OpenAI sets out scalable model as compute and revenue surge
Artificial intelligence group says pricing follows demand and outcomes, with capacity and annual recurring revenue rising sharply between 2023 and 2025 as it prepares to focus on practical adoption.
OpenAI has outlined a business model it says scales with the value of intelligence, publishing figures that show rapid growth in both compute capacity and revenue over the past three years.
In a statement accompanying the disclosure, the company said monetisation has been shaped by how customers use its products. Consumer subscriptions are priced around capability and reliability, while workplace customers pay through subscriptions and usage-based fees. For developers, OpenAI operates a platform business in which spending through application programming interfaces is designed to align with outcomes.
The group added that it is also exploring commerce and advertising, but only “for moments close to decision-making when options are clearly labelled and useful”, signalling a cautious approach to commercial expansion.
The figures show a sharp rise in infrastructure and sales. Compute capacity grew 3X year on year and 9.5X from 2023 to 2025, increasing from 0.2 gigawatts in 2023 to 0.6 gigawatts in 2024 and about 1.9 gigawatts in 2025. Revenue followed a similar trajectory, rising 3X year on year and 10X over the same period, from $2bn in annual recurring revenue in 2023 to $6bn in 2024 and more than $20bn in 2025.
“Our job is to close the distance between where intelligence is advancing and how individuals, companies, and countries actually adopt and use it,” wrote Sarah Friar, chief financial officer of OpenAI, in the statement.
To support that growth, the company said it has diversified its compute providers to secure sufficient capacity and manage costs. Frontier models are trained on premium hardware, while high-volume inference workloads are increasingly served on lower-cost infrastructure. OpenAI said this approach reduces latency and improves throughput, while helping to deliver intelligence at costs measured in cents per million tokens.
Related reading
- OpenAI hits back at Musk filing, disputes “cherry-picked” notes and sets out $130 billion structure claim
- OpenAI to test ads on free and Go versions of ChatGPT in the US
- Google releases TranslateGemma open models supporting 55 languages
The company also highlighted the breadth of its product platform, which spans text, images, voice and code, as well as APIs for developers. Agents and workflow automation are described as the next phase of product development, extending the use of its models beyond single prompts towards more complex, multi-step tasks.
Looking ahead, OpenAI said its priority for 2026 is practical adoption. The focus will be on narrowing the gap between what artificial intelligence makes possible and how it is used day to day, with particular attention on sectors such as health, science and enterprise. The emphasis, the company said, will be less on novelty and more on embedding the technology into routine workflows at scale.
The Recap
- OpenAI says its business scales with intelligence value.
- Compute grew from 0.2 GW to roughly 1.9 GW.
- The company prioritises practical adoption in 2026 across sectors.
