OpenAI reports a widening AI-usage gap among its corporate customers

OpenAI published a report arguing that a small group of corporate customers is pulling sharply ahead of the rest in how deeply they use its tools. Drawing on aggregated usage data across its enterprise customer base, the company said firms in the top decile of usage generated 8.3 times as many AI output tokens per active user as typical firms in June, up from 2.6 times in January — a gap it described as tripling in five months.

The report attributed the widening gap to depth of integration rather than raw adoption: frontier firms were far likelier to connect AI agents to internal tools and repeatable workflows, and adopted OpenAI’s Plugins and Skills features at roughly double the rate of typical firms. It also reported uneven growth by function — legal teams increased their use of OpenAI’s coding agent Codex 108-fold since February, against a fivefold rise in engineering — and said Codex now accounts for the majority of combined Codex-and-ChatGPT enterprise token usage, which OpenAI framed as a shift from employees asking AI for help toward delegating multi-step tasks to it. A further finding complicated the usual survey-based picture of adoption: usage data showed early-career employees using AI more heavily than executives, the opposite of what self-reported surveys typically find.

OpenAI is a commercially interested party in this account — the figures come from its own product telemetry, not independent measurement, and the company was explicit that the data does not yet establish whether the spending behind frontier firms’ AI use pays for itself. Set alongside a companion report on global ChatGPT usage patterns published the same week, it is best read as OpenAI’s own case for how deep its product needs to reach inside a customer’s workflows to show a return, at a moment when the company is pushing hard on enterprise revenue ahead of a prospective IPO.