Anthropic And Elon Musk Resolve Ethical Differences After Discovering Mutual Need For GPUs

MEMPHIS — Anthropic this week announced a deal to use the entire compute capacity of Elon Musk’s Colossus 1 data center, three months after Musk described the company on X as hating “Western civilization” and asked whether there was “a more hypocritical company than Anthropic.” CNBC

On the day of the announcement, Musk posted that he had spent time with senior Anthropic staff and was “impressed,” adding that “no one set off my evil detector” and that “so long as they engage in critical self-examination, Claude will probably be good.” That the detector failed to register anything is either a strong endorsement of Anthropic or a known limitation of the device when held by its operator. The detector is understood to be self-exempting.

CEO Dario Amodei explained the deal by noting that the company had planned for 10x growth and instead seen 80x, calling the surge “just crazy” and “too hard to handle.” The arrangement places Anthropic, a company founded on concerns about how OpenAI handled safety, in a data center owned by the only person in the industry who dislikes OpenAI more than Anthropic does. Fortune

SpaceX noted in its announcement that it “reserves the right to reclaim compute resources” if Anthropic’s AI harms humanity, a clause whose enforcement mechanism appears to be Elon Musk personally deciding, on a Tuesday. AIN