OpenAI’s Revenue Climbs Past $20B as Compute Expansion Powers Rapid Growth

OpenAI is no longer just a research lab with a clever chatbot. 

It has officially entered its “infrastructure era,” fueled by a massive $20 billion revenue surge and enough computing power to light up a small country.

In a weekend update, OpenAI Chief Financial Officer Sarah Friar revealed that the company’s annualized revenue has skyrocketed past $20 billion in 2025. To put that in perspective, the company was making $2 billion just two years ago.

But this growth isn’t happening in a vacuum. Friar explained in an announcement that compute capacity grew roughly threefold year over year, expanding about 9.5x from 2023 to 2025, rising from 0.2 GW in 2023 to 0.6 GW in 2024 and approximately 1.9 GW in 2025.

Revenue followed a similar trajectory, increasing about threefold year over year and roughly 10x from 2023 to 2025, climbing from $2 billion in ARR in 2023 to $6 billion in 2024 and more than $20 billion in 2025.

From one provider to a diversified infrastructure stack

For a long time, OpenAI was almost entirely dependent on Microsoft for the “brain power” needed to run its models. That is changing fast. 

To keep up with the $17 billion it reportedly spends annually to stay operational, the company has diversified. They are now working with a wide range of partners, including Nvidia, AMD, Oracle, and AWS, to ensure they never run out of juice.

This shift allows OpenAI to be more tactical. They use the most expensive, high-end chips when training a brand-new “frontier” model, but switch to cheaper, more efficient setups for the millions of people who simply ask ChatGPT to help with their homework or a work email.

Friar described the current momentum as a self-sustaining cycle: “Adoption drives revenue, and revenue funds the next wave of compute and innovation. The cycle compounds.”

Why your ChatGPT might soon have ads

Even with $20 billion coming in, running AI is incredibly expensive. To help bridge the gap, OpenAI is making significant changes to its revenue model.

The company recently confirmed it will start testing ads for users of the free version of ChatGPT and the new, low-cost ChatGPT Go plan ($8 per month). The goal is to keep the service accessible to everyone without going broke. However, if you are a “Plus” or “Pro” subscriber, you don’t have to worry — those tiers are staying ad-free for now.

OpenAI is also exploring new ways to charge big businesses, moving toward “outcome-based pricing” where companies might pay based on the actual value the AI creates for them, rather than just a flat monthly fee.

The 2026 vision: From novelty to habit

As we move further into 2026, OpenAI’s goal is to make AI a “predictable” part of life. They want to move away from the wow factor of a talking bot and toward agents.

Friar summed up the company’s trajectory and its role in the future of the economy in the OpenAI report:

“Infrastructure expands what we can deliver. Innovation expands what intelligence can do. Adoption expands who can use it. Revenue funds the next leap. This is how intelligence scales and becomes a foundation for the global economy.”

Also read: ChatGPT Translate is now live as OpenAI tests standalone tools beyond the core chatbot.