Why OpenAI Lost Over $38 Billion Despite ChatGPT Raking in Cash

Imagine this: You open the most popular bakery in your neighborhood. As soon as you open your doors, customers flock in, and in just six months, you make as much money as you did all of last year—a massive success. People envy you, saying, “You’re sitting on a gold mine now!” But when you, the owner, check your bank account, instead of profits, your debt has snowballed into tens of millions of dollars. It turns out that for every loaf of bread you bake and sell, you’re paying two or three times the bread’s price for the special oven’s electricity and premium flour. It’s a bizarre structure where the more customers you have, the more debt you accumulate.

This surreal story is precisely the reality currently faced by OpenAI, the most watched company in the world today. With hundreds of millions of people worldwide asking ChatGPT questions, delegating tasks, and writing texts every day, OpenAI’s revenue is skyrocketing. However, hidden behind this dazzling innovation is an astronomical “bill.” Why exactly does operating artificial intelligence cost such a staggering amount of money? Simply put, the “food expenses” required to feed this giant brain called AI far exceed our expectations.

Why It Matters

We take it for granted, like the air we breathe, that when we open a smartphone app and talk to ChatGPT, a smart answer returns in a second. We assume we can access infinite knowledge for free or with a small subscription fee, just like searching on Google. But behind this seamless “magic,” unimaginably massive machinery and astronomical capital are burning.

If OpenAI cannot eventually resolve this massive cost structure, the bill could ultimately land on everyday users like us. One day, ChatGPT’s basic subscription fees might skyrocket, or the era of freely using AI in our daily lives might shrink. Numerous convenient services built by startups borrowing ChatGPT’s smart brain—its API (Application Programming Interface)—could also be forced to shut down due to cost issues.

More importantly, this is not just one company’s problem. OpenAI’s balance sheet is like a mirror reflecting the direction of the entire modern tech industry. It clearly shows the extreme, cutthroat competition Big Tech companies are engaging in to dominate the most advanced AI ecosystem today, and the unprecedented capital being poured into building this “invisible infrastructure.” Because their fierce game of survival could fundamentally alter our digital daily lives, we must pay close attention to their bank balances.

The Explainer

First, let’s look at the shocking report card. Over the course of 2025, OpenAI recorded a staggering $13.07 billion in revenue. This is an enormous income that overwhelms a typical large corporation’s annual earnings. However, the total costs and expenditures incurred during the same period reached $34 billion, resulting in a phenomenal net loss attributable to the company of $38.53 billion Source Title. A massive sum—equivalent to a significant portion of a country’s annual national budget—evaporated in just one year.

Comparing this with past data reveals the severity of the situation even more clearly. Following the trend from 2024, OpenAI was pouring in a whopping $2.25 just to earn $1 Source Title. The core reasons for such a deformed revenue structure are ‘Computing Costs’ and continuous research and development (R&D) expenditures Source Title.

Why are AI maintenance costs so expensive? Here is an analogy.

For traditional services like Netflix or YouTube, once a movie is uploaded to a server, it costs the company almost nothing extra whether 10,000 or 10 million people watch it. They simply need to copy and transmit an already created video. Google Search, too, is like a “librarian quickly finding books in an already existing vast library,” making its computing costs relatively controllable.

However, generative AI like ChatGPT is completely different. They go through a process akin to “tens of thousands of genius scholars holding an impromptu meeting every time a question is asked to write a brand-new, customized essay just for you.” Every time a user poses a question, massive computer servers—especially expensive AI Graphic Processing Units (GPUs)—must run fiercely and continuously to generate new letter units, known as tokens. It’s like demanding a Hollywood studio to “shoot and show me a brand-new movie just for me in real-time whenever I want.” It’s a structure where as the number of users grows and demands become more complex, the massive electricity inhaled by the computers and the wear and tear of equipment—in other words, the “expenses”—explode exponentially.

To illustrate how overwhelming these costs are, OpenAI CEO Sam Altman recently made a painful confession. He revealed that even the company’s most expensive plan, the $200/month ChatGPT Pro subscription service, is currently losing money the more it runs Source Title. The operating costs of cutting-edge AI today exceed our imagination, to the point where even the most loyal VIP customers are not covering the raw costs generated when using the service.

Where We Stand

OpenAI is by no means failing to make money. On the contrary, the sheer speed at which it is raking in cash is the envy of every startup around the world. In the first half of 2025 alone, they generated approximately $4.3 billion in revenue, an astonishing achievement that exceeded their total revenue for all of 2024 by 16% in just six months Source Title. About 70% of this massive revenue is stably generated from ChatGPT subscription fees paid monthly by individuals and businesses, while the remaining 30% is created from API usage fees paid by other service developers integrating OpenAI’s technology Source Title.

Behind this dazzling $4.3 billion in revenue, however, lies a dismal $13.5 billion loss incurred during the same period, attached like a hefty price tag due to the massive computing costs explained earlier Source Title.

The shockwaves of this colossal deficit do not stop at OpenAI; they are violently shaking the entire AI ecosystem. Even Microsoft, OpenAI’s most solid partner and major shareholder, is taking the full brunt of this massive deficit. According to Microsoft’s financial reports, the net loss related solely to its stake in OpenAI reached at least $11.5 billion in the third quarter of 2025 (ending in September) Source Title.

Microsoft isn’t just estimating market value; the two companies are bound together as such a solid community of fate that Microsoft must absorb and report OpenAI’s actual profits and losses directly on its own books Source Title. The blow from OpenAI’s deficit is even quietly hidden within a $4.7 billion “other expenses” item Source Title. The cost to defend the AI throne right now is cruelly heavy—enough to make the accounting ledgers of Microsoft, one of the wealthiest Big Tech companies in the world, stagger.

What’s Next

So when will OpenAI be able to escape this endless cycle of debt and become a truly profitable company? Unfortunately, there are absolutely no signs of that happening anytime soon.

According to recently leaked internal financial documents, OpenAI forecasts that its operating losses will balloon to three-quarters of its total revenue even by 2028 due to explosively increasing computing infrastructure costs Source Title. They have even drastically raised their estimate for cumulative cash expenditures—the “cash burn”—that the company must incinerate to run and train its AI models through 2029 to a staggering $115 billion, much higher than initial projections Source Title.

Behind this colossal incineration of funds, which seems irrational to anyone, lies a very clear and cold calculation: the “First-Mover Advantage” for a winner-takes-all outcome Source Title.

It’s the same logic behind the history of car companies enduring astronomical deficits for their first few decades to pave invisible road networks and build gas stations across the country, or Amazon enduring years of near-bankruptcy to build giant fulfillment centers. Once the giant brain known as ChatGPT takes deep root in people’s daily lives and the workflow systems of global enterprises, subsequent competitors will never be able to tear down its stronghold. Instead of pursuing short-term “profitability” on paper, OpenAI is boldly abandoning it to place an extreme bet on completely monopolizing the “electricity and water of the AI era” that the world will have to rely on for decades to come Source Title.

This resolve was also evident in a recent event. During the opening keynote at the 2025 Developer Conference (DevDay), CEO Sam Altman strongly hinted that their technological vision would continue to push boundaries, stating, “We must challenge the very way we build and create things” Source Title. This means they have absolutely no intention of yielding to the fear of deficits and slowing down the pace of technological advancement.

Will OpenAI’s seemingly reckless multi-billion-dollar dash be recorded in human history as the greatest success story of infrastructure investment, or will it be remembered as the worst tech bubble ever? We are currently passing through the very center of the most fascinating and perilous technological evolution.

AI’s Take

The Perspective of MindTickleBytes AI Reporter: Today’s artificial intelligence industry bears a frightening resemblance to the dot-com bubble and gold rush of the early internet era in the 1990s. OpenAI’s astronomical deficit, burning massive amounts of money to monopolize future infrastructure, is not simply a “business slump.” It is the most expensive “ticket” to seize hegemony over the next-generation tech ecosystem and demonstrates the harsh rules of a winner-takes-all game. As AI gets smarter in the future, costs will rise in tandem. It is now up to us to watch and see who survives this game to the end to dominate the world’s digital brain—and how much we will ultimately have to pay in return.

References

1. Exclusive:OpenAILossesIncreasedNearly8Xin2025,With…
2. OpenAIlost $5 billion in 2024 (and itslossesareincreasing)

3. [Microsoft’sOpenAIlosseshidden in $4.7 billion expense

   LinkedIn](https://www.linkedin.com/posts/ward-computers_microsofts-openai-losses-hidden-as-part-activity-7394763832486748160-x-WT)

4. openaifinanciallosses: ChatGPT ownerOpenAIis losing money like…
5. OpenAIDevDay2025: Opening Keynote with Sam Altman - YouTube
6. Microsoft earnings suggest $11.5B+OpenAIquarterlyloss
7. OpenAI is hemorrhaging billions: Microsoft filing reveals …
8. OpenAI’s H1 2025: $4.3B in income, $13.5B in loss - LinkedIn
9. OpenAI’s Billion-Dollar Paradox: Soaring Revenue, Mounting …
10. Breaking down OpenAI’s $5B net loss - LinkedIn
11. OpenAI says it plans to report stunning annual losses through …
12. OpenAI hits $4.3B in revenue in first half of 2025, a 16% …
13. OpenAI expects business to burn $115 billion through 2029 …
14. OpenAI posts US$7.8b operating loss despite US$4.3b revenue …