OpenAI has squeezed better margins out of its paid products this year, as it races to maintain its pole position in artificial intelligence, according to a report in The Information.

The publication reported that the company improved its “compute margin,” an internal figure measuring the share of revenue after the costs of running models for paying users of its corporate and consumer products. As of October, OpenAI’s compute margins reached 70%, up from 52% at the end of 2024 and double the rate in January 2024, the publication said, citing a person familiar with the figures.

An OpenAI spokesperson said the company didn’t release the figures and declined to comment further.

Read More: OpenAI Executives Struggle to Combat AI Spending Concerns

The ChatGPT creator set off the modern AI boom, but it has yet to show a profit, one of the main indicators for investors concerned about a bubble in the industry. Last valued at $500 billion in October, OpenAI has been searching for ways to make money to cover its high computing costs and audacious infrastructure plans.

At the same time, the company is facing intense pressure over its spending and renewed competition. After the Gemini model from Alphabet Inc.’s Google performed better on benchmarks, OpenAI Chief Executive Officer Sam Altman called a “code red” to redirect internal resources to improve ChatGPT, and delayed progress on plans for an advertising service.

Most people use ChatGPT’s free version. However, the company is pushing its business version and paid software features for industries like financial services and education, where it competes with Google and rival Anthropic.

The Information reported that OpenAI has better compute margins than Anthropic for paid accounts, but that Anthropic has better efficiency on server spending overall.

OpenAI is also in early talks to raise at least $10 billion from Amazon.com Inc. and use its chips, in a deal that could value Altman’s company at north of $500 billion.