Today on the podcast, I want to talk about Oracle and OpenAI, who recently we found, OpenAI has agreed to pay Oracle $30 billion. This is an absolutely insane announcement, basically. But the interesting thing was right after it came out, it basically came up because Oracle released some violence to the SEC where they said, you know, we have a $30 billion contract. And based off of that, Oracle's stock price skyrocketed, making the owner of Oracle the second richest man in the world after Elon Musk.
So all of this is part of a much bigger $500 billion project, which the Wall Street Journal yesterday said was delayed and is off to a bad start. What exactly are they building that cost this much money? Who is the customer for this $30 billion? All of that we're going to be breaking down on the podcast today.
And of course, all of the drama from the Wall Street Journal saying that maybe this is, you know, smoking mirrors off to a bad start. This isn't going to be as good as you basically were told when they made this big kind of a start-game announcement earlier this year. So we're going to dive into all of that. Before we do, I wanted to mention, if you want to try any of the latest models from OpenAI to anthropic, to cohere deep-seat Google, I would love for you to try out my platform, which is AIBox.ai.
It's in beta right now. It essentially allows you to access the top 40 AI models. You can compare all of the models. We have benchmark data.
We have basically everything that you need for text, image, audio, and you can chat with it all in the same thread. So it's super useful. This is something that I hope saves people a lot of money. Basically, you don't need subscriptions to 20 different platforms.
You can have one subscription for $20 a month and get access to all of the top models. You can chat with them all in the same thread, which is so much more useful. You can just switch the model mid conversation. So if you want to check it out, there's a link in the description to AIBox.ai and let me know what you think.
All right. Let's get into this $30 billion deal that's going on with Oracle and OpenAI. So basically, Wall Street Journal reports that, you know, everything is not going to plan with a $500 billion dollar shargate announcement that's in Altman, Oracle, and SoftBank made earlier this year. Simaltman came on and essentially cleared the air.
He went on Twitter and said, quote, we have signed a deal for an additional 4.5 gigawatts of capacity with Oracle as part of Stargate, easy to throw around numbers. But this is a gigantic infrastructure project, some progress photos from Abilene. So this is being built out in Abilene, Texas. And he shares a couple pictures, of course, one of just like the wires of the service.
But again, another of the aerial view of exactly what they're building, which is an absolutely massive compound. This thing is huge. And there's a ton of buildings that are already up. There's a ton more space for way more capacity that is getting built out, right?
This is the way these things work is they kind of roll in phases. So it's not like everything start to finish is going to be ready at the same time, but get certain buildings down and actually get them into production before others. Now, the Wall Street Journal said this was off to a rocky start. They weren't going to be able to get things online when they were hoping for.
I think Sam Altman basically had a response that. But before I get to that, I want to say 4.5 gigawatts of capacity, that is enough, like basically energy to power 4 million different homes. It's like two Hoover dams, gigawatts of capacity. So this is like an absolutely massive project.
That is so much. And they're basically building out all of this capacity to fuel all of the AI projects that opening AI is going to be needed to do in the future. And Oracle made a great partnership and is obviously going to be benefiting a lot of their companies growing a ton based off of this. But also they have the experience, so it makes sense to make it basically a pretty good partner for OpenApp.
So I think it's funny. So back in June, Oracle told pretty much, I mean, SCC finally they said that they signed a new deal that was going to generate about $30 billion a year in revenue. So of course, they're shocked. Absolutely exploded.
One thing that I just want to mention though, like we got $30 billion, Oracle's big company, Yada, Yada. I think it's important to know how big $30 billion is even for Oracle as a company. So collectively, Oracle last year sold about $24.5 billion worth of cloud services for its fiscal year 2025 to all of its customers combined. So all of its customers generated about $24 billion so far in, I guess, fiscal year 2025.
And now they're adding $30 billion from one customer. That is absolutely massive. So what's interesting here is, you know, it's definitely part of the $500 billion data center project, but this is apparently not like this is in addition to that $500 billion. So they're doing a $500 billion project.
That's including stopping this $30 billion is not including it. It's just basically they're paying this money in addition to just to get extra gigawatts online. And I'm assuming it's because the $500 billion thing is going to take so much longer. It's a huge project.
There's a lot of stuff that's going on, but they probably could get something with $30 billion down faster and online quicker. So this is going to be pretty crazy. Definitely not a straightforward project. This is going to use a ton of time, energy and money to build this out, but it's important.
And we're seeing, you know, basically we're seeing all of these AI companies integrate in this way where you see someone like opening eye building out their own data centers, you have Amazon's and AI tools and they're relying heavily on AWS. Google is greatly benefited because they have Google Cloud. And then basically you see companies like Amazon Web Server or sorry, Anthropic that is partnering with Amazon Web Servers AWS and they kind of have to use them in order to get the compute that they need. But it's a tricky situation.
They keep getting billions of dollars invested into their company from AWS and Anthropic has, but like they're kind of stuck with them. Right. And so it's kind of interesting to see a company like opening eye partner with Oracle. Now Oracle doesn't really have it's, you know, some sort of headline AI competing product.
They're just more in kind of the data center space. And so it feels like a better partnership where it's like, Anthropic, they're going to Amazon who also builds their own AI tool. So it's almost like a direct competitor in a way. Now I will say it's not like everyone's using Amazon AI per se is not really a thing, but like built into Amazon's website, they have a roof is AI and they have a bunch of other AI tools that like I think a lot of people think about what they actually use.
And so Amazon is a competitor and I think will be more and more of a competitor in the future. So all this to say, Oracle is spending very heavily in order to build out these data centers. Last year, Oracle spent $21.2 billion just on capex. And that was last fiscal year.
They're expecting to spend about $25 billion this year. So about $50 billion total. And almost all of this was on data centers. So this also doesn't include any land purchases.
This is just like literally building out the data centers. And so I think that that is absolutely a massive sum. So one other thing that's really interesting on this, Sam Altman said that he just last month he was talking and he said that, opening I finally passed $10 billion in annual recurring revenue. So what's interesting about this is, yeah, this is huge because it's up from about $5.5 billion last year.
So they're seeing huge growth, but this $10 billion, this is only a third of the commitment they just made to Oracle just for data centers. And so in addition to all the other data center deals, they have, of course, how much money they have to spend for data, for their researchers, for everything else they have going on. It's absolutely massive. They're, you know, they're three X, their annual revenue.
They just gave away and won the old oracle. So this is a huge amount of money. But I think it really goes to like basically the picture that this is painting to me is that opening I see is absolutely massive potential, but it's all contingent on how much computing you can access to. They know that they're going to be dead if Elon Musk and X AI get access to more compute and basically can build out, you know, have more compute, build out more data centers than them.
They're going to get killed because it kind of comes to this point where the bigger your scale is, the cheaper you can make your AI the more you can compete. Otherwise you're going to get crushed by all the competitors. And we know that these AI models, basically opening I has shown that if you give more compute to an AI model, it will get a better result. And that's not cheap, right?
So they're like, if we for every question you ask, you give it $1,000 worth of compute, then the responses were like world class PhD level insane. So they're amazing. But that is very expensive to point where they were giving it like $10,000 to answer some questions and on the benchmarks, those are the greatest questions that I've ever been answered by AI. They're amazing, right?
But that's $10,000 for questions. It's like insane. So what is the solution, right? We have access to super, super smart AI, but the cost is so astronomical.
It doesn't make any sense. So what's the solution? The solution basically is to spend billions and billions of dollars building out massive data centers, massive compute. You can make these things at scale so you can bring down the costs, you can make them cheaper and faster.
And this essentially is going to be the solution that seems like every company, every major AI company is trying to build towards or work towards. It's a huge, it's a huge expense, a lot of caveats, but this is how they're trying to keep your competitive advantage and stay ahead. So all of this is amazing. If you are, if you learn anything new from this episode and if you want to give back the number one way to do it, we'll be able to leave a rating and review wherever you get your podcasts.
If you're on YouTube, drop a like and subscribe to the channel. It helps other people find the channel or find this podcast. So make sure to leave a ring and review or comment. Really, really appreciate them.
And make sure to go check out aiboxet.ai. If you want to get access to all of the top AI models in one place for one price, you don't have to pay for subscriptions to everything to test out all the top platforms. Thanks so much for tuning in today and I will catch you in the next episode.