Reading The Numbers Behind OpenAI’s EU Business Agentic AI Headline

🔍 Read the full analysis: Reading The Numbers Behind OpenAI’s EU Business Agentic AI Headline on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get privacy and security gear 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 said on Oct. 14 that its AI agents generated 55% of output tokens in its ChatGPT Enterprise and Business accounts across the 27 EU member states by Sept. 30, a figure that describes model output volume rather than how many tasks or businesses use agents. The report also records an 8.1-fold rise in weekly active agent users at businesses with 200 or fewer users since early May, while saying European frontier firms generate fewer tokens per user than comparable leading firms outside the bloc.

The figures come from OpenAI Economic Research’s report, Europe’s Agentic Opportunity: Moving from Breadth to Depth with AI Agents, released alongside a post from the company’s Global Affairs team. The data cover ChatGPT Enterprise and ChatGPT Business accounts in EU member states. OpenAI defines agentic use in this analysis as use of Codex, its coding agent, and ChatGPT Work, an agent mode launched in July 2026 that can work with files, browsers and company tools.

For output generated by those accounts through Sept. 30, OpenAI attributes 34% of tokens to Codex and 21% to ChatGPT Work. The remaining 45% came from standard ChatGPT use. The report’s measure is thus a share of generated tokens, not a count of interactions or a valuation of work completed. OpenAI’s Enterprise Signals page describes tokens as an imperfect measure of business value and says agent tasks can require substantially more computation.

At businesses with 200 or fewer users, weekly active users of Codex or ChatGPT Work increased 8.1 times between the first week of May and the week ending Oct. 5, the report says. All 27 member states recorded growth of at least fivefold. Belgium, Italy and Luxembourg had the highest reported increases, at 12.1, 10.8 and 10.0 times, respectively. OpenAI gives no absolute user counts for this measure and says a major part of the rise followed ChatGPT Work’s launch.

What the EU Usage Share Shows

The figures offer a rare, detailed public view of how one major AI provider’s business customers are using agent products across Europe. They suggest that agent tools are taking a substantial share of OpenAI’s generated output in those accounts and that active use grew quickly after a new product entered the market. For business leaders, the findings may help frame questions about adoption, employee skills and access to advanced tools.

But the headline is easy to overstate. Agents can produce many more tokens than a short chat exchange because they may inspect files, plan, draft and revise. A 55% share of tokens does not mean agents handle 55% of work, interactions or company AI use. The report also does not measure time saved, accuracy, revenue or other business outcomes. Its numbers show where output volume is concentrated inside OpenAI’s products, not whether that volume produced greater value.

OpenAI’s comparison of frontier firms points to another issue: adoption alone may not close the gap between leading companies. The report says EU firms in the top 10% by tokens per active user generated around four times as many tokens per user as typical EU firms, while frontier firms outside the EU produced more than twice the output of EU frontier firms. Those comparisons are about usage intensity, not a direct measure of productivity or competitive performance.

How OpenAI Defines Agent Use

OpenAI’s analysis is a vendor report based on activity within its own Enterprise and Business products; it does not represent every EU company or AI provider. The report labels firms in the top 10% of output tokens per active user as “frontier firms.” It says EU frontier firms make up about 10% of frontier firms worldwide, compared with about 65% in North America.

Across EU countries, the reported agentic share ranges from 85% in Lithuania and 81% in Cyprus to 22% in Slovenia and 26% in Austria. Germany is at 51%, which OpenAI ranks 17th among the 27 member states. The report’s policy argument is that Europe should move from broader adoption to deeper use by helping businesses develop worker skills and confidence and making advanced capabilities more accessible and affordable.

The timing matters when reading the growth figures. Codex accounted for roughly a fifth of EU output tokens in May, according to the report’s chart, while ChatGPT Work was not available until July and made up 21% by Sept. 30. The measured period therefore includes the launch and rapid uptake of a new OpenAI product, not just gradual changes in established business practices.

Limits in the Customer Data

OpenAI does not disclose the number of accounts or users in each country, so the country rankings cannot be assessed for sample size or representativeness. The figures describe OpenAI customers rather than national economies as a whole; large or unusually agent-heavy accounts could affect a country’s reported share. The report also provides no absolute user counts behind the 8.1-fold growth figure, making the starting scale unclear.

It remains unknown how much of the increased token use translates into completed work, improved quality, lower costs or higher productivity. The report’s frontier comparison does not establish why EU and non-EU firms differ, or whether the gap reflects company mix, product access, work patterns or other factors. Because ChatGPT Work launched during the period measured, the reported growth cannot by itself establish a lasting change in European small-business adoption.

Evidence Needed Beyond Usage

OpenAI’s report calls for businesses and policymakers to focus on moving from experimentation toward deeper agent use, including worker training and access to advanced tools. The report does not set out a specific timetable for later data releases. A clearer assessment would require absolute user and account counts, country-level sample information, and repeated measurements over a longer period after ChatGPT Work’s launch.

For readers evaluating the claims, the next useful evidence would connect product activity to workplace outcomes: what tasks agents complete, how often people review or revise their output, and whether time, cost or quality changes. Until those measures are available, the figures are best read as a snapshot of OpenAI product usage through early October 2026, not proof that agents now perform most business work in Europe.

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 a share of users, tasks or business value.

Which tools does OpenAI count as agentic?

The report counts Codex and ChatGPT Work. OpenAI says ChatGPT Work launched in July 2026 and can work with files, browsers and company tools.

How quickly did agent use grow at smaller businesses?

OpenAI reports that weekly active users of Codex or ChatGPT Work at businesses with 200 or fewer users rose 8.1 times from the first week of May to the week ending Oct. 5. It does not publish the absolute user counts in the source material.

Does the report show that agents improved productivity?

No. The report measures product activity and token output, not time saved, accuracy, revenue or productivity outcomes. Those effects remain unclear from the published figures.

Source: ThorstenMeyerAI.com