I was wrong: Elon Musk accepts Anthropic is AI leader, calls Mythos best model ever
Tech trillionaire Elon Musk has accepted that he was completely wrong about Anthropic. After questioning Anthropic for years, Musk has accepted that the Dario Amodei-led startup is now the AI leader.
Elon Musk is not afraid to speak his mind, particularly when it comes to his rivals – just look at his feud with OpenAI’s Sam Altman. But when it comes to Anthropic, the tech trillionaire seems to have made a U-turn, and accepted that he was completely wrong.
On X, a user reposted Elon Musk’s comments from September last year, where the SpaceX chief declared that “Winning was never in the set of possible outcomes for Anthropic.”
Musk, however, gave a candid response as he accepted that Anthropic was really leading the AI race. The SpaceX chief replied, “I was clearly wrong about Anthropic. They are obviously currently the leader in AI.”
In particular, Elon Musk praised Anthropic’s Claude Mythos and Fable models. The two models were previously restricted by the US following concerns over cybersecurity. While Musk’s SpaceXAI recently released Grok 4.5, and OpenAI, too, has launched GPT-5.6, the tech trillionaire believes that Anthropic remains a league ahead.
Musk added, “No company has released a model as good as Mythos/Fable and they will undoubtedly have Mythos 2 ready soon.”
Keep in mind that in February this year, Elon Musk had fired shots at Anthropic, accusing the company of being “misanthropic and evil.” To give you some context, the words anthropic and misanthropic come from the Greek word “anthropos” meaning human being. While anthropic relates to humanity, misanthropic refers to distrust towards humanity.
Can Elon Musk cutoff Anthropic from SpaceX compute?
Since Elon Musk made such comments about Anthropic, a lot has changed. Anthropic is now one of the largest customers of SpaceX. In May, the AI startup Anthropic signed a deal to lease the Colossus 1 supercomputer. The Colossus 1 includes more than 2,20,000 Nvidia GPUs with a capacity of 300 megawatts of compute.
Under the arrangement, Anthropic agreed to pay about $1.25 billion a month through May 2029, in a deal described as worth roughly $40 billion in revenue for SpaceX’s xAI unit. But either side can end the agreement on a 90-day notice.
This has raised questions that Elon Musk may now hold leverage, as he can decide to turn off the compute access at any time. Keep in mind that SpaceXAI (previously xAI) is a direct rival to Anthropic, and the Grok 4.5 model is designed for coding, a field dominated by Anthropic’s Claude Code.
The user wrote, “If Elon wanted to kill Anthropic, he could. Iirc the compute lease was short term, 6 months from May without renewal promise.” But Musk insists that he has no plans to use such a kill-switch. He said, "And I would never cut them off in a way that hurt them badly, even as a competitor. That’s not my style."
Elon Musk then cited examples from his other businesses. He pointed out that Tesla had open-sourced its patents and made the Supercharger network available to competitors, “even though we could have made it a walled garden.” The tech trillionaire also discussed the platform X where he is often criticised. Musk wrote,"Even my worst enemies can attack me on this platform."
The relationship also raises questions about what a host company can learn from supporting a rival’s systems. During Musk’s lawsuit against OpenAI, he acknowledged that AI "distilling" was real, a process described in the inputs as one company using multiple fake accounts to prompt a competitor’s model and learn how it works. When asked whether xAI had ever distilled technology from OpenAI, Musk replied, "Generally AI companies distill other AI companies." Anthropic has previously accused Chinese model makers of doing this to Claude.
Elon Musk is not afraid to speak his mind, particularly when it comes to his rivals – just look at his feud with OpenAI’s Sam Altman. But when it comes to Anthropic, the tech trillionaire seems to have made a U-turn, and accepted that he was completely wrong.
On X, a user reposted Elon Musk’s comments from September last year, where the SpaceX chief declared that “Winning was never in the set of possible outcomes for Anthropic.”
Musk, however, gave a candid response as he accepted that Anthropic was really leading the AI race. The SpaceX chief replied, “I was clearly wrong about Anthropic. They are obviously currently the leader in AI.”
In particular, Elon Musk praised Anthropic’s Claude Mythos and Fable models. The two models were previously restricted by the US following concerns over cybersecurity. While Musk’s SpaceXAI recently released Grok 4.5, and OpenAI, too, has launched GPT-5.6, the tech trillionaire believes that Anthropic remains a league ahead.
Musk added, “No company has released a model as good as Mythos/Fable and they will undoubtedly have Mythos 2 ready soon.”
Keep in mind that in February this year, Elon Musk had fired shots at Anthropic, accusing the company of being “misanthropic and evil.” To give you some context, the words anthropic and misanthropic come from the Greek word “anthropos” meaning human being. While anthropic relates to humanity, misanthropic refers to distrust towards humanity.
Can Elon Musk cutoff Anthropic from SpaceX compute?
Since Elon Musk made such comments about Anthropic, a lot has changed. Anthropic is now one of the largest customers of SpaceX. In May, the AI startup Anthropic signed a deal to lease the Colossus 1 supercomputer. The Colossus 1 includes more than 2,20,000 Nvidia GPUs with a capacity of 300 megawatts of compute.
Under the arrangement, Anthropic agreed to pay about $1.25 billion a month through May 2029, in a deal described as worth roughly $40 billion in revenue for SpaceX’s xAI unit. But either side can end the agreement on a 90-day notice.
This has raised questions that Elon Musk may now hold leverage, as he can decide to turn off the compute access at any time. Keep in mind that SpaceXAI (previously xAI) is a direct rival to Anthropic, and the Grok 4.5 model is designed for coding, a field dominated by Anthropic’s Claude Code.
The user wrote, “If Elon wanted to kill Anthropic, he could. Iirc the compute lease was short term, 6 months from May without renewal promise.” But Musk insists that he has no plans to use such a kill-switch. He said, "And I would never cut them off in a way that hurt them badly, even as a competitor. That’s not my style."
Elon Musk then cited examples from his other businesses. He pointed out that Tesla had open-sourced its patents and made the Supercharger network available to competitors, “even though we could have made it a walled garden.” The tech trillionaire also discussed the platform X where he is often criticised. Musk wrote,"Even my worst enemies can attack me on this platform."
The relationship also raises questions about what a host company can learn from supporting a rival’s systems. During Musk’s lawsuit against OpenAI, he acknowledged that AI "distilling" was real, a process described in the inputs as one company using multiple fake accounts to prompt a competitor’s model and learn how it works. When asked whether xAI had ever distilled technology from OpenAI, Musk replied, "Generally AI companies distill other AI companies." Anthropic has previously accused Chinese model makers of doing this to Claude.
