OpenAI surpasses 1 million business customers

OpenAI says more than 1 million companies now pay to use its artificial intelligence services. The figure includes organizations that use ChatGPT for work or access its models directly through the developer platform.

The company announced the figure on November 5, 2025, and listed American Express, Amgen, Booking.com, Cisco, Lowe’s, Morgan Stanley, T-Mobile, Target and Thermo Fisher Scientific among its customers. According to OpenAI, its tools are already used in sectors including banking, healthcare, retail and customer service.

ChatGPT reaches more teams

Enterprise growth is building on ChatGPT’s popularity among consumers. More than 800 million people use it every week, according to OpenAI, which reduces friction when a company tries to introduce it: many employees already know the tool and need less training.

ChatGPT for work now has more than 7 million paid seats, up 40% from two months ago. Within that figure, ChatGPT Enterprise licenses have grown ninefold in one year.

The difference matters. This is not just about one person using a chatbot on their own. It is about a company paying to integrate it into its operations, with access controls, connections to internal data and tools for teams.

From answering questions to working with company data

OpenAI is also expanding what its products can do inside an organization:

• Company knowledge allows ChatGPT to access information from Slack, SharePoint, Google Drive, GitHub and other tools. It can analyze that data, answer with citations and, in some cases, take action.
• Codex, its model for generating and modifying code, has increased its usage tenfold since August. OpenAI says Cisco cut code review time by 50% and shortened projects from weeks to days.
• AgentKit aims to make it easier to build enterprise agents, systems that can execute tasks by following instructions and connecting to different tools. Carlyle says it cut development time for a system that analyzes financial transactions by more than 50% and improved its accuracy by 30%.
• Its models for text, images, video and voice can create everything from visual content with the image generation API and Sora 2 to telephone customer service agents with gpt-realtime and Realtime API.

These figures come from OpenAI and describe specific results from individual companies, not a guarantee for every organization. Performance depends on the use case, the available data and how the technology is implemented.

What results companies are seeing

OpenAI cites a recent Wharton study in which 75% of surveyed companies reported a positive return on their AI investment, while fewer than 5% reported a negative return. This is a finding from one specific study and does not represent every company in the market.

The examples shared by the company include:

• Indeed uses OpenAI APIs in its Invite to Apply feature and says it has achieved 20% more applications and 13% more hires.
• Lowe’s gives employees at more than 1,700 stores guidance for projects through the Mylow Companion app.
• Intercom uses OpenAI models in its Fin customer service agent and says it has reduced development cycles from quarters to days.
• Databricks will integrate OpenAI models alongside the business data its customers already store on the platform.

Companies are also building on OpenAI

The shift is not limited to internal use. Companies such as Canva, Figma, Zillow and Spotify have connected their apps directly to ChatGPT so users can access them from there.

In commerce, Shopify, Etsy, Walmart, PayPal and Salesforce are developing conversational shopping experiences through the Agentic Commerce Protocol. The idea is that you can discover products, compare options and move forward with a purchase within a conversation instead of jumping between several pages.

For you, the most likely change is not that every company will immediately replace its tools. It is that AI will start appearing inside specific tasks: reviewing documents, responding to customers, writing code, searching internal information or preparing a purchase. What OpenAI needs to prove now is that this use can continue at scale, with measurable results, protected data and controlled costs.