Descriptions:
Miles Clements, a growth-stage investor at Accel who has backed Cursor, Atlassian, and Linear, joins 20VC to push back on the “Cursor is dead” meme and explain why the AI coding market is more nuanced than a straightforward battle between Cursor and Anthropic’s Claude Code.
Clements introduces a two-axis framework for evaluating AI companies: time to value and durability of value. He argues that coding tools are unique in excelling on both dimensions simultaneously — a developer can become measurably more productive within an afternoon of using Cursor, and that productivity compounds as the tool embeds deeper into team workflows. He contrasts this with early vibe-coding apps, which offered fast time to value but collapsed quickly due to zero durability, and with legal or accounting AI, which has slow adoption cycles but transformational stickiness once embedded.
On the Cursor vs. Claude Code debate, Clements contends the market is growing fast enough that both tools can thrive — they are expanding the developer population by onboarding people who would not have coded at all a year ago, while simultaneously driving consumption-based ARR growth independent of seat count. He attributes much of Claude Code’s recent momentum specifically to the Opus 4.5 and 4.6 model releases, calling it model-tethered success. Clements also addresses why Cursor being strongly identified with the IDE has created a perception problem, and closes with a warning that ARR growth can blind investors to underlying product health — usage intensity is the more important signal in the current environment.
📺 Source: 20VC with Harry Stebbings · Published March 09, 2026
🏷️ Format: Interview
