Is Most EU Business AI Agentic? OpenAI Says Yes, With A Caveat

🔍 Read the full analysis: Is Most EU Business AI Agentic? OpenAI Says Yes, With A Caveat on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

OpenAI says AI agents generated 55% of output tokens in its ChatGPT Business and Enterprise accounts across the European Union by Sept. 30, a finding that points to growing agent use among its business customers but does not show that most EU business AI tasks are agentic. The figures come from a report published Oct. 14 and describe OpenAI’s own customer accounts, not the wider European economy.

OpenAI’s Economic Research report, Europe’s Agentic Opportunity: Moving from Breadth to Depth with AI Agents, defines agentic use as activity from Codex, its coding agent, and ChatGPT Work, an agent mode that can work with files, browsers and company tools. By Sept. 30, Codex accounted for 34% of output tokens in the covered accounts and ChatGPT Work for 21%. The remaining 45% came from standard ChatGPT use.

The company also reports that weekly active users of Codex or ChatGPT Work at businesses with 200 or fewer users increased 8.1 times between the first week of May and the week ending Oct. 5. OpenAI says every EU member state grew at least fivefold over that period. Belgium recorded 12.1 times growth, Italy 10.8 times and Luxembourg 10 times. The report does not provide absolute user counts, so the growth multiples do not show how many small-business employees adopted the tools.

Agentic output shares varied considerably by country. OpenAI reports 85% in Lithuania and 81% in Cyprus, compared with 22% in Slovenia and 26% in Austria. Germany’s share was 51%, placing it 17th among the 27 member states. The report also 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. Frontier firms outside the EU generated more than twice the output per user of the EU frontier group.

What the Token Share Shows

The figures suggest that agent products are taking a substantial share of computation in the businesses already using OpenAI’s paid services. That matters to companies deciding whether to introduce agents for coding, document work or tasks involving company tools. The data also point to a gap in the intensity of use: even among OpenAI’s most active European business customers, the report says output per user trails that of comparable frontier firms outside the EU.

But a token share is not a direct measure of workplace adoption or economic benefit. Agents can generate long sequences of output as they inspect files, plan work, produce a result and check it. A single agent task may therefore use far more tokens than a short chat exchange. More tokens do not necessarily mean more tasks, more users or more value, and OpenAI’s Enterprise Signals materials describe tokens as an imperfect measure of business value.

The distinction matters for policymakers and employers. OpenAI’s conclusion is that Europe should turn broad AI adoption into deeper use by building employee skills and confidence and making advanced capabilities more accessible and affordable. The report’s data can inform that discussion, but they do not establish how widely agents are used across Europe’s businesses or whether the tools are improving productivity.

How OpenAI Defines Agentic Use

The report covers ChatGPT Enterprise and ChatGPT Business accounts in the 27 EU member states. Its definition of “agentic” is product-based: it counts Codex and ChatGPT Work, which launched in July 2026. It is not a survey of every AI system used by European businesses, nor a count of all software that might be described as an agent.

The timing affects the growth comparisons. OpenAI’s May starting point came before ChatGPT Work was available. The report’s chart shows Codex at roughly one-fifth of EU output tokens in May and 34% by Sept. 30, while ChatGPT Work contributed 21 percentage points after its launch. OpenAI says a major part of small-business growth followed that launch. The figures thus combine increased use of an existing tool with uptake of a newly released product.

OpenAI Global Affairs published the report with a separate global view on the company’s Enterprise Signals page. Both draw on OpenAI customer activity. The source material does not give country-level account counts or user counts, so national percentages may reflect differences in which customers use the services as well as differences in broader business adoption.

What the Data Cannot Establish

The report does not show what share of all EU business AI tasks is handled by agents. Its 55% figure is a share of output tokens in OpenAI’s covered accounts, not a share of interactions, users or business activity. OpenAI also notes that agentic tasks can require substantially more computation. The source material does not provide a measured conversion from tokens to tasks or economic value.

It is also unclear how many businesses or users sit behind the country figures. OpenAI gives no account counts by member state, limiting comparisons between countries and making it difficult to tell whether high shares reflect broad use or a small number of heavy users. The reported 8.1-times small-business increase has no published absolute starting or ending user total in the supplied material.

The report’s scope leaves other questions open: whether use persists after the initial period following ChatGPT Work’s launch, how much activity occurs in non-OpenAI tools, and whether the EU-to-global frontier gap reflects adoption, customer mix or other factors. The data alone cannot answer those questions or confirm productivity gains.

Whether Early Growth Persists

The next useful evidence would include follow-up data showing whether small-business use continues to rise after ChatGPT Work’s launch period, alongside absolute user counts and clearer measures of tasks completed and outcomes. Comparable information across countries would help distinguish national adoption patterns from differences in OpenAI’s customer base.

OpenAI’s report sets out a policy direction—more worker skills and confidence, and wider access to advanced capabilities—but does not announce a specific EU program or timetable. For now, the figures establish a marked shift in how OpenAI’s EU business accounts generate output; whether that shift represents broad adoption or measurable business gains remains unresolved.

Key Questions

What does OpenAI mean by agentic AI in this report?

OpenAI counts activity from Codex and ChatGPT Work. The latter can work with files, browsers and company tools. This is the report’s product-specific definition, not a count of every AI agent used in Europe.

Does the 55% figure mean most EU business AI tasks use agents?

No. It means the two agent products generated 55% of output tokens in the covered OpenAI business accounts by Sept. 30. Tokens measure generated text and related output, not the number of tasks, users or value created.

How much did small-business agent use grow?

OpenAI reports an 8.1-times increase in weekly active users of Codex or ChatGPT Work at businesses with 200 or fewer users, from the first week of May to the week ending Oct. 5, 2026. The report does not give absolute user totals in the supplied material.

Does the report show that AI agents improved productivity?

No productivity result is established by the reported token shares or growth figures. The report describes activity among OpenAI customers; it does not directly measure business outcomes or productivity gains.

Source: ThorstenMeyerAI.com