In an unlikely alliance, Elon Musk's SpaceX has partnered with Anthropic, the AI safety company Musk publicly called 'evil' just months ago.
The deal gives Anthropic control of the Colossus 1 data center in Memphis, unlocking over 300 megawatts and 220,000 Nvidia GPUs for its Claude models.
Anthropic was founded by ex-OpenAI researchers prioritizing safe, constitutional AI. In contrast, Musk's Grok was built as an anti-woke chatbot integrated with X. While Anthropic reached $30 billion annual revenue run rate, xAI lost $1.46 billion in a quarter.
Musk said SpaceXAI can reclaim compute if Anthropic causes 'harm to humanity,' a vague clause not in the official contract.