OpenAI data shows frontier enterprises, the top 10% of enterprise AI users, generate 8.3 times more output tokens per active user than typical firms, up from 2.6 times in January 2026. Codex accounted for 64% of combined Codex and ChatGPT enterprise output tokens as of June 2026, reflecting a shift toward longer, multi-step agentic tasks. Separately, average reasoning token consumption per organization rose approximately 320 times over the past twelve months, and per-token price declines have not offset rising usage intensity.
Adoption has spread well beyond engineering. Enterprise non-developer Codex users grew 189 times from August 2025 to early June 2026, surpassing the rate at which developers originally adopted the tool. Legal weekly active users grew 108 times since February 2026, sales and recruiting each 41 times, marketing 26 times, and engineering 5 times. OpenAI noted that early-career employees sent 13 more messages per week than executives after six months of adoption, a pattern it said contradicts survey-based findings showing higher executive usage. Frontier firms used Plugins in 21% of weekly sessions versus 9% at typical firms, while OpenAI's internal rate was 95%. Codex sessions invoked external tools in 60.3% of cases, compared to 21.9% for ChatGPT sessions.
Case studies across multiple enterprises illustrate the pattern. Zapier's enterprise marketing team uses ChatGPT Work to reduce lead funnel drop-offs, build campaign assets, and automate reporting. Virgin Atlantic used ChatGPT Work to conduct structured competitive research across multiple airlines' customer journeys, compressing a weeks-long process to hours, with findings directly informing recommendations for its new five-year digital strategy. RingCentral ran an AI-Native Challenge that gave every participant ChatGPT Work and Codex to build projects with no mandated workflow, resulting in thousands of employees including executives delivering functioning projects; the company's Program Management Office now uses AI-powered workflows to automate status reporting by aggregating data from Jira, Google Sheets, and CRM systems. OpenAI's CFO described rebuilding the finance function around AI, with goals including a zero-day close and continuously updated automated forecasting; a finance hackathon produced IR-GPT, a custom GPT that reduces investor relations diligence from hours to seconds, and finance professionals without coding backgrounds are using Codex to build custom dashboards and forecasting tools.