OpenAI Says Most EU Business AI Is Agentic—What Was Measured?

🔍 Read the full analysis: OpenAI Says Most EU Business AI Is Agentic—What Was Measured? on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get smart everyday buys delivered free with Prime

• Fast, free delivery on millions of items
• Prime Video, Amazon Music and more included
• Member-only deals all year

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

OpenAI says AI agents generated 55% of output tokens in its ChatGPT Enterprise and ChatGPT Business accounts across the European Union by Sept. 30, according to a report published Oct. 14. The figure covers output from Codex and ChatGPT Work; it measures generated text, not the share of users, business tasks or economic value represented by agent use.

OpenAI Economic Research’s report, Europe’s Agentic Opportunity: Moving from breadth to depth with AI agents, defines “agentic” use as output from Codex, the company’s coding agent, and ChatGPT Work, an agent mode launched in July 2026 that can work with files, browsers and company tools. By Sept. 30, Codex accounted for 34% of output tokens and ChatGPT Work for 21%. The other 45% came from standard ChatGPT use, the report says.

The company also reports rapid growth among smaller businesses. In accounts with 200 or fewer users, weekly active users of Codex or ChatGPT Work rose 8.1 times between the first week of May and the week ending Oct. 5. OpenAI says every EU member state recorded at least fivefold growth; Belgium, Italy and Luxembourg had the highest reported multiples. The report does not provide absolute user counts behind that growth measure.

Agentic output shares varied substantially by country in OpenAI’s customer data, from 85% in Lithuania and 81% in Cyprus to 22% in Slovenia and 26% in Austria. Germany’s reported share was 51%. OpenAI also compares high-output “frontier firms”: it says the top 10% of EU firms by output tokens per active user produced about four times the output of typical EU firms, while frontier firms outside the EU produced more than twice the output of the EU frontier.

What the 55% Share Measures

The report offers a view into how businesses using OpenAI’s paid work products are engaging with its newer agent tools. It suggests those tools account for a large portion of the text generated in these accounts, and the small-business trend provides a signal that use expanded after ChatGPT Work became available. For companies weighing adoption, the figures may point to growing interest in workflows that let software handle multi-step work rather than only answer prompts.

But the headline percentage should not be read as meaning that agents perform most AI work across Europe. OpenAI’s Enterprise Signals material calls tokens an imperfect measure of business value and says agentic tasks can require substantially more computation. An agent may read files, plan, draft, check its work and try again, generating many more tokens than a short chat response. A high token share can therefore reflect the length and computational intensity of agent runs rather than their share of tasks or value.

The comparison between EU and non-EU frontier firms also raises a question about the depth of adoption, not simply whether businesses have tried AI. OpenAI frames the policy challenge as moving from broader adoption toward deeper use, worker skills and access to advanced tools. The report’s data can inform that discussion, but they do not by themselves show why the output gap exists or whether it translates into a productivity gap.

How OpenAI Built the Figures

The report covers ChatGPT Enterprise and ChatGPT Business accounts in the 27 EU member states. It was released by OpenAI Global Affairs alongside a ten-page analysis from OpenAI Economic Research. OpenAI’s separate global Enterprise Signals page supplies additional comparisons. The figures describe activity within the company’s own customer base, rather than a survey of all European firms or a measure of AI use across the economy.

The timing matters for interpreting the growth claims. OpenAI says Codex accounted for roughly a fifth of EU output tokens in May and 34% by Sept. 30. ChatGPT Work launched in July and reached 21% of output tokens by the end of September. The company’s chart therefore records both increased use of Codex and the arrival of a new product during the measured period. OpenAI says a major part of the increase in small-business active users followed ChatGPT Work’s introduction.

OpenAI CFO Sarah Friar described the finding as evidence of agents’ growing role among the company’s EU business customers. The report’s broader conclusion is that Europe should turn wider adoption into deeper use, including through worker skills and more accessible advanced capabilities. Those are OpenAI’s interpretation and policy recommendations, distinct from the account-level measurements.

Limits of the EU Account Data

The report does not say how many accounts or users are represented in each country, or give absolute user counts for the small-business growth figures. That makes it difficult to judge how much a small number of large customers may influence national shares, or how large the increase was in headcount. The data also cover OpenAI’s business products only, so they cannot establish the prevalence of agent use among all EU companies.

It remains unclear how much of the rise reflects sustained changes in business practices, the launch and uptake of ChatGPT Work, changes in customer mix, or a combination of those factors. The 55% figure is a token share, not an interaction count; OpenAI does not provide a conversion from tokens to tasks, time saved, productivity or financial return. The report also does not establish that the difference between EU and non-EU frontier firms causes, or corresponds to, a wider economic performance gap.

What Further Data Could Show

OpenAI’s report sets out a case for deeper business adoption and improved access to advanced AI tools, but it does not specify a single next measurement or policy deadline. Further reporting would be needed to see whether agent use continues to grow after the initial launch period and whether activity spreads beyond the company’s existing business customers.

Useful follow-up data would include absolute user and account counts, country-level sample sizes, longer-term adoption trends and measures tied to completed tasks or business outcomes. Until those are available, the 55% figure is best understood as OpenAI’s measurement of output tokens in its EU business accounts through Sept. 30, not as a measure of how much European businesses overall rely on AI agents.

Key Questions

What does OpenAI’s 55% figure measure?

It is the share of output tokens generated by Codex and ChatGPT Work in OpenAI’s EU ChatGPT Enterprise and Business accounts by Sept. 30, 2026. It is not the share of users, tasks or business value.

Which tools does OpenAI count as agentic?

The report counts Codex, OpenAI’s coding agent, and ChatGPT Work, an agent mode launched in July 2026. Standard ChatGPT output is counted separately.

Did agent use grow 8.1 times across all EU businesses?

No. OpenAI reports an 8.1-fold rise in weekly active users of Codex or ChatGPT Work among its business accounts with 200 or fewer users, comparing the first week of May with the week ending Oct. 5. It does not report absolute user counts, and the data do not cover every EU business.

Does the report show that agents make businesses more productive?

No. The report measures product use and generated tokens, not productivity, time saved or financial returns. OpenAI itself cautions that tokens are an imperfect measure of business value.

Source: ThorstenMeyerAI.com