Transcription
Oracle may be the most disliked company in the software industry, and that's despite the fact that more than 95% of Fortune 500 companies license some Oracle software. But almost none of them describe that experience fondly. There's even a joke that goes something like, "What's the difference between working with Oracle and being held hostage?" Well, if you're a hostage, somebody eventually negotiates a release. The fact is that once you start working with Oracle, it's very hard to get out. And that's probably the reason why Oracle has been described as a law firm masquerading as a software company.
And look, Oracle does sell some very good products. I'm not disputing that. I've used Oracle products throughout my corporate career. But for some reason, which I don't understand, Oracle has gone all in on AI data centers, and now it's got itself into a very precarious situation. So in this video, I want to explain why I think Oracle is cooked and could end up being the first domino to fall when the AI data center bubble collapses.
So Oracle is right in the middle of the multi-billion dollar data center build out. Take a look at this chart showing the circular financing arrangements, and notice that Oracle is heavily exposed to OpenAI, which in turn is exposed to Nvidia. Oracle went from a company that used to generate significant free cash flow to a now a company that carries more than $150 billion in debt. But for most of its history, Oracle's debt has sat comfortably in investment grade. But as of a few weeks ago, that was downgraded to one notch above junk. That's because the board has made an oversized bet on cloud AI, and the odds on that bet get longer every month.
So what actually happens if Oracle can't fund its obligations from the cash that it generates, and it finds itself in financial difficulty? Oracle used to make its money selling licenses, and licenses are beautiful. You write the software once and you sell it a thousand times. The marginal cost is almost zero. It's probably a 70% gross margin business and it generates cash whether the customer hates you or not. But Oracle has somehow decided that it wants to be in the landlord business. It borrows money, buys GPUs, builds air-conditioned warehouses, and rents them out. But the landlord business is a very different kind of business to the license business. So they've used their high-margin cash generation machine to fund a low-margin leasing operational primarily servicing one client, OpenAI. But Oracle didn't just take on OpenAI as a customer, they got into bed with OpenAI and took on the balance sheet risk that OpenAI didn't want to carry. So now the futures of OpenAI and Oracle are intimately connected in a way that doesn't make a lot of sense if you're an Oracle shareholder.
So imagine that you're looking at the financials of an AI company such as OpenAI that is publicly committed to spending hundreds of billion dollars on computing infrastructure to service its future anticipated demand. What do you think its balance sheet would look like? Well, you'd expect either enormous piles of cash collected from investors or an enormous debt to fund that capital expenditure. But OpenAI does have investor cash and according to the draft IPO documents reviewed by The Information, it was holding about 73 billion dollars at the end of March. And that does sound like a lot of money. It just isn't sufficient not against what's already been committed. Because sitting off the balance sheet are hundreds of billions of dollars in purchase commitments for chips, energy, and data centers. So OpenAI does have some cash, but what it doesn't have is debt. And as of March 2026, it has almost none. Now, none of that is illegal. All of the required disclosures have been made. But let's put that aside. Is it still not unreasonable to wonder where all the money that OpenAI needs is going to be coming from? The answer, it seems, is Oracle, which is something I still can't get my head around.
How does a company this old, this prestigious, this arrogant end up in such a mess where their debt is just one notch above junk? What drove it to make such an oversized bet on cloud AI? Well, here's what I think might have happened. For a start, Oracle lost the first cloud war. They were They were clearly beaten by Amazon, Microsoft, and Google. So, that left Oracle in an uncomfortable place. Yes, still profitable. Yes, still enormous. But, a company where people were starting to put alongside the tech dinosaurs like IBM. But, then AI fever arrived, and that was just irresistible. So, instead of trying to mint will win over millions of small cloud customers, Oracle saw an opportunity to supply computing power to a handful of big customers with near unlimited budgets. So, for the first time in in 15 years, being behind in cloud didn't matter anymore because everybody was starting from scratch. So, Oracle looked at OpenAI AI, the biggest buyer of computing power on the planet, and then decided, "We need to win this one, whatever it costs." That's what I think happened. A combination of greed and fear of missing out.
Now, perhaps that strategy would have made sense if they just funded what they could from their own cash flow. But, they didn't. They took on enormous amounts of debt to get into a business with ridiculously low margins. And it has to build these data centers with real cash first before these AI customers can arrive. That is, of course, if they arrive at all. And to be honest, I can't see how they will. OpenAI has committed to spending hundreds of billions of dollars. But, OpenAI hasn't borrowed the money to do that. Oracle has, and that's the problem. The obligation belongs to OpenAI, but the debt belongs to Oracle. And it's the Oracle shareholders who are carrying that because Oracle has borrowed against 30 years of license revenue to build air-conditioned GPU warehouses for a customer that's lost money every single year of its existence.
And then of course there's the timing problem, and that's the part that's most worrying. Oracle's debt is long, but the GPUs inside those buildings are short. They depreciate really quickly, and nobody, not Oracle, not OpenAI, not Nvidia, can actually be certain that the AI that we're using today will be the AI that anyone actually wants in 5 years. And to be honest, that's a hell of a bet to be making with other people's money. And if it doesn't turn into cash real soon, Oracle Oracle could be the first domino to fall.
So in the span of 18 months, Oracle had turned itself from a cash-generating software gorilla into a leveraged property developer with a single tenant. Now, I'm not disputing that this was legal. I question whether it was rational, and whether Oracle has become the first victim to AI fever, the the global pandemic that's gripping the corporate world. Perhaps 2 years ago the board agreed to make a bold strategic bet because they saw the future early. But perhaps also 2 years ago the board was so scared of becoming irrelevant that it bet the entire farm on the the next thing that was shiny and new. And from here I can't tell you which one it is, and I'm not sure that Oracle shareholders can either. But personally, I think Oracle's cooked. Because they don't just need OpenAI to survive, they need OpenAI to thrive.
And if Oracle does face bankruptcy or some kind of crunch, I know that there will be a great deal of schadenfreude watching this company squirm. And that's because in corporate there's been 30 years of license audits and very aggressive behaviors, and that's going to make many CFOs and CTOs feel like they've got to see a bit of corporate justice. But we need to be very careful what we wish for, because if Oracle is the first dominator fall, it won't be the last. And that's because of this circular finance mess that somehow we've allowed to happen right under our noses. This isn't the first bubble, and it won't be the last. But, I'm just really surprised that it got this far before people started ringing the bell.
I'm Dr. Earl Brandt, the founder of Kira. If you enjoy my content, please like and subscribe. You can learn how to be involved with what we're building if you jump onto the website and sign up for the newsletter. Connect with me on LinkedIn and leave your thoughts in the comments below. Thanks again for watching, and I'll see you in the next one.