Elon Musk has rarely publicly admitted that he misjudged Anthropic on the social media platform X. He not only praised them as "clearly the leader in the AI field at the moment," but also explicitly promised not to cut off Anthropic's server services due to competitive reasons, a complete 180-degree turnaround from his September last year statement that "victory was never within Anthropic's possible outcomes."

Artificial Intelligence, Robotic Arm, AI (1)

The statement came amid SpaceX's previous collaboration with Anthropic, where SpaceX rented computing power to the latter. An X user pointed out that Anthropic completely relied on computing power rented from SpaceX AI, and if Musk wanted to destroy Anthropic, he could simply remove its business from the server with one click. In response to this "conspiracy theory"-style question, Musk directly replied that undermining competitors is "not my style" and openly admitted, "I clearly misjudged Anthropic."

The Mythos/Fable model was specifically praised, and Musk predicted the arrival of Mythos 2

Musk's praise for Anthropic was generous. He stated that no company had ever released models as good as Mythos and Fable, and he predicted that Anthropic would undoubtedly be ready for the next-generation Mythos 2 soon. From last September's dismissive attitude to now publicly acknowledging the opponent's superiority, Musk's dramatic change of heart was backed by the hard evidence of Anthropic's models continuously breaking records in multiple benchmark tests over the past few months.

As of July 2026, Anthropic had become one of SpaceX's largest clients. Musk emphasized that even though they were competitors, he would never cut off services in a way that seriously harmed the other party. He also used the practices of his other companies to support his "competitive philosophy": Tesla opened up its patents and made the supercharging network available to all competitors, and SpaceX launched a satellite system that competed with rivals without increasing prices or setting unfair terms.

From "disagreeing verbally" to "accepting the bet," there are also decent matches in Silicon Valley competition

This statement stood out in the current highly tense atmosphere of the AI industry. Over the past year, Musk had engaged in constant verbal battles with competitors such as OpenAI and Anthropic, but this time he chose to ease public concerns about the potential cutoff of computing power by "admitting mistakes and making promises," rather than continuing to attack further.