🔍 Read the full analysis: OpenAI Says Agentic AI Is Common In EU Business. Check The Denominator on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
OpenAI reports that Codex and ChatGPT Work generated 55% of output tokens across its ChatGPT Enterprise and Business accounts in the EU by September 30, 2026. The figure is a measure of generated text, not the share of business tasks, users or value handled by agents; the company’s data also reflects its own customers and a recent product launch.
OpenAI said Codex and ChatGPT Work generated 55% of output tokens in its ChatGPT Enterprise and Business accounts across the 27 EU member states by September 30, 2026. The finding, published in a new company report, points to growing use of agent products among OpenAI’s business customers, but it does not mean agents handle most tasks or create most business value.
OpenAI Economic Research’s report, Europe’s Agentic Opportunity: Moving from breadth to depth with AI agents, defines agentic use as activity through Codex, OpenAI’s coding agent, and ChatGPT Work, an agent mode launched in July that can work with files, browsers and company tools. In the September 30 snapshot, Codex accounted for 34% of output tokens and ChatGPT Work for 21%; other ChatGPT use made up the remaining 45%.
For businesses with 200 or fewer users, OpenAI reports that weekly active users of Codex or ChatGPT Work increased 8.1 times between the first week of May and the week ending October 5. The company says every EU member state recorded at least fivefold growth. Belgium had the largest reported increase, at 12.1 times, followed by Italy at 10.8 times and Luxembourg at 10.0 times. OpenAI does not provide absolute user counts alongside these growth figures.
The report also compares agentic token share by country. It ranges from 85% in Lithuania and 81% in Cyprus to 22% in Slovenia and 26% in Austria. Germany is reported at 51%, ranking 17th among the 27 member states. These are figures for OpenAI’s accounts, not a survey of all businesses in each country.
OpenAI’s Europe report · in one view
From 22% to 85%: how agentic the EU’s OpenAI business accounts are
Share of output tokens produced by Codex and ChatGPT Work in September 2026, per member state, next to how fast small businesses adopted the agents since May.
Agentic share of output tokens, September 2026
Dark tick = EU median (56%). Right column: small-business agent users vs. May; green = at or above the median (8.1×).
- 1Lithuania85%6.8×
- 2Cyprus81%7.5×
- 3Hungary79%8.1×
- 4Poland76%6.5×
- 5Sweden73%8.0×
- 6Finland72%5.1×
- 7Estonia71%7.8×
- 8Netherlands68%8.5×
- 9Croatia68%6.8×
- 10Italy65%10.8×
- 11Slovakia58%7.8×
- 12Belgium57%12.1×
- 13Denmark56%9.9×
- 14Malta56%9.5×
- 15France55%9.2×
- 16Czechia52%8.1×
- 17Germany51%7.5×
- 18Romania49%5.9×
- 19Luxembourg47%10.0×
- 20Bulgaria42%5.9×
- 21Ireland39%9.5×
- 22Greece37%9.8×
- 23Latvia35%5.8×
- 24Spain30%8.8×
- 25Portugal29%6.5×
- 26Austria26%8.2×
- 27Slovenia22%9.2×
Read as a customer map, not a league table: no account or user counts per country. Share and growth barely relate (rank correlation -0.19): Slovenia is last on share but grows 9.2×, Lithuania first on share but grows 6.8×.
Source: OpenAI Economic Research, “Europe’s Agentic Opportunity: Moving from breadth to depth with AI agents” (October 2026); ChatGPT Enterprise and Business accounts. “Agentic” = Codex + ChatGPT Work (launched July 2026). Small and medium businesses = 200 or fewer associated users. OpenAI publishes no sample sizes per country. Rank correlation: own calculation (Spearman) from Figures 2 and 4.
Powered by Thorsten Meyer AI
The data offers a view into where OpenAI’s business customers’ model output is going as the company adds tools that can perform multi-step work. It suggests these products have become a meaningful part of usage within the accounts OpenAI tracks. For firms weighing agent adoption, the report provides a signal of activity, though not a measure of whether agents are saving time, improving output or generating returns.
The distinction matters because an agent may read documents, plan, produce code or text, check its work and repeat steps. Those processes can generate substantially more tokens than a short chat response. OpenAI’s global Enterprise Signals page describes tokens as an imperfect measure of business value and says agentic tasks can require more computation. A high token share may therefore reflect the resource intensity of agent runs, rather than agents completing a majority of work.
The report’s broader policy argument is that Europe should turn adoption into deeper workplace use by building skills and confidence and making advanced tools more accessible and affordable. That is OpenAI’s interpretation of its findings. The account-level data does not establish how representative its customers are of the European economy or whether the reported usage has produced measurable productivity gains.
Popular right now
Roborock Q7 M5+ Robot Vacuum and Mop
Roborock Q7 L5 Robot Vacuum and Mop Combo
Ninja CREAMi 7-in-1 Ice Cream Maker, NC301
As an affiliate, we earn on qualifying purchases.
How OpenAI Counts Agent Use
The report was published October 14 by OpenAI’s Global Affairs team alongside the research paper from OpenAI Economic Research. Its EU figures cover ChatGPT Enterprise and ChatGPT Business accounts in the bloc. A separate OpenAI Enterprise Signals page supplies global comparisons. The material is a vendor’s analysis of activity on its own products, rather than an independent census of business AI use.
OpenAI says EU frontier firms—defined as the top 10% by output tokens per active user—generated about four times as many tokens per user as typical EU firms in September. It also says frontier firms outside the EU produced more than twice the output per user of the EU frontier group. EU firms made up about 10% of global frontier firms in the company’s data, compared with about 65% in North America. These comparisons use OpenAI’s account and token definitions; they do not directly rank companies by productivity or economic performance.
One timing detail shapes the growth figures: ChatGPT Work launched in July, after the May baseline used for the small-business comparison. OpenAI’s report says a major part of the increase followed that launch. The 8.1-times figure thus records increased activity across a period when a new product became available, not solely a change in demand for a product that had been available throughout.
What the Account Data Cannot Establish
OpenAI does not report the number of users or accounts behind each country’s figures, nor does the supplied material give absolute weekly active-user counts for the small-business growth calculation. Without those denominators, readers cannot judge how much a few large customers may affect national shares or how large the reported increase is in raw user numbers.
The report also does not provide a direct count of agentic tasks, a comparison of task outcomes, or evidence that token growth has translated into productivity or financial returns. The 55% figure is a share of generated output tokens, not a share of interactions. The article’s source offers an illustrative estimate that, under an assumed 10-to-30-times difference in output per agent run, a 55% token share could correspond to a much smaller share of interactions. That is the author’s hypothetical calculation, not a result measured by OpenAI.
Country rankings should be treated with similar care. OpenAI notes that smaller countries can have relatively high agentic use within larger firms, but does not disclose country sample sizes. It is therefore unclear how representative the country-level rates are, or whether differences reflect broad national adoption, the mix of customers using OpenAI, or a small number of high-use accounts.
What to Watch in Later Data
The next useful test will be whether OpenAI publishes later figures with absolute user counts, account counts and comparable time periods, particularly after ChatGPT Work has been available for longer. That would help distinguish sustained adoption from a launch-related rise and make the small-business growth rate easier to interpret.
Readers should also look for measures beyond tokens: how many tasks agents complete, what kinds of work they perform, and whether businesses report changes in time, quality or cost. The report does not announce a specific date for a follow-up dataset or independent validation. Until more detail is available, its findings are best read as a snapshot of usage within OpenAI’s EU business accounts—not a measure of agent adoption or productivity across Europe as a whole.
Key Questions
What does OpenAI mean by 55% of output being agentic?
OpenAI says Codex accounted for 34% and ChatGPT Work for 21% of output tokens in its EU Enterprise and Business accounts by September 30, 2026. Together, that is 55% of generated tokens—not 55% of users, tasks or business value.
Did small-business agent use grow 8.1 times across Europe?
OpenAI reports an 8.1-times rise in weekly active users of Codex or ChatGPT Work among businesses with 200 or fewer users. The comparison runs from the first week of May to the week ending October 5, 2026. The company does not provide absolute user totals, and the period includes the July launch of ChatGPT Work.
Does the report show that agents do most European business work?
No. It covers OpenAI’s own business accounts and measures output tokens. It does not count all European businesses, tasks or work completed, and it does not establish what share of business value comes from agents.
Why do the country figures need caution?
OpenAI does not publish account or user sample sizes by country. Smaller customer groups or a few high-use firms could influence national rates, so the figures should not be treated as a ranking of overall national AI readiness.
Source: ThorstenMeyerAI.com
