OpenAI's S‑1 Filing Reveals Massive Revenue Growth and Deep Losses
OpenAI filed its S-1 with the SEC this week, and leaked financial statements have given the public a rare look at the company's books. OpenAI's top line has surged, but its expenses are climbing even faster. The company reported $3.7 billion of revenue in 2024. By the end of 2025, that figure had jumped to $13.07 billion. Monthly sales were already nearing $2 billion as the year closed, indicating the upward trend continued into 2026.
Revenue growth, however, is dwarfed by spending. Research and development alone hit $19.18 billion in 2025 - more than the company's total revenue for the year. Of that R&D spend, $10.59 billion went to Microsoft for model-training work. Adding the cost of revenue, which rose from $2.65 billion in 2024 to $7.5 billion in 2025, and sales and marketing expenses, which climbed from $1.11 billion to $5.73 billion, pushes the day-to-day operating loss from $8.78 billion in 2024 to $20.92 billion in 2025. Expressed as a share of sales, the operating-loss margin improved from 237% to 160%, but it remains far above the break-even point.
The headline net loss for 2025 appears even more alarming at roughly $39 billion. A large part of that figure is a one-time accounting charge tied to OpenAI's shift from a non-profit to a for-profit structure. Independent reviewers estimate that non-recurring charge at about $30 billion. Stripping it out leaves an adjusted loss of close to $8 billion for the year - a number that still shows the business is burning cash but puts the loss in a more understandable context.
Financing activity underscores the scale of the cash burn. In March 2026, OpenAI closed a $122 billion funding round that valued the company at $852 billion. At the same time, the firm reported over 900 million weekly active ChatGPT users, yet only about 50 million of those are paying subscribers - roughly 5.5% of the total base. That low conversion rate raises questions about how much of the user base can be monetized without eroding growth.
Cost-cutting moves have already begun. OpenAI shut down its Sora video-generation model in March 2026, just fifteen months after launch, and told employees to halt "side quests" in order to concentrate on core coding and products for business customers. The decision reflects an effort to curb expenses that are currently outpacing income.
Investors evaluating the IPO should also consider how OpenAI stacks up against its peers. PitchBook's AI business-quality scorecard gives OpenAI a 4.8 out of 10, the lowest in its group. At the $852 billion valuation, that translates to $177.5 billion per quality point - 11.8 times what investors pay for Databricks. By contrast, Anthropic enjoys a higher run-rate annual recurring revenue (estimated at $47 billion versus OpenAI's $25 billion-$33 billion), a 40% enterprise market-share lead (versus OpenAI's 27%), and a clearer path to profitability according to the same analysis. PitchBook's review also notes differences in capital efficiency, compute obligations, and governance risks between the firms.
Looking ahead, the company's own projections suggest a steep hurdle. To justify the $852 billion IPO price, OpenAI would need to generate $95 billion-$105 billion in free cash flow by 2030. First-quarter 2026 results show revenue of $5.7 billion but an adjusted operating margin of negative 122%, meaning the firm spent $2.22 for every dollar earned. On that trajectory, the company would lose $10 billion in 2026 and $30 billion by 2030, instead of generating the cash needed to support the valuation.
A crucial piece of the profitability puzzle is the revenue-share agreement with Microsoft. In April 2026, the two sides renegotiated the deal, capping Microsoft's payments at $38 billion through 2030. That cap is expected to save OpenAI between $70 billion and $97 billion over the life of the agreement. Without it, the cash-flow outlook would be far bleaker, and positive free cash flow would be unlikely. Notably, OpenAI's path to profitability now hinges on a contract with an expiration date, adding a layer of uncertainty for long-term investors.
Other figures from the filing paint a picture of a company that is spending heavily to stay at the frontier of AI. Payments to Microsoft in 2025 totaled $17.2 billion, of which nearly $10.5 billion was earmarked for R&D. The firm has also pledged (though not yet spent) $1 trillion for future data-center buildouts. In its earlier communications, OpenAI had claimed it would hit $20 billion in annualized revenue by 2025. The actual number fell short at $13.07 billion.
For retail investors, the key takeaway is that while the top line is expanding rapidly, the underlying economics are still weighed down by massive research spending, a reliance on a single major partner, and a low share of paying users. The Microsoft revenue-share cap offers a potential path to profitability, but its success depends on the terms holding through 2030 and on OpenAI's ability to convert more of its vast user base into paying customers.