10x in 3 Years. OpenAI Achieves $20 Billion Annualized Revenue. But Will It Ultimately Be "Forced" to Sell Ads?

TradingKey - OpenAI CFO Sarah Friar disclosed in a blog post last Sunday that the company's 2025 Annual Recurring Revenue (ARR) exceeded $20 billion, up from the $6 billion achieved in 2024, and a staggering 10 times the $2 billion in annualized revenue recorded in 2023.

Friar believes that investing more in computing resources during this period of unprecedented growth will accelerate OpenAI's path to profitability.

Where does the 20 billion in revenue come from? Computing power is revenue!

Friar stated that revenue growth is fundamentally aligned with the expansion of compute capacity. OpenAI's compute scale has increased from 0.6GW in 2024 to 1.9GW in 2025, while weekly active users (WAU) and daily active users (DAU) continue to reach record highs.

Currently, individual subscriptions serve as OpenAI's core revenue pillar, with its monthly active users and subscription counts experiencing explosive growth in 2025. As of the end of 2025, ChatGPT's weekly active users have exceeded 800 million, while monthly active users surpassed 700 million in July.

Enterprise services were OpenAI's fastest-growing engine in 2025, with over 1 million companies now paying for its enterprise-grade AI products. Paid seats for ChatGPT workplace products—representing the number of employee users—have reached 7 million. This makes ChatGPT the fastest-growing enterprise platform in history.

Friar noted in a blog post that OpenAI's ability to serve customers is entirely linked to available compute capacity. In the AI sector, compute is the scarcest resource. Three years ago, OpenAI relied solely on Microsoft as its compute provider; it has since established partnerships with additional suppliers, ensuring compute stability and securing support on a larger scale.

Can 20 Billion in Annual Revenue Allay Market "Cash Burn" Concerns?

Although annual revenue is growing rapidly, it pales in comparison to the massive expenditures on computing power and R&D.

Deutsche Bank cited a forecast disclosed by OpenAI showing that, OpenAI's cumulative losses before achieving profitability could reach as high as $143 billion. OpenAI forecasts revenue of $345 billion between 2024 and 2029, but expenses for computing power are expected to reach as much as $488 billion. Furthermore, this projected $143 billion loss does not include OpenAI's recently announced $1.4 trillion commitment to data center investment.

The market has already priced in the expectation that OpenAI will incur massive losses before turning a profit; however, the issue is that OpenAI's spending is currently increasing alongside its revenue. This creates a paradox: the more OpenAI sells, the more it loses. Following the publication of this blog post by Friar, Tech blogger Paul Kedrosky remarked: "It is laughable to read OpenAI's CFO boasting about their success in selling dollars for $0.70 in bulk."

This points to another critical issue: OpenAI has yet to establish a healthy business closed-loop.

In terms of gross margins, OpenAI resembles a capital-intensive enterprise more than a SaaS company. Traditional software has almost zero marginal cost, allowing profits to explode exponentially as revenue grows. For OpenAI, however, the costs of computing power and electricity are so high that they cannot be amortized through the scale effects of user growth, making it difficult for the cost of inference models to decrease.

Additionally, the majority of OpenAI's revenue comes from ChatGPT subscriptions, which have an inherent ceiling; individual subscriptions, as a consumer-facing business, face the risk of growth deceleration. An October 2025 Deutsche Bank report showed that consumer spending on ChatGPT in the European market has nearly stalled since May, suggesting that the pool of users willing to pay for subscriptions may have reached saturation and that subscription revenue in certain regions has hit a ceiling.

Regarding subscription rates, as of October 2025, only about 5% of weekly active users are paying. This conversion rate implies that OpenAI is using the fees from a minority to subsidize the "free-riding" of the majority. Under this business model, non-paying users represent not only ineffective traffic but also add to the costs of computing power and electricity.

Currently, other large language models are catching up rapidly. Google's Gemini 3, released at the end of 2025, has shown the potential to surpass ChatGPT in several areas. This narrowing gap will also dampen users' willingness to pay for ChatGPT.

OpenAI "Pivots" to Advertising, but Analysts Predict Potential to Challenge Google

OpenAI once promised to build a purer and smarter way of accessing information than Google. However, considering the high costs of electricity and computing power, as well as the growth limits of the subscription model and current extremely low conversion rates, to address the urgent need for continued R&D spending, OpenAI's best path forward currently may be to sell advertisements and generate traffic revenue from free users.

OpenAI stated last week that it will begin serving advertisements to some U.S. users within ChatGPT, intensifying efforts to generate revenue to cover the high costs of technological development. OpenAI stated that in current testing, ads will appear at the bottom of responses and will be relevant to the user's current conversation.

Considering the scale of ChatGPT's user base, Mark Mahaney, a senior analyst at Evercore ISI, is highly optimistic about ChatGPT's advertising business, forecasting $25 billion in annual advertising revenue and predicting it could pose a threat to Google.

Mahaney noted in the report that by 2025, the combined advertising revenue of Google Search and YouTube could approach $300 billion, while Meta's advertising revenue will be approximately $180 billion, with operating margins for the advertising business as high as 40%. He predicts that OpenAI could achieve billions of dollars in advertising revenue by 2026 and exceed $25 billion by 2030.

Additionally, analysts proposed another advertising strategy, namely that OpenAI launch conversational ads allowing users to research and discuss potential purchase intentions within ChatGPT, which could attract advertisers to reallocate portions of their marketing budgets.