# OpenAI's Revenue Run Rate Tops $40 Billion Just Months After Doubling > Source: > Published: 2026-08-14 04:50:43+00:00 *OpenAI is already producing revenue at a scale most startups never reach, but the clean $40 billion run-rate claim does not hold up cleanly in public reporting. The better story is sharper: OpenAI is growing fast, Anthropic is closing in, and the losses are now too large to wave away.* OpenAI's revenue story is no longer a theory about what generative AI might become. Reuters reported in March, citing The Information, that the company had topped $25 billion in annualized revenue by the end of February 2026, up 17% from $21.4 billion at the end of 2025. That is real scale. It is not small. Careful with that number. Run rate is a snapshot. It takes recent revenue and stretches it across a year, which means a strong month can make a company look bigger than its full-year accounts will later show. Still, you don't get to $25 billion annualized on hype alone. OpenAI has ChatGPT subscriptions, enterprise sales, API usage and coding tools all pulling in the same direction. That is why the IPO question has become harder, not easier. Bloomberg's OpenAI news page listed a June 26 report that the company was weighing a 2027 IPO after Anthropic's expected public debut, and FutureSearch's July update put its median first trading day around mid-July 2027. Public investors can forgive heavy losses when revenue is compounding like this. They are less forgiving when the math behind the growth is foggy. ## Anthropic has made this a race Anthropic is the reason OpenAI's revenue number now gets read with a stopwatch. The company said on May 28 that its run-rate revenue had crossed $47 billion earlier that month, alongside a $65 billion Series H round at a $965 billion post-money valuation. That changes the race. Claude Code is the wedge. Anthropic said the funding would help expand compute and scale products such as Claude Code and Cowork, and recent reporting from the Financial Times and Wall Street Journal has framed the company as the more aggressive near-term IPO candidate. If Anthropic lists first, OpenAI will not be judged in isolation. It will be judged against the rival that turned enterprise coding demand into the cleaner public-markets story. There is a catch here that you should not ignore. Semafor reported in April that OpenAI and Anthropic count some cloud-partner revenue differently, which can make direct comparisons messy by billions of dollars. A dollar sold through a partner does not always show up the same way at each company. That is not a footnote. It is the difference between a clean race and a race where the lanes have different widths. ## The bill comes due This is the tension. OpenAI is pulling in more money than almost any young technology company in memory, and it is still expected to lose staggering sums. The Information reported that OpenAI's losses could rise as high as $14 billion in 2026, excluding stock compensation, and separately reported that the company expected its business to burn $115 billion through 2029. FutureSearch is even tougher, forecasting a 2026 GAAP loss near $33 billion once stock-based compensation and other accounting costs are included. That matters for you because the AI trade has been sold as a simple growth story for too long. It is not simple. OpenAI, Anthropic, Microsoft, Amazon, Oracle, Nvidia, you name it, are all tied into a compute buildout that demands huge upfront spending before anyone knows how durable the revenue will be. Greg Brockman said in May, according to Bloomberg, that OpenAI expected to spend $50 billion on computing power in 2026. That's the hard part. OpenAI is also changing the people responsible for turning demand into revenue. Axios reported this week that Denise Dresser is leaving as chief revenue officer and that Dali Rajic, formerly president and chief operating officer at Wiz, will replace her. For a company moving toward an IPO, that is not just an org chart change. It tells you the commercial machine is still being rebuilt while the valuation clock is already running. Frankly, the cleanest read is this: OpenAI has proved the demand is real, but it has not proved the business model is settled. A $25 billion-plus annualized revenue base is a serious company. A path that includes tens of billions in annual losses and a possible $115 billion burn through 2029 is a serious risk. The S-1 will settle what leaked run rates cannot. Public investors will want audited revenue, gross margin, compute commitments, Microsoft economics, customer concentration and a plain answer on how much it costs to serve all those ChatGPT users. Until then, OpenAI's growth deserves attention, but not blind belief. **Also read:** [SMIC Raises Chip Prices as AI Demand Overwhelms China's Top Foundry](https://startupfortune.com/smic-raises-chip-prices-as-ai-demand-overwhelms-chinas-top-foundry/) • [Google launches Gemini 3.7 Flash and halves its price as Gemini 3.5 Pro waits](https://startupfortune.com/google-launches-gemini-37-flash-and-halves-its-price-as-gemini-35-pro-waits/) • [Nvidia Bets Its Balance Sheet That $500 Billion in AI Chips Won't Age](https://startupfortune.com/nvidia-bets-its-balance-sheet-that-500-billion-in-ai-chips-wont-age/)