A Statistical Reality Check On OpenAI’s EU Agentic AI Claim

🔍 Read the full analysis: A Statistical Reality Check On OpenAI’s EU Agentic AI Claim on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on monitors, keyboards and dev gear

• 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 said Codex and ChatGPT Work produced 55% of output tokens in its ChatGPT Enterprise and Business accounts across the 27 EU member states by September 30, 2026. The finding, published in a report on October 14, describes the share of text generated by those tools—not the share of users, work tasks or business value attributable to AI agents.

OpenAI Economic Research’s ten-page report, Europe’s Agentic Opportunity: Moving from Breadth to Depth with AI Agents, defines agentic use as activity in Codex, OpenAI’s coding agent, and ChatGPT Work, its agent mode for working with files, browsers and company tools. Codex accounted for 34% of output tokens and ChatGPT Work for 21%; ordinary ChatGPT accounted for the remaining 45%.

Among businesses 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 October 5, the company reported. Every EU member state recorded at least fivefold growth. Belgium, Italy and Luxembourg had the largest reported increases, at 12.1, 10.8 and 10.0 times, respectively. OpenAI did not give absolute user counts alongside these growth figures.

The report also found wide differences in agentic token share by country, from 85% in Lithuania and 81% in Cyprus to 22% in Slovenia and 26% in Austria. Germany stood at 51%, ranking 17th among the 27 member states. OpenAI defines “frontier firms” as the top 10% of firms by output tokens per active user; by that measure, EU frontier firms generated about four times the output of typical EU firms, while frontier firms outside the EU generated more than twice the EU frontier level.

What the Token Share Shows

The report offers a window into how OpenAI’s European business customers are using its products, and the rapid uptake of ChatGPT Work and Codex could matter to companies deciding whether to expand agent access or train staff to use the tools. But token volume is not a direct measure of economic impact. OpenAI itself describes tokens as an imperfect measure of business value and says agentic tasks can require substantially more computation.

An agent may read files, plan, produce code or text, review its work and repeat steps. That process can generate more output than a short chat response even if it represents only one task. The report’s 55% figure therefore supports a specific conclusion: agents account for more of the generated text in these accounts than ordinary ChatGPT does. It does not establish that most AI interactions, workers’ tasks or productive value in European businesses are agentic.

The growth figures also need their time window and product history. ChatGPT Work launched in July 2026, after the May baseline used for small-business growth. OpenAI’s chart shows it contributing 21 percentage points of the 55% September token share. That launch helps explain the increase, but the report does not show whether use will persist or translate into measurable gains in productivity, revenue or costs.

How OpenAI Built Its EU Picture

The report covers accounts using ChatGPT Enterprise and ChatGPT Business in the EU’s 27 member states. It is a company analysis of activity on OpenAI’s services, not a survey of all European companies or a measure of AI adoption across the wider economy. OpenAI’s separate Enterprise Signals page offers a global comparison, while the EU report focuses on the region’s opportunity to turn broad adoption into deeper use.

OpenAI’s policy message is that European businesses should move beyond experiments, develop workers’ skills and confidence, and make advanced tools more accessible and affordable. The report says EU frontier firms produce around four times the tokens per active user of typical EU firms, but firms outside the EU’s frontier produce more than twice the output of the EU frontier. It also says EU companies represent about 10% of frontier firms worldwide, compared with about 65% in North America. These are measures based on OpenAI’s own customer activity and definition of frontier firms.

What the EU Data Cannot Establish

OpenAI does not publish account or user sample sizes for each country, making it hard to judge how representative the national comparisons are. The company notes that smaller countries including Cyprus, Estonia and Lithuania have relatively high agentic use within larger firms. In those markets, a limited number of large customers could have a substantial effect on a country-level share.

The report also provides no absolute counts for the small-business user increase and does not establish how many distinct tasks or workers the token figures represent. The period spans the launch of ChatGPT Work, so the reported growth cannot by itself show whether businesses are undergoing a lasting change in how they operate. Nor does the comparison of output tokens demonstrate differences in productivity or value between EU and non-EU firms.

The country rankings point in different directions depending on the measure. For example, Slovenia had the lowest agentic token share but reported 9.2-fold small-business user growth; Lithuania had the highest share but 6.8-fold growth. Those figures do not establish that one country is ahead or behind in overall AI readiness. OpenAI’s customer mix, sample sizes and task-level outcomes remain unreported.

Evidence Needed Beyond Token Counts

The next useful evidence would show whether the reported growth continues after the initial ChatGPT Work launch period and provide absolute user counts, country-level sample sizes and clearer information about account composition. Those details would help readers distinguish broad adoption from changes concentrated among a small number of customers.

To judge whether agent use is improving business performance, companies and researchers would also need measures tied to tasks and outcomes—for example, how often agents complete work, how much human review is required, and whether use changes time, cost or quality. OpenAI’s report does not supply those measures. For now, its findings are best read as a snapshot of token generation and product use within OpenAI’s EU business accounts, not a verdict on the European economy.

Key Questions

What does OpenAI mean by a 55% agentic share?

It means Codex and ChatGPT Work generated 55% of output tokens in the covered EU ChatGPT Enterprise and Business accounts by September 30, 2026. It is not a claim that agents handled 55% of users’ tasks or created 55% of business value.

How quickly did small-business agent use grow?

OpenAI reported an 8.1-fold increase in weekly active users of Codex or ChatGPT Work among businesses with 200 or fewer users, from the first week of May to the week ending October 5, 2026. The company did not provide the corresponding absolute user count.

Did every EU country have the same level of agent use?

No. OpenAI reported agentic shares ranging from 22% in Slovenia to 85% in Lithuania. The report does not provide country-level account or user sample sizes, so the rankings should not be treated as a measure of national AI readiness.

Does the report show that agents improved productivity?

No. It reports product activity and generated tokens, not direct measures of productivity, costs, revenue or task quality. OpenAI says tokens are an imperfect measure of business value.

Source: ThorstenMeyerAI.com