The latest figures revealed by Sarah Friar, OpenAI‘s chief financial officer, shed new light on the health of the artificial intelligence industry leader. In a blog post on Sunday, Friar reported that the company’s annual revenue in 2025 has surpassed the $20 billion barrier. This is an impressive jump from the $6 billion reported a year earlier and confirms that the market’s appetite for generative AI is not waning. Importantly from an investor perspective, the financial growth goes in close correlation with the expansion of infrastructure. The company’s computing capacity has grown from 0.6 gigawatts (GW) in 2024 to 1.9 GW this year, suggesting that the company is effectively monetising every watt of energy added.
Maintaining a so-called ‘light balance sheet’ (asset-light) remains a key element of OpenAI’s financial strategy. Friar emphasised that the company avoids owning heavy infrastructure directly, relying instead on flexible partnerships and contracts with multiple hardware providers. This approach allows for dynamic scaling without the need to freeze capital in physical assets, which is key at such an aggressive growth rate. Despite record numbers of active users, the cost of developing the technology remains gigantic. In response to this challenge, OpenAI announced last week the introduction of ChatGPT ads for selected users in the US, a clear shift in the search for new revenue streams beyond subscriptions.
However, the company’s ambitions go beyond software alone. Axios reports that Chris Lehane, OpenAI’s head of policy, has signalled that the company is on track to unveil its own device in the second half of 2026. This move would bring the ChatGPT developer into the hardware market, putting it in direct competition with consumer electronics giants.
In the near future, the company’s focus will be on ‘practical adoption’ in the health, learning and enterprise sectors. Friar has announced a move from simple text and voice interfaces to fully autonomous agents. The new phase of development is to be based on systems capable of automating complex workflows, transferring context over time and taking action across business tools. The year 2026 is therefore expected to be the year when AI will cease to be just a chatbot and become an integral, autonomous part of business processes.
