Elon Musk praises Mythos/Fable, promises not to ‘cut off’ Anthropic

Ought to Anthropic belief Elon Musk to host its fashions? After customers on X implied that Musk might get up sooner or later and easily boot the AI lab from SpaceX’s servers as a technique to kneecap a rival, Musk replied with glowing reward for the AI lab. He mentioned that such a trick was “not my model.”

“I used to be clearly flawed about Anthropic,” Musk wrote on Thursday, referring to his September 2025 put up on X through which he mentioned, “Profitable was by no means within the set of potential outcomes for Anthropic.” In fact, even at the moment, Anthropic might already be thought of a winner; the corporate was reported to have the largest AI market share with enterprises.

It appears these anti-Anthropic days are behind Musk — and never simply on X. As of July 2026, Anthropic is certainly one of SpaceX’s largest prospects.

To recap: Anthropic signed a deal in Could to purchase 300 megawatts of compute, your complete output of xAI’s Colossus 1 knowledge middle close to Memphis, Tennessee. (Musk’s xAI merged with SpaceX in February.) Anthropic agreed to pay $1.25 billion monthly via Could 2029, a deal price about $40 billion in income for SpaceX’s xAI unit. Google, by the way in which, additionally signed a deal to hire SpaceX infrastructure via June 2029, for $920 million monthly.

Musk insists that this wasn’t a harmful choice by Anthropic and that he’s stuffed with admiration for the rival.

“They’re clearly presently the chief in AI. No firm has launched a mannequin nearly as good as Mythos/Fable and they’ll undoubtedly have Mythos 2 prepared quickly. And I might by no means minimize them off in a means that harm them badly, at the same time as a competitor. That’s not my model,” he wrote.

He provided as proof of his don’t-squeeze-competitors model Tesla’s choice in 2014 (which was outlined in a now deleted firm weblog put up and now housed beneath its patent pledge) to not provoke patent lawsuits in opposition to anybody who, in good religion, desires to make use of its expertise. He additionally famous that Tesla opened its Supercharger community and charging port design to rivals.

“SpaceX launches competing satellite tv for pc techniques with no enhance in value or use of unfair phrases. Even my worst enemies can assault me on this platform,” he wrote, itemizing one other instance.

In fact, Musk will not be precisely above ways geared toward rivals, particularly these with whom he has a historical past. He sued OpenAI, as an example.

Anthropic doesn’t should depend on Musk’s sticking to his “model,” although. There would definitely be contractual penalties if Musk instantly shut down Anthropic’s infrastructure. To not point out the large advantages for SpaceX to maintain that deal intact. Not solely does Anthropic pay handsomely, but in addition SpaceX’s engineers might discover ways to construct for, and assist, Anthropic’s quickly rising AI , identical to Amazon’s engineers do.

That proximity might need different advantages as properly. Throughout his trial in opposition to OpenAI, Musk acknowledged that AI “distilling” was actual — a course of through which one mannequin maker units up many faux accounts to ship prompts to a competitor so as to be taught the way it works. Because the New York Occasions reported, when a lawyer requested him if xAI had ever distilled expertise from OpenAI, Musk replied: “Typically AI corporations distill different AI corporations.”

Anthropic in February accused three Chinese language mannequin makers of doing this to Claude. Presumably, Anthropic and Google really feel they’ve safeguards in opposition to SpaceX doing this whereas they’re utilizing its infrastructure. However internet hosting Anthropic’s compute might nonetheless give SpaceX larger visibility into how the corporate operates than most rivals would ever have.

There seems to be nothing however upside for Musk’s firm on this partnership in the mean time. As for tomorrow, and because the three-year contract ages, who is aware of?