For most of its existence, OpenAI was a consumer story. ChatGPT’s 100-million-user surge, the viral moments, the endless comparisons with Google Search. Enterprise customers were there, but they were not the headline.

That has changed. On April 8, OpenAI published an update titled “The Next Phase of Enterprise AI,” confirming that enterprise now accounts for more than 40% of the company’s total revenue — and that it is on track to reach parity with consumer revenue by the end of 2026.

For business owners and data leaders wondering whether enterprise AI is still in pilot mode or has crossed into real production: OpenAI is answering that question with its own financials.

What OpenAI Actually Announced

The April 8 update covered three substantive areas.

Revenue composition. Enterprise has crossed 40% of OpenAI’s total revenue. This is notable both as a number and as a trajectory. As of early 2025, enterprise was a much smaller share. The acceleration reflects a broader market reality: businesses are no longer experimenting with AI — they are deploying it, and paying for it at scale.

OpenAI Frontier. OpenAI formalized Frontier as its end-to-end platform for enterprises to build and manage AI agents. Frontier is designed to give large organizations a governed environment for deploying agentic workflows — with controls, audit trails, and integration into existing enterprise systems. Think of it as the management layer that enterprises have been asking for since agents became capable enough to run real processes.

GPT-5.4 in agentic workflows. GPT-5.4 is driving what OpenAI describes as record engagement across agentic workflows. The model is being used inside Frontier and ChatGPT Enterprise to handle multi-step tasks — things like research, drafting, data extraction, and process execution — without requiring human intervention at each step.

Why the 40% Number Matters

Enterprise revenue is structurally different from consumer revenue. A consumer subscription is $20 a month, and a user can cancel in 30 seconds. An enterprise contract is typically annual, with IT integration, procurement approval, and a committed spend. When OpenAI says enterprise is 40% and rising, it means the business is becoming more durable, more predictable, and more deeply embedded in its customers’ operations.

It also means OpenAI is competing directly for the same procurement budgets as Salesforce, SAP, Microsoft, and every other enterprise software vendor. That competition will shape the next two years of the AI market more than any benchmark or model release.

The contrast with Anthropic is worth noting. Anthropic’s enterprise share is estimated at roughly 80% of its total revenue — a much higher proportion, built primarily through API access and Claude integrations in tools like Cursor, VS Code extensions, and enterprise platforms. OpenAI’s 40% is lower, but OpenAI is starting from a larger total revenue base ($25B ARR at last reporting), so the absolute dollars flowing from enterprise to OpenAI are substantial.

Both companies are betting that the enterprise market is where AI monetization ultimately concentrates. Based on current trajectories, they are probably right.

The Frontier Platform: What It Actually Solves

The agents problem in enterprise has never really been about the models. Models got capable fast. The problem has been governance: how do you give an AI agent access to real systems without losing control of what it does?

OpenAI Frontier addresses this directly. The platform is designed around three enterprise requirements:

Deployment control. Organizations can define what agents can access, what they can execute, and under what conditions they escalate to a human. This is the permission and sandboxing layer that most businesses have had to build themselves when deploying agents via API.

Audit and compliance. Frontier maintains records of what agents did, why they did it, and what data they accessed. For regulated industries — finance, healthcare, legal — this is not optional. It is the table stakes for getting AI agents past the legal and compliance teams.

Integration. The platform connects to existing enterprise systems rather than requiring businesses to rebuild workflows around a new toolset. This matters because most enterprise AI failures happen at the integration layer, not the model layer.

OpenAI has not released detailed pricing for Frontier, but positions it as part of the ChatGPT Enterprise family of products rather than a separate offering.

What This Means for Your Business

The practical implications depend on where your organization sits in its AI journey.

If you have not yet deployed AI agents: The fact that OpenAI is formalizing an enterprise-grade management layer, and that Anthropic launched Managed Agents last week, means the deployment infrastructure is maturing fast. The barrier that previously stopped many businesses from moving past pilots — the difficulty of governing agents in production — is being addressed at the platform level.

If you are already deploying agents: The emergence of Frontier and similar platforms is worth evaluating as an alternative to custom-built agent infrastructure. Building your own sandboxing, permission management, and audit trails is expensive and time-consuming. If a platform provides those capabilities out of the box, the build-vs-buy calculus shifts.

If you are a data or AI leader in a larger organization: The enterprise revenue figures from OpenAI and Anthropic signal that your C-suite is going to encounter these numbers. They will be asked why their company is or is not capturing the productivity gains that other enterprises are evidently experiencing. Having a clear answer to that question — and a credible roadmap — is increasingly important.

The Broader Pattern

The past week has been unusually eventful in enterprise AI. Anthropic launched Managed Agents (April 9), Project Glasswing went live with 12 major tech partners (April 7), OpenAI published its enterprise inflection point update (April 8), and both companies are racing to deploy cybersecurity-grade AI capabilities.

The common thread is that enterprise AI is moving from a market of experiments to a market of operations. The companies that treated AI as something to evaluate are now watching competitors treat it as something to run.

The question for most businesses is not whether to adopt AI agents. The question is how to do it in a way that is governed, measurable, and connected to real business outcomes rather than just capability demonstrations.

That is a strategy question before it is a technology question. Enterprise DNA works with business leaders at exactly that stage — helping organizations build the data literacy, the AI understanding, and the operational frameworks that make AI adoption durable rather than just fast.

Enterprise DNA put together a free field guide on exactly this: the full Claude ecosystem, Claude Code, and how to roll agents out without breaking things. Get the guide.