OpenAI Report: Significant Gap in AI Agent Usage Intensity Among EU Enterprises
BIBIBI
AT A GLANCE
OpenAI said that AI agent usage by EU enterprises has grown significantly, but usage intensity varies widely among enterprises and Member States.
Article
【OpenAI: Significant Gap in AI Agent Usage Intensity Among EU Enterprises】OpenAI Economic Research Team October 14 published a report stating that EU enterprises are using AI agent tools to expand their AI use cases, rather than merely asking AI questions, and are instead using agents to complete tasks from planning through execution. The report examined the growth of agent usage by large and small enterprises, differences in usage intensity, and the types of work people complete with ChatGPT. Among the EU sample of ChatGPT Enterprise customers,Codex and ChatGPT work currently account for more than half of output tokens. Since 2026 May year, the number of weekly active users of Codex and ChatGPT work among EU small and medium-sized enterprises has increased to more than 8.1 times the original level, with growth in every Member State. The article defines small and medium-sized enterprises as businesses with no more than 200 associated users. The lowest share of relevant output tokens in the sample was 22%in Slovenia, and the highest was 85%in Lithuania. Leading enterprises are defined as those in the top 10%by output tokens per active user; the sample includes only ChatGPT Enterprise and ChatGPT Business accounts.September The figure for leading EU enterprises was more than four times that of typical EU enterprises, while leading enterprises outside the EU still exceeded leading EU enterprises by more than twofold. Enterprises headquartered in the EU accounted for only 10%。
Key points
01
OpenAI An economic research team October 14 published a report examining AI agent usage by EU enterprises.
02
Among the ChatGPT Enterprise EU sample of customers,Codex and ChatGPT work currently account for more than half of output tokens.
03
Since 2026 year,May the number of weekly active users of Codex and ChatGPT work among EU small and medium-sized enterprises has increased to more than 8.1 times the original level, with growth in every Member State.
04
The article defines small and medium-sized enterprises as businesses with no more than 200 associated users.
05
The lowest share of relevant output tokens in the sample was 22%in Slovenia, and the highest was 85%。
06
Leading enterprises are defined as those in the top 10%by output tokens per active user; the sample includes only ChatGPT Enterprise and ChatGPT Business accounts.
07
September The output tokens per active user at leading EU enterprises were more than four times those at typical EU enterprises.
08
The same metric at leading enterprises outside the EU exceeded that of leading EU enterprises by more than twofold.
09
Enterprises headquartered in the EU accounted for only 10%。
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
AI agent tools do more than answer questions; they can also participate in tasks from planning through execution. The report uses metrics such as output tokens and weekly active users to observe enterprise usage. The sample shows that relevant user growth among EU small and medium-sized enterprises has been relatively rapid, but significant gaps remain among Member States, typical enterprises, and leading enterprises.
Why it matters to readers
These data reflect the activity level and internal disparities in EU enterprises’ use of AI agent tools, and also show that user growth does not mean all enterprises have reached similar usage intensity.
Beginners can understand weekly active users as 1 the number of people who actually used the product during the week, and output tokens as AI a unit for measuring the amount of generated content; the two metrics respectively reflect the number of users and usage intensity.
Risks and unknowns
The existing evidence is a retelling of the report and does not provide OpenAI the link to the original report.
The evidence does not establish the report’s complete title, full publication date, or specific statistical methodology.
The sample includes only ChatGPT Enterprise and ChatGPT Business accounts, so it cannot be confirmed whether it represents all EU enterprises.
The evidence does not specify the exact product or functional scope of “ChatGPT work”.
Related Developments
Loading event timeline…
The reasons behind the differences in output token share and multiples are not explained in the evidence.
Related concepts
AI Agents
This term is not in the glossary yet. Browse related concepts in the glossary.