Before Taking OpenAI’s Agentic AI Claim At Face Value, Check The Denominator

🔍 Read the full analysis: Before Taking OpenAI’s Agentic AI Claim At Face Value, Check The Denominator 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 said Codex and ChatGPT Work generated 55% of output tokens in its ChatGPT Enterprise and Business accounts across the EU by Sept. 30, 2026. The figure, in a report published Oct. 14, measures the company’s generated text and code—not the share of European businesses, user interactions or economic value now involving AI agents.

OpenAI Economic Research’s report, Europe’s Agentic Opportunity: Moving from Breadth to Depth with AI Agents, covers business 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. Codex accounted for 34% of output tokens by Sept. 30 and ChatGPT Work for 21%; standard ChatGPT accounted for the remaining 45%.

OpenAI also reported that weekly active users of Codex or ChatGPT Work in businesses with 200 or fewer users grew 8.1 times from the first week of May to the week ending Oct. 5. The company said every member state grew at least fivefold, with Belgium at 12.1 times, Italy at 10.8 times and Luxembourg at 10 times. It did not provide absolute user counts for this measure.

Agentic output shares varied across countries, from 85% in Lithuania and 81% in Cyprus to 22% in Slovenia and 26% in Austria. Germany recorded 51%, which the report places 17th among the 27 countries. OpenAI defined “frontier firms” as the top 10% of firms by output tokens per active user. By that measure, EU frontier firms produced about four times the output of typical EU firms, while frontier firms outside the EU produced more than twice the output of EU frontier firms.

What Token Share Can Tell Businesses

The report offers a view into how customers of one AI provider use its products, but token volume is not a direct measure of adoption or outcomes. Agents may read documents, plan, generate code and check results in multiple steps, producing more output than a short chat response. OpenAI’s Enterprise Signals page itself describes tokens as an imperfect measure of business value and says agentic tasks can require substantially more computation.

That distinction matters for decision-makers comparing AI use across companies or countries. A high output-token share could reflect sustained agent use, but it can also reflect the length and computational demands of agent runs. It does not establish how many tasks agents handle, how many workers rely on them, or whether the work saves time or improves results. The data is best read as a measure of activity within OpenAI’s business accounts—not a census of European business adoption.

The report’s other finding points to a separate issue: EU frontier firms trail their counterparts elsewhere on OpenAI’s output-per-user measure. That is a signal about use among this provider’s customers, not a complete measure of European productivity or AI capability. Still, it frames OpenAI’s policy argument: it says Europe needs to move from broader adoption to deeper use, worker skills and more affordable access to advanced tools.

How OpenAI Defines Agent Use

OpenAI published the report alongside its Global Affairs post, “The EU’s next chapter in AI is agentic,” and its Enterprise Signals page provides a worldwide comparison. The figures are based on OpenAI’s own customer accounts, so they do not represent all EU companies or all AI providers. The underlying report focuses on ChatGPT Enterprise and Business accounts rather than a representative survey of businesses across Europe.

The timing also affects interpretation. Codex already contributed output in May, but ChatGPT Work was launched in July and reached 21% of EU account output tokens by Sept. 30. OpenAI’s chart indicates that Codex rose from roughly a fifth of output tokens in May to 34% in September, while the newly launched ChatGPT Work added a substantial share during the period. The reported 8.1-fold small-business increase starts in the first week of May, before Work was available; OpenAI says a major part of the growth followed the launch.

What the EU Figures Leave Out

OpenAI has not published the absolute account or user counts behind the country comparisons, nor enough detail to assess how representative each country’s customer mix is. The report’s country rankings could be affected by a small number of large customers, particularly in smaller markets; the available figures do not show how much.

The data also does not establish what share of interactions or business tasks were handled by agents, how much value those tasks produced, or whether firms achieved measurable productivity gains. The 8.1-fold figure is a relative change from a stated May baseline, but without absolute user numbers readers cannot tell the scale of the increase. OpenAI’s figures describe activity on its products, and the company’s own caveats do not resolve these broader questions.

Finally, the EU-to-global frontier comparison uses output tokens per active user as its measure. The report does not show whether firms’ industries, task mixes, account sizes or usage patterns are comparable across regions. The reported gap is confirmed as an OpenAI metric; its causes and implications for overall regional competitiveness remain uncertain.

Evidence Needed Beyond Token Counts

OpenAI’s report calls for Europe to deepen AI use by building worker skills and confidence and making advanced tools more accessible and affordable. The company has not, in the material cited here, set out a timetable for updated results or additional country-level sample information. The next useful evidence would include absolute user counts, account coverage and task-level measures, alongside outcomes such as time saved or work quality.

For now, readers can treat the report as a dated snapshot of OpenAI business-account activity: EU data through Sept. 30 for output shares, and small-business active-user growth through the week ending Oct. 5. Further data from OpenAI—or independent research covering a broader range of companies and AI providers—would help establish whether the product-level growth translates into wider adoption and measurable business effects.

Key Questions

What does OpenAI mean by 55% of output being agentic?

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

How did OpenAI calculate small-business growth?

OpenAI compared weekly active users of Codex or ChatGPT Work at businesses with 200 or fewer users in the first week of May with the week ending Oct. 5. It reported 8.1 times growth, but did not give the absolute number of users.

Does the report show that European firms are becoming more productive?

No. The report measures product use and output tokens, not verified productivity, financial returns or work quality. OpenAI also describes tokens as an imperfect measure of business value.

What does OpenAI say about Europe’s leading AI users?

Using its definition of frontier firms—the top 10% by output tokens per active user—OpenAI says EU frontier firms generated about four times the output of typical EU firms. It reports that frontier firms outside the EU generated more than twice as much per user as the EU frontier group.

Source: ThorstenMeyerAI.com