Last month, OpenAI proclaimed itself the “fastest-growing business platform in history” after hitting a milestone one million business customers, 800 million weekly ChatGPT users, and 7 million paid workplace accounts.

Millions of people are now using ChatGPT at work, either officially or otherwise, but for the first time, OpenAI is giving users a peek behind the curtain by sharing a comprehensive study on how enterprises are adopting its AI.  

Grounded in real-world usage data coupled with a survey of 9,000 workers across nearly 100 businesses, the AI firm’s State of the Enterprise AI report shows that adoption is not only accelerating but deepening, with firms leaning on the tech for more complex, high‑value work.

Three‑quarters of workers (75%) said that AI has allowed them to take on new tasks they couldn’t previously perform, ranging from programming support and code review to spreadsheet automation, technical troubleshooting, and even custom GPT or agent design.

According to OpenAI, the broadening of individual capabilities is most obvious among non-technical teams, who are now taking on coding, analysis, and data work once left to the IT experts. Among ChatGPT Enterprise users, coding‑related messages have risen across all functions, with non‑engineering teams posting a 36% increase in the past six months alone.

Across enterprises, 75% of workers reported that AI has improved either the speed or quality of their output, saving them up to an hour a day, with 87% of IT workers claiming faster issue resolution, and 73% of engineers quicker code delivery.

But firms aren’t just chasing productivity gains, they’re tapping into ChatGPT to design highly customised, controllable products and systems, moving away from experimentation toward full deployment.

In the last twelve months, OpenAI saw average token use per organisation surge, increasing 320×, which it said is evidence of more intelligent models being integrated across new products and services.

Overall, the report claims that 9,000 organisations have processed more than 10 billion tokens, with nearly 200 exceeding 1 trillion tokens. The adoption of Custom GPTs and Projects has risen 19× since the start of the year, with 20% of all Enterprise messages processed via a Custom GPT.

The findings suggest that even as companies rush to build bespoke AI tools, the heavy lifting will continue to be done by giants like OpenAI.

“Looking ahead, the next phase of enterprise AI will be shaped by stronger performance on economically valuable tasks, better understanding of organisational context, and a shift from asking models for outputs to delegating complex, multi-step workflows,” said Ronnie Chatterji, OpenAI’s chief economist.

“As these capabilities mature, we expect organisations to not only improve efficiency, but also discover new ways to serve customers and deliver value.”

Recommended reading

• Report: 96% of Enterprises Running AI Across Core Processes
• Enterprise AI Adoption Surges 282%, Finds Salesforce
• Will AI Agents Be the Enterprise Disruptor of 2026?

The publishing of this new report, one which serves as a reminder of how extensive OpenAI’s platforms have become, comes at a critical moment when the company finds itself with competitors nipping at its heels.

Anthropic now counts more than 300,000 business customers, with high‑value accounts generating over $100,000 in run‑rate revenue, a figure that has grown nearly seven‑fold in the past year. 

Meanwhile, Gemini has seen rapid adoption, with 70% of Google’s cloud customers using the AI, and the bot beating GPT models over a variety of benchmarks, culminating in industry heavyweights like Salesforce’s Marc Benioff declaring Gemini 3 so far ahead that he is abandoning the OpenAI chatbot.

The figures have rattled Sam Altman enough to declare a ‘code red’ and push for faster development of new AI models, predicting “temporary economic headwinds” for the company if it falls short.