OpenAI Says Compute and Revenue Both Grew 10X From 2023 to 2025

OpenAI says its revenue grew roughly 10X between 2023 and 2025 — from $2 billion ARR to more than $20 billion — and that the curve maps almost exactly onto its compute capacity, which expanded from 0.2 gigawatts in 2023 to approximately 1.9 gigawatts in 2025. The company published those figures in a strategy essay, "A business that scales with the value of intelligence," that lays out its business model, infrastructure economics, and monetization roadmap through 2026 and beyond.

The numbers are the sharpest picture OpenAI has given of the relationship between its capital spending and its income. Compute grew 3X year over year, or 9.5X from 2023 to 2025: 0.2 GW in 2023, 0.6 GW in 2024, and ~1.9 GW in 2025. Revenue followed the same trajectory, tripling year over year to $6 billion in 2024 and $20 billion-plus in 2025. OpenAI's assessment is blunt: "This is never-before-seen growth at such scale."

The company also makes a claim that will shape the next round of data-center deals: "We firmly believe that more compute in these periods would have led to faster customer adoption and monetization." In other words, OpenAI is arguing it was supply-constrained, not demand-constrained — a signal that it intends to keep committing capital to capacity ahead of demand.

Why the compute-revenue link matters

The essay frames compute as the binding constraint on the entire AI economy. "Compute is the scarcest resource in AI," OpenAI writes. Three years ago the company relied on a single compute provider — a reference to its foundational Microsoft partnership. Today it says it works "with providers across a diversified ecosystem," which it says delivers resilience and "compute certainty."

That diversification has an economic logic. OpenAI now treats compute "from a fixed constraint into an actively managed portfolio": frontier training runs on premium hardware when capability matters most, while high-volume serving workloads run on lower-cost infrastructure when efficiency matters more. The result, the company says, is useful intelligence delivered "at costs measured in cents per million tokens" — the price point that makes AI viable for everyday workflows rather than elite use cases.

The stakes extend past one company. Access to gigawatt-scale compute increasingly determines which AI labs can ship frontier models and which cannot, and OpenAI's claim that revenue tracks compute so tightly is an argument for why the capital arms race is self-funding — at least for the market leader.

What happened to ChatGPT after the research preview?

The essay opens with a retrospective on ChatGPT, which launched as a research preview "to understand what would happen if we put frontier intelligence directly in people's hands." The answer, OpenAI says, was "broad adoption and deep usage on a scale that no one predicted."

The usage patterns the company describes moved from personal life into work:

• Students used ChatGPT to work through homework late at night.
• Parents used it to plan trips and manage budgets.
• Writers used it to break through blank pages.
• People used it to prepare for doctor visits and to think clearly "when they were tired, stressed, or unsure."

Then, per OpenAI, users "brought that leverage to work." Engineers reasoned through code faster; marketers shaped campaigns; finance teams modeled scenarios; managers prepared for hard conversations. The company's summary: "What began as a tool for curiosity became infrastructure that helps people create more, decide faster, and operate at a higher level."

Both weekly active users and daily active users continue to hit all-time highs, according to the company, driven by what it calls a flywheel: compute powers research, research improves models, models drive product adoption, adoption drives revenue, and revenue funds the next wave of compute. "The cycle compounds," OpenAI writes.

How OpenAI plans to monetize decisions, not just questions

The most forward-looking part of the essay concerns commerce and advertising. OpenAI states that people now come to ChatGPT "not just to ask questions, but to decide what to do next. What to buy. Where to go. Which option to choose." Helping users move from exploration to action, the company argues, creates value for users and for the partners who serve them.

Advertising follows the same logic. "When people are close to a decision, relevant options have real value, as long as they are clearly labeled and genuinely useful," OpenAI writes. The company applies one standard across every monetization path: "Monetization should feel native to the experience. If it does not add value, it does not belong."

The current revenue architecture is already multi-tiered:

• Consumer subscriptions.
• Workplace and team subscriptions with usage-based pricing.
• APIs where spend grows in proportion to outcomes delivered.
• A free, ad- and commerce-supported tier that drives broad adoption.

The founding principle, according to the company, has stayed constant since ChatGPT became a daily work tool: "Our business model should scale with the value intelligence delivers."

What comes after subscriptions?

OpenAI says the next economic models will extend beyond what it sells today. As intelligence moves into scientific research, drug discovery, energy systems, and financial modeling, the company expects new deal structures to emerge: licensing, IP-based agreements, and outcome-based pricing that shares in the value created. Its historical analogy is direct — "That is how the internet evolved. Intelligence will follow the same path."

On the product side, the next phase is agents and workflow automation: systems that "run continuously, carry context over time, and take action across tools." For individuals, that means AI that manages projects, coordinates plans, and executes tasks. For organizations, OpenAI describes it as "an operating layer for knowledge work." The platform today spans text, images, voice, code, and APIs.

How OpenAI manages the capacity-demand mismatch

The essay is candid that growth "does not move in a perfectly smooth line." Capacity sometimes leads usage; usage sometimes leads capacity. OpenAI says it manages that mismatch three ways:

• Keeping the balance sheet light.
• Partnering rather than owning infrastructure.
• Structuring contracts with flexibility across providers and hardware types.

Capital is committed in tranches against real demand signals, which the company says lets it "lean forward when growth is there without locking in more of the future than the market has earned." Securing world-class compute, it notes, requires commitments made years in advance — a discipline requirement for any lab betting on the compute-revenue flywheel.

What is the 2026 priority?

OpenAI names its 2026 focus explicitly: "practical adoption." The priority is closing the gap between what AI makes possible and how people, companies, and countries actually use it day to day. The company points to health, science, and enterprise as the areas of largest and most immediate opportunity, "where better intelligence translates directly into better outcomes."

The essay closes with the flywheel restated as strategy: infrastructure expands what OpenAI can deliver, innovation expands what intelligence can do, adoption expands who can use it, and revenue funds the next leap. "This is how intelligence scales and becomes a foundation for the global economy."

If the 2023-2025 pattern holds — revenue tracking compute at roughly 10X over three years — the practical question for 2026 is whether OpenAI can keep the flywheel spinning as its capacity commitments grow into the multi-gigawatt range, and whether advertising and commerce can mature fast enough to fund it.