06-reference

stratechery anthropic revenue nvidia earnings

Tue Mar 03 2026 19:00:00 GMT-0500 (Eastern Standard Time) ·reference ·source: Stratechery ·by Ben Thompson
anthropicnvidiaearningsrevenueagentic-aiclaude-codetokens

Anthropic’s Skyrocketing Revenue, Nvidia Earnings

Two major stories. First, Anthropic hit ~$19B annualized revenue run rate, more than doubling from $9B at end of 2025 — driven largely by Claude Code adoption. Ramp data shows Anthropic dominating OpenAI in both API spend and enterprise subscriptions. Thompson notes Anthropic’s rapid growth raises stakes for the DoW standoff: the company doesn’t need government revenue, which means government must use more coercive means if it wants control. He sees positive signs of de-escalation from Amodei’s Morgan Stanley comments, and suggests OpenAI’s contract language (specific legal terms vs. vague restrictions) could provide a template for compromise.

Second, Nvidia reported $68.1B quarterly revenue (up 73%), $43B net income (up 94%). Wall Street still unimpressed (stock down 5.5%). Thompson identifies three AI inflection points expanding Nvidia’s addressable market: (1) LLMs (ChatGPT moment — tokens as answers), (2) reasoning models (o1 moment — exponential token increase for thinking), (3) functional agents (Opus 4.5 moment — agents multiply token demand squared, since humans run multiple agents that each trigger multiple reasoning chains).

Key Huang quote: “In this new world of AI, compute equals revenues.” The SaaS bear case and the Nvidia bear case are contradictory — if agents replace software companies, that means massively more tokens needed.

RDCO Mapping

Critical Anthropic data point for moat thesis cross-reference with Jaya Gupta analysis. The three inflection points framework is useful for understanding compute demand trajectory and RDCO infrastructure planning. Claude Code’s role in Anthropic’s revenue dominance validates our own tooling decisions.