OpenAI Says Most EU Business AI Is Agentic—But Check The Denominator

🔍 Read the full analysis: OpenAI Says Most EU Business AI Is Agentic—But Check The Denominator on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get movie nights 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 its agent products generated 55% of output tokens in ChatGPT Enterprise and Business accounts across the European Union by September 30, according to a report published October 14. The figure combines Codex and ChatGPT Work, but it counts model-generated text—not the share of business users, tasks or economic value that involved agents.

OpenAI Economic Research’s report, Europe’s Agentic Opportunity: Moving from Breadth to Depth with AI Agents, covers accounts in the 27 EU member states. It classifies Codex, OpenAI’s coding agent, and ChatGPT Work, an agent mode launched in July 2026, as agentic tools. The report says Codex accounted for 34% of output tokens and ChatGPT Work for 21% by September 30. The remaining 45% came from plain ChatGPT.

OpenAI also reports a sharp increase in weekly active users of Codex or ChatGPT Work at businesses with 200 or fewer users: 8.1 times the level recorded in the first week of May, measured in the week ending October 5. Every EU member state recorded at least fivefold growth in that measure, the report says. Belgium, Italy and Luxembourg had the highest reported multiples, at 12.1, 10.8 and 10.0 respectively. OpenAI does not provide absolute user counts alongside those growth rates.

The report compares the highest-output companies with typical firms. In September, 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. OpenAI says frontier firms outside the EU generated more than twice as many as the EU frontier. These are measures of activity on OpenAI’s products, not a direct comparison of productivity or business results.

What the Token Share Can Show

The report offers a view into how OpenAI’s business customers are using its products, and suggests that agent tools have become a substantial part of the company’s EU business-account activity. For companies deciding whether to adopt agents, the reported growth may signal rising interest and a widening gap between typical firms and OpenAI’s highest-use customers.

But the 55% figure is a token share, not evidence that agents now handle most business AI work in Europe. OpenAI’s Enterprise Signals page describes tokens as an imperfect measure of business value and notes that agentic tasks can require substantially more computation. An agent may inspect files, plan, generate output and revise it, producing more tokens than a brief chat exchange. A larger share of tokens can therefore reflect more intensive runs, without showing how many distinct tasks or users rely on agents.

The distinction matters for policymakers and business leaders who may use the headline to judge adoption or readiness. The report’s evidence describes activity among OpenAI customers; it does not establish how representative those customers are of all European businesses, or whether greater agent use improves results. OpenAI’s policy conclusion is that Europe should turn broad AI adoption into deeper use by building skills and confidence and making advanced capabilities more accessible and affordable.

How OpenAI Defines Agent Use

OpenAI released the report through its Economic Research team alongside a Global Affairs post titled The EU’s Next Chapter in AI Is Agentic. The report’s definition is specific to two OpenAI products: Codex and ChatGPT Work. ChatGPT Work, which can work with files, browsers and company tools, launched in July 2026; Codex is OpenAI’s coding agent. The worldwide Enterprise Signals page provides additional company data and cautions that token volume is not a direct measure of value.

The timing of the product launch is relevant to the growth figures. The small-business comparison starts in the first week of May, before ChatGPT Work existed, and ends in the week ending October 5. OpenAI says a major part of the increase followed ChatGPT Work’s introduction. The report therefore tracks changing use of a product set that expanded during the measurement period; the 8.1-fold rise should not be read on its own as proof of a broad, lasting change in the European small-business economy.

Country figures vary widely: Lithuania had an 85% agentic share of output tokens and Cyprus 81%, while Slovenia had 22% and Austria 26%. Germany recorded 51%, ranking 17th among the 27 member states. OpenAI provides no country-level account or user sample sizes in the material described, and says some smaller countries have relatively high agentic use within larger firms. The chart reflects the company’s customer mix as well as product use.

What the EU Figures Leave Out

The published figures do not show how many individual users, interactions or business tasks were agentic, nor the economic value or outcomes tied to those uses. OpenAI itself cautions that output tokens are an imperfect proxy for business value. The report does not provide the absolute small-business user totals behind the 8.1-fold growth, so the starting scale and the number of businesses represented cannot be determined from that multiplier alone.

It is also unclear how representative the accounts are of businesses across the EU. OpenAI does not give country-level sample sizes in the supplied report material, making the national rankings difficult to interpret as comparisons of whole economies. The data also cover OpenAI products only, not agent use across competing AI services or internally developed systems.

Finally, the period includes ChatGPT Work’s July launch. That makes it difficult to separate growth in underlying demand for agents from uptake of a newly available product. The report shows increased activity on OpenAI’s tools; it does not establish whether that growth will persist or deliver measurable gains in productivity.

Whether Agent Use Deepens

OpenAI’s report points to a policy and business agenda built around skills, confidence and access to advanced tools. The next useful evidence would include absolute user and task counts, clearer information about the customer samples behind country comparisons, and outcome measures that connect agent use with work completed or business results.

OpenAI has not, in the supplied material, announced a follow-up measurement date or a new dataset addressing those gaps. For now, the report is a snapshot through September 30 for output-token shares and through the week ending October 5 for small-business user growth. Whether the reported rise continues, and whether it translates into broader adoption or demonstrable economic value, remains to be seen.

Key Questions

What does OpenAI mean by “55% of output tokens”?

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

Which tools count as agentic in the report?

OpenAI counts Codex, its coding agent, and ChatGPT Work, an agent mode launched in July 2026. The report compares their combined output with plain ChatGPT.

What does the 8.1-fold small-business growth measure?

It is the reported increase 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 October 5, 2026. OpenAI does not provide the absolute user totals in the supplied material, and the period begins before ChatGPT Work launched.

Does the report prove EU businesses are getting more productive?

No. The report describes use of OpenAI’s products and token output. It does not establish productivity gains, business value or economic impact, and OpenAI says tokens are an imperfect measure of business value.

Source: ThorstenMeyerAI.com