OpenAI Releases First Enterprise AI Usage Data: Large Companies Lead, New Employees Most Active

In mid-August, OpenAI released two reports: the working paper 'How Organizations Use AI' and the companion report 'Enterprise Signals', marking the first time it has answered 'who in enterprises uses AI and for what' based on real usage data. The paper matches ChatGPT Enterprise account records with employee job titles, tasks, and financial data from public companies, covering over 1,500 organizations and 17 million messages. The task classification sample covers 973 organizations and nearly 8.7 million messages, matched with financial data from 417 U.S. public companies. All results are anonymized and aggregated; researchers did not manually review any enterprise messages.

The paper tracks organizations adopting ChatGPT Enterprise from January 2024 to March 2026, using 'per company, per calendar week' as the unit of analysis. In aggregate, output tokens grew about 7x from June 2025 to March 2026, with about half of that growth coming from existing adopters 'intensifying' their usage—the same cohort that adopted between January 2024 and June 2025 saw output tokens grow about 4x over the same period. This indicates that enterprise AI usage deepens over time rather than being a one-time deployment. OpenAI acknowledges that tokens are a rough metric, but 'using more over time' is a clear directional signal.

Comparing U.S. public companies that adopted AI versus those that didn't, adopters have a median revenue of $2.275 billion versus $209.6 million for non-adopters, a roughly 10x difference; total assets are $4.394 billion versus $668 million, and employee counts are 2,934 versus 424. Even after controlling for industry and size, companies with higher per-capita revenue are more likely to adopt, while those with higher per-capita fixed assets are less likely, indicating that 'asset-light, knowledge-heavy' firms lead. Firms in the top 25% of revenue have a 6.9 percentage point higher adoption probability, and the top 5% have a 9.8 percentage point higher probability. Additionally, thicker 'organizational foundations'—such as higher SG&A, R&D, and software spending—correlate with higher adoption, confirming that AI requires complementary capabilities to be effectively utilized.

The paper breaks down active users six months after adoption by job function and seniority: engineering and technology account for about 11% of weekly active users, executives about 9%; by seniority, managers and directors about 24%, and general employees about 15%. But the real gap lies in intensity—entry-level employees and interns send 8 to 9 more messages per week than the average, while executives send significantly fewer, and this pattern holds within the same company. Task classification shows that over half of active users use AI for document and technical writing, and nearly half for technical and digital work, spanning dozens of tasks. In finance and insurance, financial and tax tasks dominate; in entertainment and retail, sales and marketing tasks are more common.

The companion report shows that by June 2026, Codex output tokens accounted for 64% of total enterprise customer output tokens, indicating that agents are taking on substantive tasks. Since February, weekly active Codex users in legal, sales, recruiting, and marketing departments have grown 108x, 41x, 41x, and 26x respectively, while engineering has grown only 5x. The gap in per-capita output tokens between frontier firms (top 10% in monthly per-capita output tokens) and typical firms (middle 10%) widened from 2.6x in January to 8.3x in June, reaching 11.7x in the information technology sector. Frontier firms give agents context, tools, and persistence, and set clear boundaries; 21% of weekly active users in frontier firms use plugins, compared to 9% in typical firms, and 95% within OpenAI itself.