The EU’s next chapter in AI is agentic
Oct 13, 2026, 5:00 PM · OpenAI
OpenAI’s EU data says Codex and ChatGPT Work already drive more than half of enterprise output tokens—while typical European firms still trail the global frontier.
Why it matters
OpenAI’s Economic Research report on the EU finds that among ChatGPT Enterprise customers sampled, Codex and ChatGPT Work now account for more than half of output tokens. Weekly active users of those agentic tools among EU small and medium businesses have grown more than eightfold since May 2026, across every member state.
Adoption is uneven. The share of output tokens through Codex and ChatGPT Work ranged from 22% in Slovenia to 85% in Lithuania in the enterprise sample. Frontier firms—top 10% by output tokens per active user—generated more than four times as many tokens per user as typical EU firms in September, yet still produced less than half the intensity of frontier firms outside the EU. EU-based firms were only 10% of global frontier firms in the sample.
OpenAI CFO Sarah Friar framed the moment as a new era of tools that carry out tasks, with room for deeper use across Europe.
From the desk
We’re treating this as a productivity map, not a victory lap. When more than half of enterprise output tokens in OpenAI’s EU sample already flow through agentic surfaces, the chat box is no longer the main event for the firms that are all-in.
The gap story is the one that matters for ministers and boards. Frontier EU firms outrun their local peers and still lag global frontier intensity. That is a skills, process, and access problem as much as a model problem. Eightfold growth in SMB agentic users since May shows curiosity is spreading; token intensity shows who actually rebuilt workflows.
OpenAI has an interest in painting Europe as under-adopting so policy stays friendly and seats keep filling. Fair. The comparative numbers still land: agentic use is real, lumpy by country, and concentrated among firms that already treat AI as infrastructure.
Our read: the EU’s next chapter is less about another awareness campaign and more about making advanced agentic use affordable, trainable, and safe enough for the median firm—not only the 10% that already live in the tools.
Context
Businesses in Europe are shifting from Q&A chat to agents that plan and complete work in tools like Codex and ChatGPT Work. OpenAI is publishing usage research as that shift accelerates.
Who feels it
- EU enterprises and SMBs
- Agentic tools are moving from experiment to default for heavy users; laggards risk a widening productivity gap.
- Policymakers
- Skills, access, and trust—not just model availability—will determine whether adoption deepens.
- Workers
- Task-completing agents change job design; training and oversight become table stakes.
- OpenAI competitors
- EU agentic share is a benchmark other platforms will be measured against.
What to watch
- Whether EU token intensity closes on the global frontier over the next quarters
- Country-level policy responses that target skills and SMB access
- Independent surveys that validate or challenge OpenAI’s sample
- How Codex and ChatGPT Work usage mixes shift by sector