Agentic AI In EU Business: What OpenAI’s ‘Most’ Claim Leaves Out

🔍 Read the full analysis: Agentic AI In EU Business: What OpenAI’s ‘Most’ Claim Leaves Out on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get everyday essentials 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 reported Oct. 14 that its AI agents generated 55% of output tokens in ChatGPT Enterprise and Business accounts across the European Union by Sept. 30. The finding describes activity among OpenAI’s business customers, not the share of all European business AI use that is agentic, and it does not measure tasks completed or value created.

OpenAI Economic Research’s 10-page report, Europe’s Agentic Opportunity: Moving from Breadth to Depth with AI Agents, covers accounts in the 27 EU member states. It counts Codex, OpenAI’s coding agent, and ChatGPT Work, an agent mode launched in July, as agentic tools. By Sept. 30, Codex accounted for 34% of output tokens and ChatGPT Work for 21%; standard ChatGPT accounted for the remaining 45%.

OpenAI also reported that weekly active users of Codex or ChatGPT Work among businesses with 200 or fewer users rose 8.1 times between the first week of May and the week ending Oct. 5. The company said every member state recorded at least fivefold growth. Belgium led at 12.1 times, followed by Italy at 10.8 and Luxembourg at 10.0. The report does not give absolute user counts for this measure.

The report’s country figures vary widely: the agentic share of output ranged from 85% in Lithuania and 81% in Cyprus to 22% in Slovenia and 26% in Austria. Germany recorded 51%, ranking 17th among the 27 states. OpenAI defines frontier firms as the top 10% by output tokens per active user. By that measure, EU frontier firms generated about four times the tokens per user of typical EU firms, while frontier firms outside the EU generated more than twice as many as the EU frontier.

What the Token Share Can Show

The report offers a view into how OpenAI’s business customers are using its newer tools, but token volume is not a measure of economic impact. An agent can read documents, plan, write and revise before returning a result, generating far more text than a short chat response. OpenAI’s Enterprise Signals page calls tokens an imperfect measure of business value and notes that agentic tasks can require substantially more computation.

That distinction matters for business leaders and policymakers using the figures to judge adoption. A majority share of tokens does not establish that agents handle most work, reach most employees or deliver a measurable return. The figures do suggest that a growing amount of computation within these accounts is going to agent tools. OpenAI’s policy argument is that Europe should turn broader adoption into deeper use, through skills, confidence and access to advanced capabilities. That is the company’s interpretation of its data, not a finding that the report establishes about Europe’s economy as a whole.

A Launch Shapes the Growth Window

The comparison period for small-business growth begins in the first week of May, before ChatGPT Work launched in July. The report says a major part of the increase followed that launch. In the output-token figures, Codex rose from roughly one-fifth of EU account output in May to 34% by Sept. 30, while ChatGPT Work added 21 percentage points after its introduction. The launch therefore accounts for a substantial part of the reported change, though the supplied figures do not isolate its effect from other factors.

OpenAI’s Global Affairs team published the report alongside a global Enterprise Signals page. The report frames Europe’s challenge as moving from broad adoption to deeper business use. It also says EU firms account for about 10% of frontier firms worldwide, compared with about 65% in North America. Those figures use OpenAI’s customer data and its definition of frontier firms; they are not a census of companies in either region.

What the EU Figures Do Not Measure

The report does not disclose country-level account or user counts, making it difficult to assess how representative each national percentage is. Smaller countries’ results could be shaped by a limited number of large customers, a possibility OpenAI itself raises when discussing high agentic use among larger firms in countries including Cyprus, Estonia and Lithuania. The report does not provide enough detail to test that explanation.

Nor does the 55% figure reveal how many interactions, employees or business processes involved an agent. The comparison between token shares and task shares cannot be calculated from the data provided. One illustrative estimate in the source article assumes an agent run produces 10 to 30 times the output of a chat reply; that is the author’s assumption, not a finding in OpenAI’s report. Absolute user numbers, customer mix and business outcomes are also not supplied, so the reported growth cannot by itself establish a broad shift across European small businesses.

The Next Test Is Business Impact

OpenAI has published adoption and usage measures, but the report does not say when it will release account counts, country sample sizes or evidence linking agent use to business outcomes. Those details would help readers judge whether the patterns reflect widespread changes or activity concentrated among particular customers.

For now, the report’s next stated challenge is to deepen use by building worker skills and confidence and making advanced tools more accessible and affordable. Whether that produces sustained adoption or measurable productivity gains remains to be tested. Future reporting will need comparable measures over time and clearer information about the customers behind the figures to show whether the trend extends beyond OpenAI’s own accounts.

Key Questions

What does OpenAI mean by saying agents produced 55% of EU business output?

It means Codex and ChatGPT Work generated 55% of output tokens in the covered ChatGPT Enterprise and Business accounts by Sept. 30, 2026. It does not mean agents handled 55% of tasks or users.

Which tools did the report count as agents?

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

How fast did agent use grow at small businesses?

OpenAI reported an 8.1-fold rise in weekly active users of Codex or ChatGPT Work at businesses with 200 or fewer users, comparing the first week of May with the week ending Oct. 5. It did not provide absolute user counts.

Does the report show that agents improved business productivity?

No. The report describes usage and output-token measures. It does not establish that agent use improved productivity, reduced costs or generated a particular business return.

Why should country rankings be treated cautiously?

OpenAI did not publish account or user sample sizes by country. The rankings describe its customer accounts and may not represent the level of AI readiness across each national economy.

Source: ThorstenMeyerAI.com