Oracle declines on concerns about fulfilling AI cloud demand

Oracle has inked multibillion-dollar deals to develop data centers to power artificial intelligence work for customers like OpenAI, Meta Platforms Inc. and Elon Musk’s xAI. Last month, the company said the cloud infrastructure business would produce $144 billion in sales by fiscal 2030. On Thursday, Oracle said overall annual revenue would total $225 billion by then.

The main question is how quickly Oracle can supply the data centers needed to capitalize on all this demand, wrote Brad Sills, an analyst at Bank of America. This is due to “supply constraints across land, buildings, energy and GPUs,” he wrote.

The shares fell as much as 8.2% Friday in New York, the steepest intraday decline since Jan. 27. The stock had gained 88% this year through Thursday’s close.

While these AI cloud bookings have boosted the company’s valuation, investors have expressed concern about the profitability of the effort. Oracle tried to address that issue during its analyst day Thursday in Las Vegas.

An infrastructure project for AI that generates $60 billion in total revenue over six years, for example, would have a gross margin of 35%, the company said during a presentation. Gross margin represents the percentage of revenue remaining after the cost of producing goods and services.

The margin profile on this example is “illustrative of even the very largest customers,” co-Chief Executive Officer Clay Magouyrk said while speaking with analysts.

The disclosure “can help quell concerns about lower profitability,” wrote Anurag Rana, an analyst at Bloomberg Intelligence. The Information reported last week that some of Oracle’s AI cloud arrangements recently had a 14% margin. “Given that this business is still in its infancy, it’s highly likely that profit will improve over the next few years,” Rana wrote.