OpenAI Grew 50% to $10 Billion in a Quarter, but Anthropic at $17 Billion Is Still Ahead

For two years running, the market has argued over who will win the AI lab race: the one with the smarter model or the one with better economics. The latest quarterly figures reported by the WSJ and The Information tip the scales toward the latter. OpenAI posted its sharpest growth in some time, yet still ended up behind Anthropic.

What the Numbers Show

OpenAI's third-quarter revenue grew roughly 50% from the previous quarter and reached $10 billion. In the second quarter, growth was just 17–18%. A year ago the company earned $3.9 billion, meaning revenue has grown 2.5 times in a year.

Profitability is a different story. The adjusted operating loss rose from $10.1 billion to $10.9 billion. That figure excludes stock-based compensation, which came to $2.6 billion. On the other hand, the margin improved: minus 150% in the second quarter versus minus 109% now. For comparison, a year ago it was minus 97%.

Anthropic's picture is different. It overtook OpenAI in revenue back in the second quarter and even showed a small operating profit then. In the third quarter, revenue grew 47% to $17 billion, while the operating loss, this time including stock-based compensation, was just $567 million. The metrics are calculated differently: OpenAI reports an adjusted loss, Anthropic a full one. So the real efficiency gap is most likely even larger than it appears at first glance.

Why It Turned Out This Way

The explanation is simple: different business models. Anthropic focused on enterprise customers from the start and won them over with Claude Code and the Claude Cowork agent. These are products companies are happy to pay for and that can be scaled without endless subsidies for users.

OpenAI simultaneously funds broad research, including robotics, and runs a huge consumer business. ChatGPT has 1.2 billion weekly active users and 100 million paying subscribers, with 22 million added over the quarter. Advertising in ChatGPT has already reached $1 billion in annual revenue. But serving such an audience is expensive.

To catch up in the enterprise segment, OpenAI shifted its priorities this year. The company shut down a number of projects, including a video app, added resources to Codex, hired Dali Rajic as head of sales and cut prices on new models. According to third-party data, companies were looking for cheaper alternatives to Claude, and that boosted API sales. Greg Brockman called what is happening a turning point and cited JPMorgan Chase. According to him, enterprise revenue doubled over the quarter and now makes up the bulk of income.

What This Means in Practice

For developers and product managers, the main takeaway is that a price war between the two leaders is already underway. If your product is tied to a single API, now is the time to recalculate your unit economics and ask your provider for new terms. OpenAI is currently ready to bargain for enterprise customers, while Anthropic has no need to undercut prices yet.

Second: the cost of switching between models has become a real competitive advantage. Those with an abstraction layer over providers and their own quality test suites can quickly move to a cheaper option. Those locked tightly to a single vendor pay full price.

Third concerns the coding market. The Codex vs. Claude Code battle is now the main one for both companies, so tools will get cheaper and improve quickly. Subscriptions and limits will most likely keep changing.

There are caveats, too. The figures come from leaks and retellings by two publications, not from audited financial statements. The tech stock market is currently down: OpenAI raised less than half of its planned $30 billion at a $1.25 trillion valuation and is aiming for an IPO next year, while Anthropic, which is preparing an IPO next week, has noticeably lowered its target valuation. If investors keep pressing on margins, prices and terms for developers could change again, and not necessarily in your favor.