Elon Musk Praises Anthropic’s Mythos/Fable Amid Massive Compute Deal
Elon Musk has officially pivoted from skeptic to supporter regarding Anthropic, praising their cutting-edge model performance while reassuring the industry that he will not weaponize infrastructure against the AI lab. This unexpected endorsement comes as Anthropic becomes one of SpaceX’s most significant revenue drivers through a massive multi-year compute agreement.
A Massive Financial Anchor for SpaceX and xAI
The relationship between Musk’s empire and Anthropic has evolved from ideological friction to a massive commercial alliance. Following the February merger of xAI and SpaceX, Anthropic secured a landmark deal in May to purchase 300 megawatts of compute. This capacity represents the entire output of xAI’s Colossus 1 data center located near Memphis, Tennessee.
The financial scale of this partnership is staggering. Anthropic has committed to paying $1.25 billion per month through May 2029, a total deal value of approximately $40 billion for SpaceX’s xAI unit. This infrastructure-as-a-service model is also being utilized by Google, which signed a separate deal to rent SpaceX infrastructure for $920 million per month through June 2029.
Technical Admiration for Mythos and Fable
Musk’s recent commentary on X marks a sharp reversal from his September 2025 stance, where he doubted Anthropic's ability to compete. He now identifies Anthropic as the current leader in the AI landscape, specifically highlighting the technical superiority of their latest models.
"No company has released a model as good as Mythos/Fable and they will undoubtedly have Mythos 2 ready soon," Musk stated. By acknowledging the dominance of the Mythos/Fable architecture, Musk has signaled that even his own xAI ventures recognize Anthropic's current lead in model capabilities and enterprise market share.
The "Style" of Competition and the Risk of Distillation
Addressing concerns that Musk might "cut off" Anthropic to kneecap a rival, Musk cited Tesla’s open-patent pledge and SpaceX’s competitive satellite pricing as evidence of his "style." He insists he would not use his control over compute to inflict undue harm on competitors.
However, the tech community remains cautious due to the inherent risks of hosting a rival's core infrastructure. Beyond the possibility of sudden service termination—which would carry massive contractual penalties—there is the technical risk of "AI distilling." This process involves using a model to interact with a competitor's AI to learn its underlying logic and patterns. While Anthropic and Google likely have safeguards in place, hosting their compute gives SpaceX unprecedented visibility into the operational patterns of the industry's most advanced models.
Why This Partnership Matters for AI Infrastructure
This development highlights a growing trend in the AI industry: the concentration of compute power within a few massive infrastructure providers. As model training requires increasingly gargantuan amounts of electricity and hardware, the line between AI model developers and AI infrastructure providers is blurring. The symbiotic relationship between Anthropic’s software prowess and SpaceX’s hardware capacity sets a new precedent for how the next generation of frontier models will be built and scaled.
Key Takeaways
- Massive Revenue Stream: Anthropic’s $40 billion compute deal through 2029 provides a massive financial windfall for SpaceX’s xAI unit.
- Technical Validation: Elon Musk has publicly endorsed Anthropic’s Mythos/Fable models as the current industry leaders.
- Infrastructure Risks: While contractual protections exist, the proximity of SpaceX to Anthropic’s compute raises questions regarding competitive intelligence and AI distillation.