Claude Code as Distribution Wedge
SOTA model nightclub hype cycle
Anthropic’s breakout showed that the winning AI lab was no longer just the one with the best model, it was the one that could turn its own coding model into an internal software factory. Claude Code gave Anthropic a fast way to build, test, and ship adjacent products, then reuse the same agent workflow in Excel, PowerPoint, and knowledge work. That turned coding strength into product velocity, and product velocity into much faster revenue expansion.
-
Claude Code became Anthropic’s distribution wedge. It grew to an estimated $2.5B annualized revenue by February 2026, ahead of Cursor at $2B and Codex at $1B in January, which gave Anthropic a large, high usage base of developers generating constant feedback and internal dogfooding.
-
The new products were concrete extensions of the same core interaction. Claude for Excel and PowerPoint moved the agent into office workflows, while Claude Cowork applied the Claude Code style interface to finance, legal, and code review tasks, pushing Anthropic beyond developers into higher value enterprise seats.
-
OpenAI had broader consumer distribution, but Anthropic’s growth engine was more expansion heavy. OpenAI rose from an estimated $20B annualized revenue at the end of 2025 to $25B by February 2026, while Anthropic jumped from $9B to about $30B by March 2026 as coding and agent workflows drove much higher token usage.
The next phase is a race to turn coding agents into the default interface for all desktop work. Anthropic has shown that a lab can use AI coding to speed up its own shipping cadence, then export that workflow into adjacent enterprise products. OpenAI is now following the same path with Codex, so the market is converging on full stack agent operating systems for work.