OpenAI enterprise data shows AI use shifting from assistance to execution

Enterprise AI is becoming more execution-oriented

OpenAI's August 2026 enterprise update argues that business use of AI is moving beyond one-off assistance toward systems that execute more substantial work. The company combines product telemetry and organizational research in two reports, Enterprise Signals and How Organizations Use AI: Evidence from ChatGPT.

One of the clearest signals is coding activity. OpenAI reports that, as of June 2026, Codex generated 64% of the combined output tokens produced through Codex and ChatGPT among enterprise customers in the analysis. That does not mean coding represents 64% of all enterprise AI value, but it does show how quickly execution-oriented developer workflows have grown inside the measured customer base.

Frontier adopters are using much more AI

OpenAI describes the top 10% of firms by AI usage as frontier firms. In the reported sample, these organizations generated 8.3 times as many output tokens per active user as typical firms, up from a 2.6-times gap in January. Weekly plugin usage was also higher: 21% at frontier firms compared with 9% at typical firms. OpenAI itself reported 95% weekly plugin usage internally.

These are adoption and usage measures, not proof that heavier AI usage automatically causes stronger business performance. The research is most useful as evidence about how deployment patterns differ across organizations and roles.

Adoption also varies by seniority

The accompanying research says that six months after adoption, early-career employees sent about 13 more ChatGPT messages per week than executives in the studied organizations. That pattern suggests AI may be especially embedded in the day-to-day task volume of employees closer to hands-on execution, although role mix and organizational context matter.

What employers and job seekers should watch

For employers, the practical lesson is that value increasingly depends on workflow integration, governance and access to execution-capable tools rather than simply giving employees a chatbot account. For job seekers, the shift increases the value of skills that combine domain expertise with AI-assisted coding, agent workflows, automation, tool use, verification and responsible deployment.

The strongest takeaway is not that every organization should maximize token usage. It is that the frontier of enterprise adoption is moving toward deeper integration of AI into repeatable work, and the gap between light and heavy adopters is widening on OpenAI's own usage measures.