OpenAI, the company behind ChatGPT, has achieved extraordinary global reach but at an equally extraordinary cost. Despite being valued at an eye-watering $500 billion, OpenAI is reportedly losing nearly three times more money than it earns, even as it becomes the most dominant name in artificial intelligence.
According to reports from The Information and The Financial Times, OpenAI generated $4.3 billion in revenue during the first half of 2025 but posted a staggering $13.5 billion net loss over the same period. Much of that red ink stems from convertible equity arrangements and massive data center commitments that could reshape the entire AI infrastructure landscape if the company can sustain them.
Of OpenAI’s $13.5 billion in reported losses, over half reportedly comes from the “remeasurement of convertible interest rights” in other words, billions in equity obligations issued to investors. Stripping those out, OpenAI’s operating loss for the first half of 2025 still stands near $8 billion, according to The Financial Times.
The paper also reports that OpenAI is projecting $13 billion in annual recurring revenue (ARR) based on more than $1 billion in monthly revenue. But even that optimistic figure is under scrutiny from analysts who question how sustainable such growth can be, especially with rising infrastructure and compute costs.
Roughly 70 percent of OpenAI’s recurring revenue comes from ChatGPT subscriptions, which range from free access to paid tiers at $20/month (Plus) and $200/month (Pro). However, out of the 800 million ChatGPT users, only 5 percent are paying customers.
That translates to around 40 million subscribers, a figure consistent with estimates from Menlo Ventures. By comparison, the firm notes that across the wider AI landscape, only about 3 percent of the 1.8 billion users of generative AI tools are willing to pay.
Menlo Ventures optimistically called this gap between usage and monetization “a major opportunity.” But according to a recent ZDNET/Aberdeen study, that optimism may be misplaced only 8 percent of respondents said they would pay extra for AI at all.
To sustain its AI empire, OpenAI CEO Sam Altman has committed to purchasing more than 26 gigawatts of datacenter capacity from major hardware partners including AMD, Broadcom, Nvidia, and Oracle by the end of the decade.
The cost? More than $1 trillion, largely financed through partnerships and future commitments rather than existing cash flow.
Backing this expansion are industry giants such as Microsoft and Nvidia, both deeply invested in OpenAI’s growth. Nvidia alone has pledged to pour $100 billion into the company much of it in the form of GPU credits, prompting some analysts to describe the arrangement as a “circular investment scheme” reminiscent of speculative tech bubbles.
In addition to subscriptions, OpenAI is reportedly exploring commissions on ChatGPT e-commerce transactions and potentially advertising integrations, an idea CEO Sam Altman initially dismissed but now appears open to reconsidering.
While “AI shopping assistants” may one day drive new profits, similar models have struggled elsewhere. For instance, AI search rival Perplexity recently paused its ad program to rethink monetization after finding limited success with sponsored content.
Despite its financial headwinds, OpenAI’s influence in the AI market is unmatched. Data from SimilarWeb shows that OpenAI platforms account for 80 percent of all generative AI web traffic, with 190 million of 240 million daily visits attributed to its tools primarily ChatGPT.
But even with this dominance, profitability remains elusive. As one observer noted: “The path to profitability is easier said than done make a product so compelling that people will pay for it. We’re not there yet.”