Mechanism: The company tracks usage of Codex and ChatGPT Work in its enterprise and small-business accounts, using weekly active users and output-token shares as proxies for agentic activity. Why now: The report turns broad claims about an “agentic shift” into customer-account measures, while exposing substantial variation by country and firm type.
OpenAI reports that weekly active users of Codex and ChatGPT Work among EU small and medium businesses grew 8.1× since May 2026, with growth across every member state. The report defines an SMB as a business with 200 or fewer associated users. In sampled ChatGPT Enterprise accounts, the share of output tokens generated through Codex and ChatGPT Work ranged from 22% in Slovenia to 85% in Lithuania. These are concrete observations from OpenAI’s analysis, not independent measurements of all EU firms.
First implication: For builders and business leaders, the figures make a case to examine actual task-level adoption and country-level differences rather than treating “the EU” as one adoption market. Token output and active-user counts do not establish completed work, productivity gains, or economic impact. The analysis is based on OpenAI customer accounts and its own definitions; it cannot establish representative adoption across EU businesses or show that increased usage caused improved outcomes. First watch condition: Watch for independent, comparable data on agentic task completion and business outcomes across firms beyond OpenAI’s customer base. Such evidence would help determine whether usage growth is translating into measurable operational value.