OpenAI revenue hits $10 billion in a quarter for the first time, growing nearly 50%

Some news stories matter less for their details than for their scale. This is one of them: according to Bloomberg, OpenAI’s revenue in the third quarter, the three months ending in September, reached $10 billion for the first time. Growth was nearly 50%. A source cited by the publication attributed it to strong demand for the company’s latest AI products.

What we know, and what we don’t

The report contains relatively few facts, and it is worth separating the concrete figures from the caveats.

The concrete part: the $10 billion quarterly revenue figure and the claim that it is a first. The caveat: the source is described as “a person familiar with the matter,” meaning this is neither a financial report nor a company statement. OpenAI is not a publicly traded company in the conventional sense, so the figure cannot be checked against public quarterly financial statements. The report also does not specify whether the nearly 50% growth is measured against the previous quarter or the same period last year. Until that is clear, comparisons with other companies’ growth should be made cautiously.

Another important omission is profit. Revenue tells us nothing about how much the company spends on model training, inference, and computing capacity. High revenue and losses can easily coexist, so record sales do not mean record profitability.

What this means for the market

The main takeaway is straightforward: people are paying for AI models, and spending more. For founders and product managers, this eases one of the fears of recent years: that interest in generative AI consists of pilots and demos rather than recurring payments. When a leading lab grows by nearly half from an already substantial revenue base, demand for models as a service looks resilient.

There is a flip side. If you build a product on a single provider’s API, your business is tied to its economics. The stronger the supplier’s growth, the greater its market power over prices, limits, and terms. Yesterday’s cheap plan is no guarantee for next year.

What practitioners should do

One news story is no reason to overhaul your plans, but it does highlight a few useful habits.

Build a buffer into your unit economics. Include a scenario in your cost model where token prices rise or strict limits return. If your product can weather that scenario, it is resilient.

Avoid hardwiring your product to a single provider. A simple abstraction layer over models, a common prompt format, and your own quality tests can let you switch to another supplier or an open model in days rather than months if conditions change.

Watch what customers are paying for. The report says the latest products drove growth. That suggests customers are readily moving to newer versions. If your product depends on an outdated model, check whether you are falling behind users’ expectations.

Wait for confirmation. As long as the figures come from an anonymous source, treat them as a useful guide rather than an established fact. Business decisions that depend on the precise number are best made after official figures become available.

Takeaway

$10 billion in quarterly revenue is an impressive, headline-grabbing milestone, but it is still just a milestone. It tells us that the market for paid AI services is large and continues to grow. It does not tell us how profitable that market is, who will make money over the long term, or how prices will evolve. For anyone building on other companies’ models, the practical takeaway remains the same: benefit from a growing market, but do not hand a single supplier the keys to your own economics.