[chatmasta]

Datacenter operators who rent space are selling electricity. SpaceX is selling a fully built datacenter with compute designed for a specific purpose. They’re operating at a higher level of the value chain and can charge accordingly.

[SecretDreams]

What's their novelty or moat to maintain the value chain? And why do we only see google, who already owns it, raising their hand to rent at these prices?

[chatmasta]

I’m not sure they need novelty or moat. AI compute resources are so scarce that inference providers will buy whatever is available. SpaceX sells inference hardware in bulk, with a proven track record of running inference and training workloads at scale.

[mlyle]

Without a moat, P settles to MC. No one makes significant profit.

[chatmasta]

xAI covers their cost of N-1 datacenter while running their own models in N and building out N+1.

[SecretDreams]

And they make all of their money from the N-1 data center they are renting which is sand moat.

What point are you making?

[chatmasta]

What? They make money from their own inference and models too, which they can train effectively for free by funding their operations with rental income from their last gen datacenter.

[Brybry]

Anthropic is also paying $1.25 billion a month for xAI datacenter compute (though Google does own ~14%? of Anthropic too).

[1] https://www.businessinsider.com/spacex-ipo-anthropic-paying-...

[2] https://www.nytimes.com/2025/03/11/technology/google-investm...

[SecretDreams]

I'm not a big fan of this level of circular financing and ownership. The transparency is severely obscured.

[ralph84]

SpaceX and Tesla used aggressive vertical integration, manufacturing simplification, and reuse to radically lower the cost of building rockets and EVs. It's not unreasonable to speculate they might be able to do the same for hyperscale compute.