Musk’s Latest AI Twist: The Company He Criticised Is Now Using His Computing Power

Elon Musk’s SpaceXAI has agreed to supply computing power to Anthropic, the developer of the Claude chatbot, despite Musk’s past public criticism of the company. Musk previously called Claude ‘misanthropic,’ ‘evil,’ and ‘a threat to Western civilization’ on X (formerly Twitter), but he reversed course after meeting Anthropic’s senior leadership. He stated that ‘no one set off my evil detector’ following discussions. The deal involves Anthropic using Colossus 1, a Tennessee-based supercomputer operated by SpaceXAI, which runs on over 220,000 NVIDIA GPUs. These GPUs are essential for training advanced AI models, and the partnership aims to address Anthropic’s infrastructure constraints as demand for its AI tools grows. The company, valued at nearly $400 billion and backed by Google and Amazon, reportedly needed larger computing capacity to compete with rivals like OpenAI. The agreement includes a ‘Humanity Clause,’ allowing SpaceXAI to revoke computing access if Claude engages in actions harmful to humanity. This reflects Musk’s long-standing concerns about AI safety and bias. The clause also serves as a safeguard for SpaceXAI, aligning with Musk’s public stance on AI risks. From a business perspective, the deal helps SpaceXAI generate revenue by leasing Colossus 1, which would otherwise remain underused. Meanwhile, SpaceXAI has already shifted Grok’s training operations to its more advanced Colossus 2 system. The partnership marks a strategic shift for both companies amid intensifying competition in the AI sector.