OpenAI’s enterprise revenue surpasses 40% as AI agents become integral to business workflows

OpenAI says enterprise products now account for more than 40% of its revenue, a sign that the company’s push into business software is gathering pace as it leans harder into AI agents and automated workflows.

Denise Dresser, OpenAI’s chief revenue officer, said in a company note on Wednesday that the shift has been striking. She said the firms moving fastest have gone far beyond using AI for routine tasks such as drafting emails or condensing documents and are instead deplo...

ying coordinated groups of agents that can preserve context, work across sessions and act inside business systems with limited human supervision.

OpenAI said its annualised revenue reached $25 billion in February, up from $20 billion at the end of 2025, and it expects enterprise revenue to draw level with consumer revenue by the end of 2026. That trajectory suggests the company sees business adoption, rather than mass-market subscriptions alone, as central to its next phase of growth.

The strategy is increasingly built around agents as a default interface for companies. OpenAI has launched tools aimed at helping businesses create and manage these systems, including Frontier, a platform for building and governing enterprise agents. The company also rolled out ChatGPT Agent, which it says can handle tasks such as trip planning, hotel bookings, competitor research, slide creation and online purchasing without direct intervention.

Dresser argued that many organisations are still missing a simple way to deploy agents as effective teammates without having to redesign their operations from scratch. OpenAI’s pitch is that its platform can bridge that gap.

The company’s internal figures underline how quickly the market is developing. Dresser said Codex, OpenAI’s coding agent, has passed 3 million users, having been close to zero at the start of the quarter. Paying business users reached 9 million in February, up from 5 million in August, while weekly active users across OpenAI’s products climbed to 910 million.

OpenAI’s own guidance on agents suggests the company is still encouraging businesses to start small. In a practical guide published on its website, it recommends first maximising what a single agent can do before splitting work across several systems. It describes two common designs: a central manager agent coordinating specialist agents, and a decentralised setup in which agents operate as peers and hand work off according to their strengths.

That technical framing fits a broader industry move towards multi-agent systems. Analysts and executives have increasingly portrayed them as the next stage of enterprise AI, with specialised systems breaking complex work into smaller tasks. OpenAI chief executive Sam Altman has positioned such architectures at the heart of the company’s future product roadmap.

The company’s hiring also points in the same direction. It recently brought in Peter Steinberger, founder of the open-source agent platform OpenClaw, to help lead its efforts in personal AI agents, indicating that OpenAI’s ambitions extend beyond corporate customers.

The business momentum comes as OpenAI is also preparing for a possible public listing. Chief financial officer Sarah Friar confirmed this week that retail investors will be included in the share allocation. OpenAI has said it is targeting $85 billion in revenue by 2030, a scale that would depend heavily on agents becoming embedded in everyday business operations rather than remaining an experimental add-on to chat-based tools.

For now, the message from OpenAI is clear: the company believes the enterprise market is moving from curiosity to infrastructure, and that the real test is no longer whether businesses will use AI, but how deeply they will wire agents into the way they work.

Source: Noah Wire Services