AI Coding Tools Weekly Showdown

Issue #1: Copilot goes usage-based, Claude Code keeps winning hearts, Cursor ships hard

GitHub Copilot flips to usage-based billing — the same move that burned Cursor users a year ago — while Claude Code drops internal data showing 80%+ of Anthropic's code is now AI-authored. Cursor 3.7 ships nested subagents, custom tools, and Design Mode voice input. Who got better this week, where, and whether it changes the switching calculus.