After users questioned whether Musk could pull the plug on Anthropic's infrastructure, he publicly praised the lab and pledged not to cut off a paying customer worth $40 billion to SpaceX.
Elon Musk publicly defended Anthropic’s decision to host its compute on SpaceX infrastructure on Thursday, calling the AI lab “obviously currently the leader in AI” and promising he would never weaponize that access against a competitor [1].
The remarks matter because the stakes are unusually high: Anthropic signed a deal in May 2026 to purchase 300 megawatts of compute — the full output of xAI’s Colossus 1 data center near Memphis, Tennessee — at $1.25 billion per month through May 2029, a contract worth roughly $40 billion in total revenue for SpaceX’s xAI unit [1]. Google also signed a separate deal to rent SpaceX infrastructure through June 2029 for $920 million per month [1].
What Prompted the Post
Users on X raised the scenario that Musk could simply boot Anthropic from his servers to kneecap a rival, according to TechCrunch’s reporting [1]. Musk pushed back directly, writing that such a move was “not my style” [1].
He also walked back earlier skepticism about the company. “I was clearly wrong about Anthropic,” Musk wrote, referencing a September 2025 post in which he had said, “Winning was never in the set of possible outcomes for Anthropic” [1]. Even at that earlier date, Anthropic was reported to hold the largest AI market share among enterprise customers [1].
On the model side, Musk was effusive: “No company has released a model as good as Mythos/Fable and they will undoubtedly have Mythos 2 ready soon,” he wrote [1]. Mythos and Fable are Anthropic model names referenced in his post [1].
Why Anthropic Isn’t Solely Relying on Musk’s Word
TechCrunch notes that contractual protections and the sheer financial incentive for SpaceX give Anthropic structural leverage beyond Musk’s stated goodwill [1]. SpaceX engineers supporting Anthropic’s infrastructure may also gain operational expertise similar to what Amazon’s engineers have developed through their own close relationship with the lab [1].
There is, however, a subtler risk. During his trial against OpenAI, Musk acknowledged that AI “distilling” — a practice in which one model maker sends large volumes of prompts to a competitor’s model to learn how it works — is real and widespread among AI companies [1]. Anthropic itself accused three Chinese model makers of distilling its Claude models in February 2026 [1]. Hosting Anthropic’s compute could give SpaceX greater visibility into the company’s operations than most rivals would ever have, TechCrunch reported [1].
Musk cited Tesla’s 2014 patent pledge and its decision to open the Supercharger network and charging port design to competitors as evidence of his non-aggressive competitive posture, and added that “SpaceX launches competing satellite systems with no increase in price or use of unfair terms” [1]. He has, nonetheless, sued OpenAI [1].
The Anthropic deal runs through May 2029 [1].
Sources
This article was drafted with AI from the cited sources and checked against them before publication. Spot an error? Let us know.
