Everyone knows OpenAI is burning money. The company lost $8 billion in 2025 alone, and internal projections suggest losses will triple to $14 billion in 2026. Cumulative losses through 2029 are expected to hit $115 billion before the company reaches profitability sometime in the 2030s. For context, that is more than the inflation-adjusted cost of the Manhattan Project. Yet somehow, investors keep writing bigger and bigger checks. OpenAI's valuation has climbed from $29 billion in early 2023 to over $730 billion in early 2026, and the company has raised more than $168 billion in total funding. How does a company that has never turned a profit command that kind of price tag? And what happens if the music stops?

The money machine

OpenAI's revenue story is genuinely impressive on the surface. The company generated $13.1 billion in actual revenue in 2025, up from roughly $3.7 billion in 2024. By December 2025, its annualized run rate had crossed $20 billion. ChatGPT now has over 800 million weekly active users, and more than 1 million companies are paying for enterprise AI products. That makes it the fastest-growing enterprise platform in history. But revenue growth alone does not explain the valuation. OpenAI trades at roughly 50 to 60 times its revenue, a multiple that dwarfs even the most aggressive SaaS valuations, which typically land between 5 and 10 times annual recurring revenue. The justification from investors is simple: they believe AI will become the most important technology layer of the next decade, and OpenAI is the company best positioned to capture that value. The bet is not on what OpenAI earns today, but on what it could earn if AI penetrates every industry.

How the fundraising keeps working

OpenAI's fundraising rounds have escalated at an almost absurd pace. In March 2025, SoftBank led a $40 billion round at a $300 billion valuation, the largest private funding round in history at the time. By August 2025, the company raised another $8.3 billion. Then in February 2026, OpenAI closed a staggering $110 billion round with Nvidia, SoftBank, and Amazon as strategic investors. What keeps new money flowing is a combination of FOMO and strategic self-interest. Nvidia invested up to $30 billion in OpenAI, but OpenAI then uses that capital to buy Nvidia's chips. SoftBank poured in tens of billions, partly to secure AI infrastructure for its own portfolio companies. Amazon invested to strengthen its cloud and AI services. These are not purely financial bets. They are strategic plays where investors are also customers and suppliers. The company has also signaled that it expects to spend around $600 billion on compute infrastructure by 2030. That number alone keeps the capital markets engaged, because it implies a scale of operation that only a handful of companies on Earth could sustain.

The circular economy of AI spending

This is where things start to look fragile. One of the most criticized aspects of the AI investment landscape is the circular nature of the money flows. Here is how it works: Nvidia invests billions in OpenAI. OpenAI uses that money to buy Nvidia GPUs. Nvidia reports record revenue, which drives up its stock price, which makes it easier for Nvidia to invest even more. Meanwhile, OpenAI points to its growing compute capacity as evidence of demand, which justifies its next fundraising round. Analysts have compared this dynamic to the vendor financing that fueled the dot-com bubble, where telecom equipment makers lent money to their own customers to buy their products. The scale is different, orders of magnitude larger, but the structure is similar. As one analyst put it, "The last time this was prevalent was during the dot-com bubble." Nvidia currently consumes roughly two-thirds of the world's advanced chip packaging capacity. The AI chip ecosystem is effectively a one-supplier market, and that supplier is also one of the largest investors in its biggest customer.

The moat problem

Google, Apple, and Microsoft all have something OpenAI does not: distribution and diversified revenue. Google's AI advantage comes from its embedded distribution. Gemini shows up in Search, Gmail, Docs, Chrome, and the entire Workspace suite. Users do not need to download a new app or create a new account. Google also benefits from a data flywheel, where billions of daily interactions improve its AI, which makes products better, which drives more usage. Google earns the majority of its revenue from advertising and cloud services, meaning its AI efforts are subsidized by existing profitable businesses. Apple controls the hardware layer. Microsoft owns the enterprise productivity suite and cloud infrastructure. Both can bundle AI into products that hundreds of millions of people already use every day. OpenAI, by contrast, is a standalone AI company. JPMorgan described its competitive moat as "increasingly fragile." OpenAI's own strategy memos acknowledge that platform giants could block ChatGPT or push their own AI assistants without giving users fair alternatives. The company views Apple, Google, and Microsoft as existential threats. To address this, OpenAI has started pursuing vertical integration. It is designing custom AI chips to reduce dependence on Nvidia, acquiring companies across data, design, and collaboration tools, and exploring advertising as a revenue stream. But building a platform from scratch while burning $14 billion a year is a very different challenge than bolting AI onto an existing empire.

The AI bubble question

The broader question is whether the entire AI investment cycle is sustainable, or whether it resembles past technology bubbles. The bull case is straightforward. AI adoption is real. Enterprises are deploying it. Revenue is growing. The technology is improving rapidly, and the potential applications in healthcare, scientific research, finance, and energy are enormous. McKinsey projects that $750 billion in US revenue will funnel through AI-powered search alone by 2028. The bear case is equally compelling. Almost none of the major AI companies are profitable. The infrastructure costs are unprecedented. The circular investment flows between chip makers and AI labs inflate apparent demand. And competition is intensifying. DeepSeek, a Chinese AI lab, has emerged as a serious challenger, offering comparable capabilities at a fraction of the cost. Kai-Fu Lee, CEO of 01.AI, has publicly called OpenAI's business model "not sustainable." The truth probably lies somewhere in between. AI is not a bubble in the sense that the technology is fake or useless. But the valuations and spending levels may be running well ahead of what the actual market can support. The gap between what AI companies spend and what they earn is wider than anything the tech industry has ever seen.

What the future holds

OpenAI's internal projections paint an optimistic picture: $100 billion in revenue by 2029, profitability in the early 2030s, and a position as the dominant AI platform company. The company is expanding into enterprise services, advertising, API monetization, and even outcome-based pricing for specialized applications like drug discovery and financial modeling. But the path from here to there requires everything to go right. Revenue growth needs to continue at an extraordinary pace. Compute costs need to come down as the company builds its own chip supply chain. Competitors need to not catch up. And investors need to keep funding losses that are projected to reach $218 billion between 2026 and 2029. The Economist called 2026 a "make-or-break year" for OpenAI. Al Jazeera reported in March 2026 that the fundraising boom is already showing signs of slowing, with Nvidia suggesting its recent investment "might be the last time." If the capital markets tighten, or if a recession hits, or if a competitor delivers comparable AI at dramatically lower cost, the entire financial structure could come under pressure very quickly. OpenAI is not going bankrupt tomorrow. It has over $100 billion in fresh capital and the strongest brand in AI. But the company is engaged in what might be the most expensive bet in the history of technology, spending trillions on the assumption that artificial intelligence will become as fundamental as electricity. If that bet pays off, the current losses will look like a rounding error. If it does not, the losses will be historic.

References