OpenAI’s annualised revenue crossed the $20 billion mark in 2025, rising sharply from about $2 billion in 2023, underscoring the extraordinary pace at which the company has scaled both its business and its underlying infrastructure. The milestone reflects not just growing demand for AI tools, but a deliberate effort by OpenAI to tie financial growth to the tangible value customers derive from using its systems at scale.
In a company blog post, Chief Financial Officer Sarah Friar explained that OpenAI has intentionally aligned revenue expansion with real-world usage rather than abstract growth targets. She noted that revenue growth is closely linked to how much meaningful work customers actually complete using OpenAI’s technology, positioning AI as a productivity layer rather than a standalone product.
A central theme in Friar’s commentary was the role of compute in enabling this growth. OpenAI’s total compute capacity has expanded nearly threefold year over year, reaching approximately 1.9 gigawatts in 2025. “Our ability to serve customers—as measured by revenue—directly tracks available compute,” she wrote, highlighting infrastructure as the primary constraint on faster adoption. She added that if broader access to compute had been available earlier, usage and monetisation could have accelerated even more quickly.
The surge in capacity has supported a dramatic increase in usage. Friar pointed to record daily and weekly active user numbers as ChatGPT has evolved from an experimental research preview into what she described as “infrastructure that helps people create more, decide faster, and operate at a higher level.” This shift reflects ChatGPT’s growing role in everyday workflows across industries.
As adoption expanded from individual users to teams and large enterprises, OpenAI adapted its commercial approach. The company introduced workplace subscriptions, usage-based API pricing, and newer advertising and commerce features designed to align costs with actual usage. “As AI moved into teams and workflows, we created workplace subscriptions and added usage-based pricing so costs scale with real work getting done,” Friar said. She also emphasised that advertising will only be added when it is “clearly labelled and genuinely useful.”
Looking ahead, Friar described compute as “the scarcest resource in AI” and signalled that OpenAI’s next phase will prioritise practical, real-world adoption across sectors such as healthcare, science, and enterprise. She also pointed to future monetisation models, including licensing and outcome-based pricing, as OpenAI continues to evolve from an AI research lab into core digital infrastructure for the global economy.
