Consumers Built ChatGPT, but Businesses Now Supply Over Half of OpenAI's Revenue
Individual users made ChatGPT a household name, yet most of OpenAI's revenue now comes from companies. The yardstick for valuing AI firms may shift from user counts to the work they actually do and what it earns.
The 20-second read
Who is AI's customer: the person using it, or the company paying for it?
When people explain ChatGPT's success, they usually picture individual users. The experience of asking a question and getting an answer brought AI into daily life.
But the center of revenue has already moved. OpenAI co-founder and president Greg Brockman said on Nov. 4 that more than half of OpenAI's revenue comes from business customers. He spoke at the Wall Street Journal's tech conference in Napa, California. He did not disclose the exact share or the revenue figure.
An individual keeps using a service if the answers please them, and looks elsewhere if not. Companies are different. If AI saves time, cuts errors and finishes tasks, there is a reason to pay every month.
If mass adoption created the users, business adoption is the test of whether recurring revenue follows.
An AI that writes code is not the same as an AI that finishes the job
One product OpenAI is pushing hard in the enterprise market is its coding agent, Codex. This is where the economic value of AI changes.
A tool that helps: Developers write code with AI. The AI drafts, and people continue with edits, testing and documentation. What gets saved is mostly writing time.
A system that does the work: The AI handles everything from the fix through verification. It edits code, runs tests and writes documentation, and the developer checks the result and gives final approval. What gets saved is the whole work process.
Companies pay less for plausible sentences than for completed work. But the more steps the AI takes on, the more computing cost for execution and verification must be counted too.
The competitiveness of enterprise AI may be measured less by how impressive an answer is than by the range of work that can be trusted to it.
If OpenAI earns, does Microsoft earn too?
OpenAI is expanding its sales effort in enterprise AI, competing with Anthropic. But it is hard to name a single winner in this market.
Microsoft (MSFT) and Amazon (AMZN) compete with others over AI services, and can also supply cloud infrastructure and a route to corporate customers. Even as AI use grows, the profit does not all stay with the model developers.
Once companies pay for AI, the questions split in three. Which platform books the revenue? Who bears the computing cost? Where does profit remain after costs?
Cloud providers can gain from demand for computing, but they also carry the burden of capital spending. Model developers have a chance to secure customer relationships, but must absorb competition and the cost of execution.
Growth in AI revenue is not the same as growth in profit left for shareholders.
The numbers to read before a 2027 IPO
Asked whether OpenAI could be ready for an initial public offering in 2027, Brockman answered, in effect, that it could. That does not mean an IPO date has been set.
The prospect of a listing could become an occasion to compare and assess OpenAI's value. But a large number of business customers alone does not justify a high valuation. Investors need to check whether contracts keep renewing, and whether profit grows as the AI handles more work.
More important than listing hopes is how long enterprise revenue lasts, and at what profit margin.
- No. The consumer service is a channel for spreading the product experience. But in terms of revenue mix, it means business customers have become more important.
- Whether AI replaces developers is not the point of this article. The point is that value may shift from developers performing every step themselves to AI handling multiple steps while developers handle verification and approval.
- It is hard to say who is ahead from the information given. Beyond the opportunities in business customer access and cloud revenue, you also need to weigh compute costs and the burden of capital spending.
What to watch
Insight Times Editorial Desk
