Does OpenAI’s EU Business AI Claim Show How Widespread Agents Are?

🔍 Read the full analysis: Does OpenAI’s EU Business AI Claim Show How Widespread Agents Are? on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

• 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 its AI agents generated 55% of output tokens in its ChatGPT Enterprise and Business accounts across the European Union by September 30, according to a report published October 14. The figures offer a detailed view of activity among OpenAI’s business customers, but do not establish how widely agents are used across European businesses or how much value they create.

OpenAI Economic Research’s report, Europe’s Agentic Opportunity: Moving from Breadth to Depth with AI Agents, counts output from Codex, the company’s coding agent, and ChatGPT Work, an agent mode launched in July that can work with files, browsers and company tools. By September 30, Codex accounted for 34% of output tokens in the covered accounts and ChatGPT Work for 21%. Standard ChatGPT accounted for the remaining 45%.

The company also reports fast growth in smaller businesses. Among accounts with 200 or fewer users, weekly active users of Codex or ChatGPT Work increased 8.1 times from the first week of May to the week ending October 5. OpenAI says every EU member state recorded at least fivefold growth. Belgium, Italy and Luxembourg had the largest reported multiples, at 12.1, 10.8 and 10.0 times, respectively. The report does not provide absolute user counts for this measure.

Use also varied by country. The agentic share of output tokens ranged from 85% in Lithuania and 81% in Cyprus to 22% in Slovenia and 26% in Austria. Germany’s share was 51%, placing it 17th among the 27 member states. These are figures from OpenAI’s accounts, not estimates of each country’s overall business use of AI.

What the Token Share Can Show

The report matters because it provides one of the clearest public snapshots of how a major AI provider’s business customers are using agent products. A rising share of agent-generated output suggests that some companies are assigning these tools work that can involve multiple steps, such as reading files, drafting or coding, and checking results. It also signals a change in the mix of activity taking place on OpenAI’s business platform.

But output tokens are not tasks, users or business value. Agent runs can produce much more text than a short chat exchange because they may plan, use tools and revise their work. OpenAI’s Enterprise Signals page itself describes tokens as an imperfect measure of business value and notes that agentic tasks can require substantially more computation. The 55% figure therefore means agents accounted for more than half of measured output—not that most interactions or business activity were agent-led.

That distinction matters for policymakers and business leaders weighing whether AI adoption is spreading or improving productivity. The figures can inform questions about product uptake among OpenAI customers, but they cannot, on their own, establish economy-wide adoption, job effects, cost savings or returns. OpenAI’s broader message is that Europe should move from wider adoption to deeper use, including building workers’ skills and making advanced tools more accessible and affordable.

How the Report Defines Agent Use

The report covers ChatGPT Enterprise and ChatGPT Business accounts in the EU’s 27 member states. Its definition of “agentic” is specific to two OpenAI products: Codex and ChatGPT Work. It should not be read as a count of all AI agents offered by other vendors, internally built company systems, or every form of automated work across Europe.

The timing also shapes the growth figures. ChatGPT Work launched in July 2026, after the small-business comparison period began in early May. OpenAI’s report says a major part of the increase followed the product’s introduction. The company’s chart shows Codex’s share of EU output tokens rising from roughly one-fifth in May to 34% by September 30, while ChatGPT Work added 21 percentage points after launch. The product release is therefore part of the story behind the change, rather than evidence by itself of a broad economic shift.

OpenAI also compares firms at the high end of its own usage data. It defines “frontier firms” as the top 10% by output tokens per active user. In September, EU frontier firms produced about four times as many tokens per user as typical EU firms; frontier firms outside the EU produced more than twice as many as the EU frontier. OpenAI says EU companies represent about 10% of frontier firms worldwide, compared with about 65% in North America. These comparisons concern token output within the company’s customer data.

What OpenAI’s Data Cannot Establish

OpenAI does not publish country-level sample sizes for the charts, or absolute user counts behind the 8.1-times small-business growth figure. That makes it difficult to judge how representative individual country results are or how much a small starting base influenced the growth multiple. The figures may reflect differences in the mix of OpenAI customers as well as differences in use across national economies.

The report also does not show what share of all EU businesses use agents, how many tasks they perform, or whether those tasks produce measurable savings or better results. Its data covers OpenAI business accounts and two products, and the token measure can rise when agents generate longer outputs or use more computation. The report’s comparisons do not establish why the EU frontier trails non-EU frontier firms, or whether the gap reflects access, company size, industry mix, spending or other factors.

OpenAI’s figures are company-reported measures of its own products. No independent validation or comparable account-level data from other providers is included in the supplied report material. As a result, the findings offer a view of activity on OpenAI’s platform, not a census of AI-agent adoption in Europe.

Evidence Needed Beyond OpenAI

The next step for readers assessing Europe-wide adoption is to compare OpenAI’s account data with independent surveys, company-level research and measures that track tasks and outcomes rather than tokens alone. Published sample sizes, absolute user counts and consistent definitions of “agent use” would help show whether the reported growth extends beyond OpenAI’s customer base and how adoption differs by company size and industry.

OpenAI’s report points to a continuing policy and business focus on moving from experimentation to more sustained use, including worker training and access to advanced capabilities. The material provided does not set out a next data-release date or a specific follow-up measurement. Until comparable evidence is available, the 55% figure is best understood as a measure of OpenAI agent output within its EU business accounts, not a verdict on how widely agents have spread through Europe’s economy.

Key Questions

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

OpenAI says Codex and ChatGPT Work generated 55% of output tokens in the EU ChatGPT Enterprise and Business accounts covered by the report as of September 30, 2026. It is a share of generated text, not a share of users, tasks or economic value.

Does the report show that most EU businesses use AI agents?

No. The report covers OpenAI’s business accounts in the EU, not every business in the region. It does not provide an economy-wide estimate of how many companies use agents.

How did small-business agent use change?

OpenAI reports that weekly active users of Codex or ChatGPT Work in accounts with 200 or fewer users rose 8.1 times between the first week of May and the week ending October 5. The report gives no absolute user total, and says a major part of the growth followed ChatGPT Work’s July launch.

Which EU countries had the highest agentic output share?

In OpenAI’s account data, Lithuania had the highest reported share at 85%, followed by Cyprus at 81%. Slovenia had the lowest at 22%. OpenAI does not provide country-level sample sizes, so these figures should not be treated as a ranking of national AI readiness.

What remains unknown about the findings?

The report does not establish how many EU businesses use agents, how many tasks agents handle, or whether use produces measurable productivity gains. It also does not give absolute user counts for the small-business growth measure or independent data for comparison.

Source: ThorstenMeyerAI.com