OpenAI crosses $25 billion in annualized revenue and eyes a 2026 IPO

OpenAI has crossed $25 billion in annualized revenue, and the company is now reportedly taking early steps toward a public listing by late 2026. That number is worth sitting with for a moment. Two years ago, OpenAI was a research lab with a chatbot. Today it is generating more annual revenue than many Fortune 500 companies, and it has not even gone public yet.

The revenue figure comes alongside reports that OpenAI is generating $2 billion in monthly revenue and is approaching 1 billion weekly active users. Those are not projections. They are operational numbers from a company that launched its first consumer product, ChatGPT, in November 2022. The pace from zero to $25 billion in annualized revenue is faster than Google, Facebook, or any other major consumer tech company managed in their early years.

What the IPO timeline actually means

An IPO by late 2026 would be a compressed timeline for a company of this complexity. OpenAI raised $122 billion at an $852 billion post-money valuation in its most recent private funding round. Going public at that valuation would make it one of the largest technology IPOs in history, sitting above Alibaba's 2014 listing of roughly $25 billion raised and comparable in scale only to Saudi Aramco's 2019 debut.

The reason to go public at all, given how much private capital is available, is likely tied to liquidity for early employees and investors. OpenAI has been operating as a capped-profit entity, which complicates how equity is structured and when people can actually cash out. A public market solves that problem in a way that secondary share sales only partially address.

Anthropic at $19 billion: the gap is narrowing

Anthropic is approaching $19 billion in annualized revenue. That is a $6 billion gap behind OpenAI, which sounds large until you consider that Anthropic launched Claude, its primary commercial product, more than a year after ChatGPT. The company has grown its enterprise revenue significantly through API access and through its Claude.ai subscription tiers, and it has locked in major cloud partnerships with Amazon Web Services and Google.

Anthropic's revenue concentration is worth watching. A large share of its income flows through AWS, where Claude is embedded into Amazon Bedrock. That makes Anthropic somewhat dependent on Amazon's sales motion, which is different from OpenAI's more direct consumer and enterprise relationship. Both models have advantages. OpenAI's direct relationship gives it better margin visibility. Anthropic's cloud distribution gives it reach into enterprise accounts that might never have found Claude on their own.

Why revenue growth at this scale is unusual

Software companies typically grow fast. AI model companies are growing faster than normal software benchmarks by a significant margin. Salesforce took about 10 years to reach $10 billion in annual revenue. ServiceNow took around 13 years. OpenAI reached $25 billion annualized within roughly two years of launching a commercial product. The comparison is not entirely fair, because OpenAI benefited from existing cloud infrastructure and a distribution network that earlier companies had to build themselves. But the pace still stands out.

The revenue comes from three main sources: ChatGPT subscriptions at $20 per month for Plus and $200 per month for Pro, API access for developers and enterprises, and larger enterprise contracts. The $200 monthly Pro tier launched in late 2024 and targeted heavy users who needed access to OpenAI's o1 reasoning model without usage caps. Enough people signed up at that price point to materially affect the monthly revenue run rate.

Profitability is still an open question

Revenue at $25 billion does not mean profit. OpenAI's compute costs are substantial. Training a frontier model like GPT-4 costs somewhere in the range of $50 million to $100 million per run, and inference costs scale with usage. With nearly 1 billion weekly active users, the inference bill is significant. OpenAI has been spending heavily on data center capacity, talent, and research, and the company is reportedly still operating at a loss despite its revenue scale.

That tension between revenue and profitability will be one of the central questions any IPO prospectus has to answer. Public market investors are generally willing to accept losses from high-growth technology companies, but they want a credible path to margin improvement. OpenAI will need to show that inference costs are declining as a percentage of revenue, which is plausible given ongoing efficiency improvements in model architecture, or that enterprise contracts carry margins high enough to offset consumer losses.

For now, the $25 billion figure is the headline. If OpenAI files for an IPO in 2026, the S-1 document will be one of the most closely read financial disclosures in tech history.