Anthropic

Overview

Frontier foundation-model lab; maker of Claude. Per 2026-05-01-entree-capital-enterprise-ai-spend-map (citing Sacra and Menlo), Anthropic is the API + coding-agent archetype: ~80% of revenue is enterprise, 70–75% via the API. The lab the team’s tooling actually rides on day-to-day — Claude Code, Claude managed agents in cursor, the Brain prototype’s own LLM calls.

Relationships

• One of three labs that capture ~$35B (42%) of all 2025 enterprise AI spend, per 2026-05-01-entree-capital-enterprise-ai-spend-map.
• Direct upstream of cursor (Cursor is one of the AI-native apps named as paying token costs back to the labs); also the model behind Claude Code and the Brain prototype’s working harness.
• Saastr/ETR enterprise-share trend: 21% (Sep 25) → 48% (Mar 26) — sharpest enterprise-share gain among the three labs in the period covered.

Notes

• Revenue mix (Feb 2026, annualized): ~$14Btotal.API/Developer$10B; Claude Code $2.5B(>50$1B; Consumer subs $0.5B. Heavily API-led, the inverse of openai.
• Trajectory: Reached $30BannualizedbyApril2026(Sacra).SapphireVenturesforecasts$70B ARR by 2028.
• Claude Code is the breakout product within the lab — fastest-growing line and >50% enterprise share. Most directly relevant to the Brain team because Claude managed agents are the pattern cursor is reportedly trying to replicate (per 2026-05-06-brain-os-strategy-brainstorm).
• Strategic implication for the Brain. If the team picks a vertical and rides Anthropic’s API as the model layer, the unit economics are token-cost driven — the cursor-style ~20% margin on tokens is the realistic ceiling, not a SaaS-style 80% margin. See foundation-labs-monetization-shift.

Mentioned in

• 2026-05-01-entree-capital-enterprise-ai-spend-map — Entrée Capital’s enterprise AI spend map (May 2026); detailed revenue breakdown on slide 5; flagged as the “labs concentrate” growth story.