Analysis: OpenAI is a loss-making machine, with estimates that it has no road to profitability by 2030 — and will need a further $207 billion in funding even if it gets there

by

Don’t call it a bubble, friends!

It feels like AI is everywhere online right now – from silly memes on Instagram to automated bots on X, and even governments using it to help write policies. AI is the latest big tech trend, and it doesn’t look like that will change anytime soon.

Microsoft has invested a lot in OpenAI to power its AI initiatives, most notably Microsoft Copilot. However, Copilot hasn’t yet proven very useful for practical tasks, often needing people to double-check its work and fix mistakes. Similar issues plague Google Gemini, Grok, and other AI platforms – they’re currently better at creating amusing content than significantly improving productivity. Despite these limitations, people are still talking about the potential of these technologies.

Many people believe that tools like ChatGPT and Google Gemini will eventually become incredibly helpful for getting real work done. They’re already pretty good at summarizing information on popular topics, but they still sometimes make things up or just tell you what you want to hear. This is especially problematic when accuracy and responsibility are important.

Even with current challenges, many companies are excited about the potential of using inexpensive virtual humans instead of real people who require salaries, benefits, and rest. Tech giants are rushing to make this a reality, and recent reports from the Financial Times and HSBC (as highlighted by Fortune) indicate that the situation is becoming somewhat…unrealistic.

New information reveals that both companies and OpenAI itself are using borrowed money more and more to cover their obligations to each other. OpenAI, for example, has promised to spend a massive $1.4 trillion on computing power to support its future growth, even though it’s only brought in about $20 billion in revenue this year. That means its commitments are 70 times larger than its income – a huge difference!

OpenAI is looking for ways to increase revenue, and one idea is to include ads directly within ChatGPT. A major focus is on using AI to automate jobs currently done by people, especially in areas like hotels and customer support. It could be significantly cheaper for large companies to use ChatGPT instead of human employees, even if it means paying a substantial amount for the service – as long as customers are happy with the AI’s performance. However, Gartner has noted that some companies who initially tried replacing workers with AI are now reconsidering that approach. But that’s a separate point for now.

HSBC estimates that even with $200 billion in revenue by 2030, OpenAI would still need over $207 billion in funding just to cover its costs. As OpenAI grows, its expenses are increasing dramatically. Running advanced models like Sora 2 and GPT-5 is incredibly expensive – costing millions of dollars each day in computing power. OpenAI currently offers these models at cost to encourage widespread use, build customer loyalty, and ultimately change how people behave. This strategy is similar to Spotify, which operated at a loss for over a decade before fundamentally changing the music industry and how music is distributed. However, OpenAI is aiming for a much larger impact – and potentially one with far more significant consequences.

Okay, as someone who follows tech, I don’t think Spotify going under would have been a big deal for the world’s finances. But this feels different. If it turns out the money being poured into AI and these large language models just isn’t working out – if these companies can’t actually turn a profit – it could be *really* bad. There’s so much investment right now, and if they can’t pay back their debts, it could cause a huge ripple effect and really shake up the markets, almost like what happened with the dot-com bust or the 2008 financial crisis.

Microsoft and others are increasingly turning to debt to fund its AI craze

AI might not actually become profitable. Issues like limited computing power, a lack of diverse data leading to poor quality results, or the fact that it’s currently more of a trendy tool than a practical solution, could all be contributing factors. This is why Microsoft is prioritizing efficiency and lower energy use in its own AI models, recognizing that computing resources and energy are major limitations. However, Microsoft is still largely relying on OpenAI for its long-term AI strategy, which is essentially a significant gamble.

It’s becoming clear that the financial foundation supporting OpenAI’s massive computing needs is surprisingly delicate. According to the Financial Times, companies partnering with OpenAI are borrowing heavily – a total of $96 billion in 2025 – to fulfill OpenAI’s $1.4 trillion in commitments for computing power. Firms like Softbank, Oracle, and CoreWeave are taking on this debt, and OpenAI is obligated to cover these costs even if the expected demand doesn’t materialize.

Over the next eight years, OpenAI has significant financial obligations that could lead to a cash shortage if they can’t secure enough funding. HSBC predicts OpenAI could reach 3 billion weekly users by 2030, with 300 million paying for subscriptions. However, even at that level, Microsoft and other partners might need to invest hundreds of billions of dollars more to keep OpenAI financially stable. OpenAI recently revised its agreement with Microsoft to help find additional income and computing resources.

Current analyses probably don’t fully account for new challenges, such as the rising cost of DRAM caused by the surge in AI demand. Instead of becoming cheaper, computing power is likely to become more expensive due to limited production capacity of both wafers and silicon. Microsoft CEO Satya Nadella recently stated that the company is unable to use all of its available computing resources because it can’t secure enough electricity to power them. This raises the question of whether companies like Microsoft will need to start generating their own electricity. Furthermore, energy costs overall are unpredictable, influenced by global events and climate change.

OpenAI and similar AI firms are trying to get governments to view large language models as vital to national security, hoping for significant financial support and protection. At the same time, a risky combination of debt and investment could lead to billions of dollars in losses if companies in the AI space fail, potentially causing major problems for industry leaders like NVIDIA and Microsoft.

The success of these new technologies hinges on people actually using them. That’s why companies like Meta are pushing their AI chatbots into apps like WhatsApp and Instagram, and why even simple programs like Notepad and Paint are getting AI features. Google’s Gemini wants to manage your email, too – they’re all trying to get us hooked. These efforts are based on the idea that people will become reliant on the technology, and whether that’s beneficial for society isn’t really part of the plan.

We’re being pressured to use this system because if we don’t, it could cause a massive financial collapse – billions in loans could go unpaid. As always, banks would likely be rescued with taxpayer money, leading to higher prices, fewer jobs, increased interest rates, and potentially more inflation. Of course, those in power will likely profit before any of these problems occur.

Large language models (LLMs) are likely here to stay, but current limitations are becoming unsustainable. The biggest challenge isn’t the technology itself, but the massive resources – electricity and water – needed to run these models at a large scale, which ultimately impacts taxpayers. A key question is whether people will be willing to pay for these services. It’s concerning that OpenAI’s success might rely on companies using the technology to significantly reduce their workforce. If widespread job losses occur, how will people afford to pay for OpenAI’s services? And if the current system of information creation by humans breaks down, where will OpenAI get the data needed to improve its models? This business model appears to have some potentially damaging consequences, at least initially.

For this to be successful, costs need to decrease quickly. The next big advancements likely won’t come solely from improving large language models; instead, we’ll probably see breakthroughs in energy and server technology. However, a deeper look into those technologies is a topic for another time.

I’m really curious to see if companies like OpenAI can become profitable before the current economic issues cause problems. The next five years seem crucial, and I’m watching closely to see how things unfold.

Read More

2025-11-29 12:41