OpenAI's third-quarter numbers look impressive on the surface. According to reports from the Wall Street Journal and The Information, revenue grew about 50 percent quarter over quarter to $10 billion, after growth of roughly 17 to 18 percent in the second quarter. A year earlier the company brought in $3.9 billion, so revenue has more than doubled in that span, and the reports say it has roughly tripled within a year. Those are the kinds of gains any company would want to brag about.
The problem is what sits underneath. Adjusted operating losses rose slightly, from $10.1 billion to $10.9 billion, and that figure excludes stock-based compensation, which climbed to $2.6 billion. The adjusted operating margin did improve, moving from negative 150 percent in the second quarter to negative 109 percent, but it was negative 97 percent a year earlier. Growing revenue while losses grow alongside it is not yet a business that pays for itself, and the reports do not suggest a clear path to that point.
The comparison with Anthropic is the sharper lesson. Anthropic overtook OpenAI on revenue in the second quarter and posted a small operating profit at the time. In the third quarter it grew revenue 47 percent to $17 billion, with an operating loss including stock compensation of just $567 million. The reports attribute much of that efficiency to an enterprise focus and to tools such as Claude Code and the Claude Cowork agent, while OpenAI funds broad research into robotics and runs an expensive consumer business. Those choices have costs, and OpenAI is now paying them.
OpenAI's enterprise push is the right instinct, and the reported results suggest it is working. President Greg Brockman said the enterprise business doubled in the third quarter and now accounts for the bulk of revenue, and he pointed to customers such as JPMorgan Chase. Cheaper model prices and a hire of Dali Rajic as head of sales show a company trying to win business it previously neglected. Still, cutting prices to win customers can squeeze margins, and the reports do not show whether that trade will eventually pay off.
My view is simple: OpenAI should stop treating its consumer ambitions and its research bets as a free pass on spending. The 1.2 billion weekly users and 100 million paying subscribers are a strong asset, but scale alone will not fix a widening loss. Investors weighing its planned IPO next year, and a fundraising round at a $1.25 trillion valuation that has so far raised less than half of the planned $30 billion, should demand clearer evidence that the operating margin is heading toward zero. Until then, the revenue headline is only half the story.