OpenAI finds uneven agentic AI adoption across EU businesses

Agentic tools account for most output in OpenAI’s EU business sample

OpenAI says Codex and ChatGPT Work now account for more than half of output tokens among ChatGPT Enterprise customers in its EU sample. The company’s latest Economic Research report describes a shift from asking AI questions to using agents that can support work from planning through completion.

The change is also visible among smaller organisations. OpenAI defines small and medium businesses in this analysis as those with 200 or fewer associated users. Since May 2026, weekly active users of Codex and ChatGPT Work in that segment have increased more than eightfold across the EU, with growth in every member state.

Usage intensity differs sharply between member states

OpenAI found that agentic use is not evenly distributed across the bloc. Within the sampled ChatGPT Enterprise accounts, the share of output tokens generated by Codex and ChatGPT Work ranged from 22% in Slovenia to 85% in Lithuania. The company also says that the types of work users bring to ChatGPT differ by country.

The findings point to differences not only in whether organisations adopt AI, but in how deeply they apply it to business tasks. OpenAI frames agentic systems as tools that can help users carry work through multiple stages rather than simply provide responses to individual prompts.

EU frontier firms remain behind the global frontier

OpenAI defines frontier firms as the top 10% of companies by output tokens per active user; this part of the sample includes ChatGPT Enterprise and Business accounts. In September, EU frontier firms generated more than four times as many output tokens per active user as typical EU firms.

However, frontier companies outside the EU generated more than twice as many output tokens per active user as frontier firms within the EU. EU-based businesses represented only 10% of frontier firms globally, despite the EU accounting for a higher share of registered companies worldwide, OpenAI said.

From experimentation to operational use

Sarah Friar, OpenAI’s chief financial officer, said Europe had entered an era in which AI tools can carry out tasks for people. She said wider benefits would depend on giving people access to tools, skills and confidence to use them well.

For business leaders, the practical implication is to assess where agentic tools can support complete work processes, not just isolated experiments, while ensuring employees have the capability and confidence to use those tools.