OpenAI tells investors enterprise revenue has overtaken ChatGPT consumer business

OpenAI's business has quietly crossed a milestone that reshapes how to think about the company: enterprise revenue now exceeds consumer, according to CFO Sarah Friar's remarks to investors reported by CNBC. The crossover arrived ahead of OpenAI's own year-end 2026 forecast, with an annualized revenue run rate of $40 billion, 20% month-over-month growth, and a 32% jump in business customers during July 2026.

The engine of the shift is agentic coding. Codex, OpenAI's coding agent, accounted for 64% of combined Codex-plus-ChatGPT output tokens among business customers as of June 2026 — a striking indicator that enterprises are burning far more inference on autonomous coding workflows than on chat. That token mix explains why enterprise monetizes so differently from consumer: agent workloads run long, multi-step jobs that consume orders of magnitude more compute per user than a chat session.

Competitively, the disclosure lands as rivals chase the same enterprise coding budget: Google positioned Gemini 3.7 Flash explicitly for coding and agents, xAI pushed Grok 4.6 into GitHub Copilot, and Anthropic's Claude Code remains a developer favorite. OpenAI's enterprise-over-consumer flip mirrors the trajectory of prior cloud platforms where developer tooling became the durable revenue base.

Separately reported context: OpenAI also lifted text-conversation limits for free ChatGPT users tied to a GPT-5.6 update, and introduced Ultrafast mode running GPT-5.6 Sol at up to 750 tokens/second via a Cerebras partnership. Skeptics will note run-rate figures shared with investors are not audited revenue, and the 20% monthly growth rate is unlikely to be sustainable indefinitely. Watch whether OpenAI formalizes these numbers ahead of any public financing.