Anthropic Files Confidential IPO and Proposes Sector-Wide Pause on Advanced AI

Overview

• The company filed confidential U.S. IPO paperwork this week and selected Morgan Stanley and Goldman Sachs to lead the offering with JPMorgan participating.
• Anthropic executives say the primary reason for going public is to raise cash to cover soaring costs for chips, data centers and the large-scale compute that trains its models.
• In recent months the firm closed a massive funding round and signed multibillion-dollar compute agreements with partners including SpaceX and Akamai to lock in chip and data-center capacity.
• Anthropic published a public proposal urging major labs to create a verifiable mechanism to temporarily slow advanced AI development, citing risks of 'recursive self-improvement' and internal data showing engineers produce far more code with AI assistance.
• The company is contesting a March U.S. Department of Defense 'supply-chain risk' designation in court even as ties with parts of the White House have warmed, a dynamic that could affect government contracts and investor confidence.