Particle.news
Download on the App Store

Musk Calls Anthropic the AI Leader After Signing Large SpaceX Compute Deal

His public reversal highlights a new commercial tie that concentrates Anthropic’s model training on SpaceX’s Colossus 1 and raises security and market risks.

Overview

  • On Thursday, July 9, Elon Musk said he was “clearly wrong” about Anthropic, called its Mythos and Fable models unmatched by rivals, and predicted a Mythos 2 release would arrive soon.
  • Anthropic in May agreed to lease the full Colossus 1 capacity at SpaceX/xAI—reported as roughly 300 megawatts or more than 220,000 Nvidia GPUs—with a SpaceX S‑1 filing showing payments of $1.25 billion per month through May 2029.
  • The U.S. Commerce Department recently lifted temporary export‑control restrictions that had limited access to Anthropic’s newest models after a federal review of the company’s technology.
  • Analysts warn that hosting Anthropic’s compute on SpaceX systems creates commercial leverage and greater visibility into Anthropic’s operations and raises risks that a host could 'distill' competitor models, though contracts and revenue make an abrupt cutoff unlikely.
  • Anthropic’s reported $65 billion fundraising, near‑$1 trillion valuation and a confidential IPO filing heighten the economic stakes and mean future regulatory, operational, or contractual disruptions could have wide market and security consequences.