OpenAI finds widening gap in enterprise AI adoption

Picture: ReutersOpenAI says the gap between companies making limited use of artificial intelligence and those deploying it extensively is widening, with its most active enterprise customers increasingly using AI to carry out tasks rather than simply answer questions. In two new studies examining enterprise.

By June, these companies were generating 8.3 times as many output tokens per active user as typical firms, compared with 2.6 times in January. The findings come from Enterprise Signals, which examines the use of agentic AI among OpenAI’s enterprise customers, and a companion.

What Happened

The research draws on millions of interactions and looks at how AI adoption varies across organisations, job functions and levels of seniority. One of the clearest shifts has been towards agentic AI, where systems can undertake multi-step tasks instead of simply providing.

• OpenAI classifies the top 10 per cent of enterprise customers by monthly output tokens per active user as “frontier” firms.

• OpenAI said Codex accounted for 64 per cent of the combined output tokens generated by Codex and ChatGPT among enterprise customers in June.

• Weekly Plugin usage stood at 21 per cent among active users at frontier firms, compared with 9 per cent at typical companies.

Key Details

The company said this reflects a broader move from asking AI for assistance to delegating parts of a job to it. Such systems can, for example, gather information from different sources, create files and produce a first version of a presentation for.

• Skills were used by 19 per cent of users at frontier firms, against 3 per cent at typical firms.

• OpenAI’s own workforce provides an indication of how much further such usage could go: the company said 95 per cent of its employees use Plugins each week.

• The aim, it said, is to turn successful individual uses of AI into repeatable processes across organisations.

Why It Matters

The adoption of more advanced capabilities is also concentrated among the companies making heavier use of AI. Agentic AI is also moving beyond software development, which was an early area of adoption.

• The company’s broader message is that businesses need to give AI systems access to appropriate context and tools while maintaining permissions, human oversight and clear governance.

• Training, shared workflows, data infrastructure and governance could play a role in helping organisations move beyond individual experimentation.

What Reports Say

Coverage of the story so far points to:

• Continued reporting by t2ONLINE as more details emerge