Here's a classic startup story with a twist: you build a world-changing technology, you partner with a tech giant to fund it and power it, and then you have to tell potential investors that your relationship with that giant is one of your biggest problems. That's where OpenAI finds itself as it eyes a public offering.
According to a document from its latest funding round, OpenAI has warned investors that its "heavy dependence" on Microsoft Corp. (MSFT) for "a substantial portion of our financing and compute" could pose a significant business risk. It's the kind of disclosure you make when you're getting your house in order for Wall Street—acknowledging that your most important relationship is also a potential single point of failure.
The Scale of the Reliance
Think about what that means. A "substantial portion" of financing and compute. For a company that just secured $110 billion in funding and is reportedly seeking another $10 billion, that's a lot of eggs in one basket. The company's future performance, it says, depends on its ability to diversify its partnerships beyond Microsoft. It's a delicate dance: you need the partner to grow, but you can't be seen as wholly dependent on them when you're asking the public markets for money.
This isn't OpenAI's only headache, of course. The document also highlighted rising compute costs, potential disruptions in the supply of critical chips, and ongoing litigation as other key risks. It's a preview of the kind of disclosures you'd expect in an S-1 filing, offering a candid look at the challenges behind the AI hype.
Revenue Soars, But So Do Costs
And there is plenty of hype to manage. OpenAI's growth has been nothing short of spectacular. The company has surpassed $25 billion in annualized revenue as of last month, up from $21.4 billion a year earlier. That's the kind of trajectory that gets investors excited.
But here's the other side of the coin: building and running advanced AI models is incredibly expensive. The company is now projecting compute spending of about $600 billion through 2030. Let that number sink in. It's a reminder that in the AI gold rush, the companies selling the shovels—like the chipmakers—might be in an even better position than the miners.











