OpenAI’s 2025 ‘The State of Enterprise AI’ report provides an in-depth look at how businesses are using AI tools within real companies. Drawing on anonymized usage data from more than 1 million business customers, along with a survey of 9,000 workers at nearly 100 organizations, the report presents a picture of increased AI adoption and integration in the workplace. 

«Across surveyed enterprises, 75% of workers report that using AI at work has improved either the speed or quality of their output,» the report states. Also, the report says that «75% of users report being able to complete new tasks they previously could not perform.» 

However, the productivity gains might not be as universal and widespread as anticipated: on average, ChatGPT Enterprise users save less than an hour of time per day, according to the report.

Below is a breakdown of the report’s major findings.

Don’t miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source on Chrome.

Report shows productivity gains, but it’s not universal 

Despite the hype surrounding AI at work, the latest data from OpenAI suggests that the reality for most employees is modest. In its report, the company says that on average, ChatGPT Enterprise users save only about 40 to 60 minutes per active workday.

That’s not nothing, but it’s nowhere near the sweeping productivity overhaul that many hoped for. In a workday filled with meetings, emails and tool overload, an hour reclaimed can feel like a minimal benefit rather than a tidal shift in productivity.

(Disclosure: Ziff Davis, CNET’s parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)

A few key findings 

The report finds AI adoption within companies is growing fast. Weekly messages in ChatGPT Enterprise have increased nearly eightfold in the past year, and the use of structured workflows, such as custom GPTs, has risen 19 times. Companies are pushing more complex prompts, too, with reasoning-token usage increasing more than 320-fold.

But the outcomes don’t scale at the same rate. Workers say they complete certain tasks more quickly — like IT troubleshooting, campaign creation and coding improvements — yet the day-to-day gains still add up to roughly an hour on average.

A divide between heavy AI users and everyone else

OpenAI’s data shows a widening gap between «frontier» users — defined by OpenAI as those in the 95th percentile of adoption intensity — and the average worker, however.

Frontier employees send about six times more messages than average users. Unsurprisingly, these heavy users report bigger gains of over 10 hours a week. They build workflows around AI, automate routine tasks and turn the tool into a dependable co-worker instead of an occasional assistant. Though arguably, around 2 hours per day of saved time is still relatively moderate. 

OpenAI frames the report as a snapshot of where enterprise AI stands today, rather than a final verdict. The company suggests that future gains could come not from the model itself, but from how organizations reshape processes and workflows around it. 

But for most workers, AI is still a sidekick. Useful, but not transformative. It helps speed things up. It may even make some work less tedious. But the typical worker saving under an hour a day points to a technology that is powerful, yet still limited. The big question now is whether those numbers will keep climbing, or whether an hour a day is closer to the ceiling than AI enthusiasts want to admit.