Enterprise adoption of generative AI across the European Union is shifting from simple text queries toward execution-oriented agentic workflows. A report from OpenAI Economic Research reveals that agentic tools now generate over 50 percent of tokens among sample ChatGPT Enterprise clients in the bloc, driven by surging usage in small businesses despite persistent cross-border adoption disparities.
Enterprise artificial intelligence usage across the European Union is rapidly transitioning from passive conversational queries toward autonomous systems designed to execute specific tasks. According to an analysis published by OpenAI Economic Research, agentic tools including Codex and ChatGPT Work now generate more than 50 percent of all tokens consumed by a sample of European ChatGPT Enterprise customers. The rapid expansion occurred despite the implementation of strict governance standards under the EU AI Act, demonstrating that regulatory oversight has channeled corporate demand toward structured, controllable processes such as software development support and workflow automation.
Adoption grew most aggressively among smaller organizations facing labor constraints. Between May 2026 and October 2026, weekly active users of Codex and ChatGPT Work surged by more than 8.1 times within a sample of EU small and medium-sized enterprises with fewer than 200 employees. However, adoption rates across member states remain highly fragmented. Lithuania led the region with agentic tools accounting for 85 percent of ChatGPT Enterprise output tokens, whereas Slovenia recorded just 22 percent, reflecting a 63 percentage point divide driven by differences in national digital maturity and software engineering talent pools.
Despite the rapid growth in token consumption, proving tangible economic returns remains a key challenge for enterprise leaders. Industry commentators have noted that output token volumes and user counts do not directly demonstrate error-free task completion rates, productivity gains, or net profitability improvements. Over the next six months, chief technology officers and financial leadership teams across Europe are expected to scrutinize token-based expenditures to determine whether agentic implementations deliver sustained economic value beyond conversational experimentation.
โ AIDEN Editorial Team ยท Reviewed by ์ดํ๋ฏผ
What this means for the market
The shift toward agentic AI indicates that global enterprises are increasingly prioritizing operational workflow integration over general conversational chatbots. While aggressive token growth in small and medium-sized organizations highlights the appeal of labor-augmenting automation tools, stark regional adoption gaps underline the influence of local infrastructure and technical talent. For AI developers and providers, future enterprise retention will rely on proving accurate, error-free task completion rather than driving raw token usage.
How this issue is unfolding
In the global artificial intelligence race, Europe faced frequent concerns that stringent frameworks such as the EU AI Act would cause enterprise adoption to lag behind the United States and China. However, OpenAI data demonstrates that the region is transitioning rapidly toward agentic AI that plans and delegates concrete tasks rather than relying on informational queries. Demand for code generation via Codex and administrative automation through ChatGPT Work expanded significantly in small and medium-sized enterprises over a short period. Nevertheless, sharp adoption disparities persist among member states, illustrated by the contrast between Slovenia at 22 percent token share and Lithuania at 85 percent. Furthermore, enterprise practitioners still face the critical task of verifying whether rising token volumes actually deliver measurable productivity enhancements and operational profitability.
