B shares > vesting requires both market cap milestones ($1.065T to $6.565T) and SpaceX completing non-Earth data centers capable of 100 terawatts of compute per year anthropic compute deal: > $1.25B/month through May 2029 for capacity across Colossus + Colossus II, with reduced fees during May/June 2026 ramp. either party can terminate on 90 days notice. colossus: > Colossus 1: ~100k H100s, ~130 MW, online in 122 days > Colossus II first cluster: ~110k GB200s, ~210 MW, online in 91 days > Colossus II second cluster: ~110k GB300s, ~220 MW, online in 64 days > next expansion: at least 220k additional GB300s, 400+ MW AI capex: > 2023: $463M > 2024: $5.6B > 2025: $12.7B > Q1 2026: $7.7B AI revenue/loss: > 2025 revenue: $3.2B, operating loss: $6.4B > Q1 2026 revenue: $818M, operating loss: $2.5B compute lease: > SpaceX/xAI renting a lot of compute hardware from Valor, run by SpaceX board member Antonio Gracias. total payments over time: ~$20.2B infra: > April 2026 turbine deal: ~$2.0B for mobile gas turbines to power data centers > additional turbine commitments: $925M through 2029 > servers + networking on balance sheet: $23.85B > data center infrastructure: $2.965B > construction-in-progress: $14.045B > Terafab target with Tesla/Intel: one terawatt of compute hardware per year
Called it!
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
@elonmusk Agreed…and VR should be even better (and cheaper)
The GB300 is the best AI computer
There is no mandate of God. Nobody knows who will win. It’s going to be fun.
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
More compute! Higher usage limits!
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
What do Anthropic people think?
I’d love to know the internal discussions on this 🤔
not even 3 months ago Elon was rabidly chudposting at Amanda Askell now, all water under the bridge! (water goes into Colossus to be annihilated by Anthropic's inference demand) Not evil at all! Highly impressive people!
Free markets are the ultimate tool for peace between tribes.
Based move from all parties involved. The acceleration continues!
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
xAI e/acc infra engineers in the trenches with the Anthropic EA MTS's

We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
@rpoo @_sholtodouglas Wow!
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
And there it is.
xAI entering the neocloud era.
Alliance between the kingdoms fighting OpenAI
The Game of AI Thrones intensifies...
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
If you think your rent in SF is expensive, just wait until you hear how much Claude is paying in Memphis
Anthropic is paying $1.25B a month to SpaceX for compute
@_sholtodouglas @zacharyhorn SICK 🔥🚀
https://x.ai/news/anthropic-compute-partnership Incredibly excited to partner with SpaceXAI on Colossus 1 - and looking forward to scaling to many terawatts in orbit :)
The age of the Cloud is over.
The age of the compute Halo has begun.
As the recently expanded partnership with @AnthropicAI demonstrates, @SpaceX is offering AI compute as a service at significant scale. We are in discussions with other companies to do the same. Over time, especially with orbital data centers, we expect to serve AI at extremely high scale.
Unfathomably based and accelerated move by all parties involved!!
GPU-maxxing
We’re expanding our partnership with @SpaceX, and will be scaling up on GB200 capacity in Colossus 2 throughout June. Appreciate @elonmusk and the team helping us find good homes for the Claudes.
@elonmusk @AnthropicAI @SpaceX Can you buy a compute marketplace company like @PrimeIntellect and offer more GPUs on the open market
As the recently expanded partnership with @AnthropicAI demonstrates, @SpaceX is offering AI compute as a service at significant scale. We are in discussions with other companies to do the same. Over time, especially with orbital data centers, we expect to serve AI at extremely high scale.
@elonmusk @Scobleizer @nottombrown Wow! This is huge!
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
My two favorite AI companies working together!
In the next few days we'll be ramping up Claude inference on Colossus. Grateful to be partnering with SpaceX here. We are going to need to move a lot of atoms in order to keep up with AI demand, and there's nobody better at quickly moving atoms (on or off planet Earth)
We're winding back our peak hours limit reduction and doubling 5 hour limits.
Excited to partner with SpaceX to bring you more compute and we'll keep pushing to bring you the best coding agent in the world.
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
just hit me: what happens with the cursor deal?
Proof that securing compute is not necessarily a competitive advantage at the frontier. Every hyperscaler will want to work with the top lab.
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
post aged better than some answers…
just hit me: what happens with the cursor deal?
colossus 2 immediately secured: why would anthropic only pay for the previous gen cluster and let xai train competitors on the good one.

post aged better than some answers…
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars
Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute
Onward towards building a future we're all excited about
SpaceXAI will provide @AnthropicAI with access to Colossus 1, one of the world’s largest and fastest-deployed AI supercomputers, to provide additional capacity for Claude → http://x.ai/news/anthropic-compute-partnership
From the SpaceX IPO filing:
Anthropic is paying SpaceX $1.25B a month untill May 2029 for the COLOSSUS 1/2 compute

spacex? the cloud company?

As the recently expanded partnership with @AnthropicAI demonstrates, @SpaceX is offering AI compute as a service at significant scale. We are in discussions with other companies to do the same. Over time, especially with orbital data centers, we expect to serve AI at extremely high scale.
@chamath Bink!
Called it!
EWS: Elon Web Services
Colossus + Spaceolossus = Money printing machine
SpaceXAI will provide @AnthropicAI with access to Colossus 1, one of the world’s largest and fastest-deployed AI supercomputers, to provide additional capacity for Claude → http://x.ai/news/anthropic-compute-partnership
@rpoo Dammit ross, after all these years
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
@elonmusk @AnthropicAI @SpaceX would you ever sell compute to openai?
As the recently expanded partnership with @AnthropicAI demonstrates, @SpaceX is offering AI compute as a service at significant scale. We are in discussions with other companies to do the same. Over time, especially with orbital data centers, we expect to serve AI at extremely high scale.
That’s like 3/4 of a Cursor
@nottombrown @SpaceX @elonmusk 🚀
We’re expanding our partnership with @SpaceX, and will be scaling up on GB200 capacity in Colossus 2 throughout June. Appreciate @elonmusk and the team helping us find good homes for the Claudes.
People are generally bad at thinking nonlinearly
I forecast people will be very surprised by the growth that’s about to come
👀 🚀
As the recently expanded partnership with @AnthropicAI demonstrates, @SpaceX is offering AI compute as a service at significant scale. We are in discussions with other companies to do the same. Over time, especially with orbital data centers, we expect to serve AI at extremely high scale.
🚀
We’ve agreed to a partnership with @SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API.
@rpoo @_sholtodouglas 🚀
Incredibly proud the xAI team's execution in bringing on an immense amount of compute so quickly & couldn't be more excited for Elon's vision of an abundant future w/ intelligence spreading to the stars Over the past weeks I've had the pleasure of getting to know the team at Anthropic & share similar vision - I'm happy to announce I'm joining the team there to focus on compute Onward towards building a future we're all excited about
@_sholtodouglas Let's gooo!!!!
https://x.ai/news/anthropic-compute-partnership Incredibly excited to partner with SpaceXAI on Colossus 1 - and looking forward to scaling to many terawatts in orbit :)
This is the most ambitious crossover in AI history
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
@animesh_garg Folks will soon start betting on them 🤪
Predicting future is hard (for everyone)! I'd be great to have a maintained tracker to evaluate: specificity vs sensitivity of predictions, especially from leaders 😇 OG post:
@deanwball That's a principled observation, but if you want America to do well the pragmatist's answer is to let them all climb down gracefully
But, but… I thought they were morally depraved purveyors of Woke AI (Jk; capital in a functioning market will be allocated to its highest and best use, but I do encourage you to remember all the people with supposedly principled opposition to ant who look foolish now)
@beffjezos Looks like a surrender to me.
Free markets are the ultimate tool for peace between tribes. Based move from all parties involved. The acceleration continues!
Holy: After Anthropic secured compute capacity from Colossus 1, it is now also getting access to compute from Colossus 2.
But to be honest: I somehow expected to his. Grok certainly doesn’t need all that compute.
We’re expanding our partnership with @SpaceX, and will be scaling up on GB200 capacity in Colossus 2 throughout June. Appreciate @elonmusk and the team helping us find good homes for the Claudes.
Dario&Daniela Amodei Interview: The 80x growth reportedly caught them completely off guard, and that's the reason for the compute constraint.
The compute deal with SpaceX is the first attempt to address the shortage and continues to search for solutions.
As far as I can tell, Colossus 1/2 have major expansion plans xAI said that they want to put over 1M GPUs in those DCs So capacity will be much larger than 1.2GW
@AndrewCurran_ Yeah, $1.25B for 1.2GW Very expensive
@zephyr_z9 They are paying Elon $1.25 billion a month for the Colossus usage.
SpaceX has a $15 Billion/yr compute side business
@nottombrown @SpaceX @elonmusk how much compute do you need exactly. i have a guy.
We’re expanding our partnership with @SpaceX, and will be scaling up on GB200 capacity in Colossus 2 throughout June. Appreciate @elonmusk and the team helping us find good homes for the Claudes.
@elonmusk @AnthropicAI @SpaceX Gavin (@GavinSBaker) on SpaceX's orbital "data centers":
"I've spent a lot of time at Starbase, and I've talked to a lot of SpaceX engineers. It is the most talented group of engineers on Earth, and they're very confident they have solved this."
In our last conversation, Gavin said data centers in space will be the most important thing in 3-4 years. He explains that means "racks in space" and thinks orbital compute will solve the watts shortage: "When people hear data centers in space, they picture a Pentagon-sized building in space. That's not what it is. A Blackwell rack weighs 3,000 pounds. It's eight feet high. Four feet deep. Three feet wide. It's racks in space. It has these solar wings that are probably 500 feet long on each side. You keep it in a Sun-synchronous orbit, so those solar panels are always in the sun. And then because it's in an exactly Sun-synchronous orbit, the radiator, which extends behind it for hundreds of feet is in the shade. You link these racks using lasers traveling through vacuum which are already on every Starlink. SpaceX operates the world's largest satellite fleet, which is 98 or 99% of all satellites in orbit. Every Starlink, they're cooling it today. I think Starlink V3 is going to operate at 20 kilowatts. A Blackwell rack is only 100 kilowatts. And people talk a lot about density. Well, if you're connecting the racks with lasers through vacuum, you can make the rack bigger physically. In space, there's all sorts of things that SpaceX can do. They also now operate the largest data center on Earth. I've spent a lot of time at Starbase over the years, and I've talked to a lot of SpaceX engineers. It is the most talented group of engineers on planet Earth, and they're very confident they have solved this."
Is this… good?
This makes things go faster. The labs collaborating on sharing computer does not make things go better if they don’t actually make the systems safe and I think Elon is just terribly wrong about his risk assessment (AI will not be evil because the people working on it are evil).
Happy to see this positivity between Elon and Anthropic! This sort of collaborative, win-win dynamic seems more likely to usher in a good future than adversarial or winner-take-all dynamics
@daniel_271828 I do think it’s a good sign of an easing of race dynamics, though Elon and Anthropic teaming up to race faster against others is bad.
Is this… good? This makes things go faster. The labs collaborating on sharing computer does not make things go better if they don’t actually make the systems safe and I think Elon is just terribly wrong about his risk assessment (AI will not be evil because the people working on it are evil).
Happy to see this positivity between Elon and Anthropic! This sort of collaborative, win-win dynamic seems more likely to usher in a good future than adversarial or winner-take-all dynamics
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
@tszzl “Things move faster as powerful factions aren’t at each others’ throats in a match to the death but instead on the same broad team as interests are somewhat aligned” strikes me as safer than the inverse
@daniel_271828 I think you have no idea what’s going on man. what this shows is that everyone is more in thrall to technocapital optimization ghosts than to any hint of something resembling human values or politics. good for acceleration and bad for you
@elonmusk @nottombrown I agree that folks at Anthropic are well-intentioned and trying hard. But alignment looks hard, and I expect them to fail regardless of intent. You don't get what you train for, not even if you're kind.
Same here. By way of background for those who care, I spent a lot of time last week with senior members of the Anthropic team to understand what they do to ensure Claude is good for humanity and was impressed. Everyone I met was highly competent and cared a great deal about doing the right thing. No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good. After that, I was ok leasing Colossus 1 to Anthropic, as SpaceXAI had already moved training to Colossus 2.
Gotta say it’s a bit weird watching all the Anthropic people cheerfully announcing their partnership with the guy who killed the most people this millennium.

We’re expanding our partnership with @SpaceX, and will be scaling up on GB200 capacity in Colossus 2 throughout June. Appreciate @elonmusk and the team helping us find good homes for the Claudes.
I get all the arguments for why this partnership is counterfactually good, but, my god, is the mood missing.
And what anthropic is trying to do with that compute is render us obsolete — something wildly dangerous and democratically illegitimate (more on that in bio).
Gotta say it’s a bit weird watching all the Anthropic people cheerfully announcing their partnership with the guy who killed the most people this millennium.



