The Mark Cuban Podcast

Episode Title: Global AI Infrastructure Race
Release Date: October 2, 2025

Overview

In this episode, Mark Cuban delves into the escalating race to build global AI infrastructure, spotlighting OpenAI's massive Stargate data center project in partnership with Oracle and SoftBank. Cuban unpacks the key players, mind-boggling investments, behind-the-scenes financial mechanisms, and the broader ramifications for the AI, energy, and stock markets. The conversation is data-driven, candid, and brimming with real-time insights on the business maneuvers powering the next era of AI.

Key Discussion Points & Insights

1. OpenAI's Stargate Data Center Expansion

(Starts at 01:30)

OpenAI announces five new Stargate data centers — all in the U.S. — in partnership with Oracle and SoftBank.
- Locations: Shackelford, Texas; New Donna Ana, New Mexico; Midwest (unspecified); Milham County, Texas; and Ohio.
Each site planned at approx. 7 gigawatts capacity (“enough energy to power more than 5 million homes”).
These data centers are core to OpenAI's ambitions to scale AI infrastructure.

“OpenAI has said that they are building five new Stargate data centers with Oracle and SoftBank. This is massive from a company that is already doing so much with the Stargate project.”
— Mark Cuban (01:01)

2. Funding and Strategic Partnerships

(04:00)

Nvidia’s Role:
- Nvidia to invest $100 billion in OpenAI, announced as part of the infrastructure deal.
- Nvidia benefits from filling these centers with their GPUs, fueling their own growth.
- Tangled investments: Nvidia funds OpenAI; OpenAI contracts Oracle; Oracle invests back into Nvidia & OpenAI—a “giant circle.”

“It seems like Nvidia's investment here just is the gift that keeps on giving.”
— Mark Cuban (06:12)

Oracle’s $300 Billion Deal:
- OpenAI signed a $300 billion letter of intent (LOI) with Oracle for future compute resources.
- Not immediate cash—more a roadmap for future spend—boosting Oracle’s value and credibility in the market.
- Oracle stock up over 90% in the past six months, driven by these headline-grabbing deals.

“These $300 billion dollar LOI deals are boosting Oracle stock up a ton... basically the whole stock has doubled in six months.”
— Mark Cuban (08:30)

3. Market Impacts and Corporate Shake-Ups

(09:00)

The mutual reinforcement of investment and promises (“LOIs”) across Nvidia, Oracle, and OpenAI has led to skyrocketing stock prices, especially for Oracle.
Oracle’s executive shift:
- Magic (joined 2014; formerly AWS) and Mike Silke (with Oracle since Primavera acquisition, 2018) named as new co-CEOs.
- Safra Katz transitions from CEO to executive vice chair of the board.

“It feels like new breath is being breathed into Oracle in an insane new way...”
— Mark Cuban (11:02)

Oracle to raise $15 billion in corporate bond sales, benefiting from its higher stock price and prominence.

4. Financial Engineering & Industry Ripples

(13:00)

Using the $300B LOI, Oracle is able to attract bond buyers and scale up infrastructure spend.
- The bond sale could be split into seven tranches, possibly including a rare 40-year bond.
Oracle also engaged in separate talks with Meta for a $20 billion compute deal, showing the rapidly growing scale of compute demand.

“OpenAI is like, all right, here's a $300 billion LOI. Everyone was asking like where are they going to get the money to build the compute? ...But using that LOI, they can now raise these corporate bonds.”
— Mark Cuban (14:22)

5. Energy Use, Concerns, and Policy

(16:40)

The new centers’ 7 GW energy draw could strain local and national energy grids.
Public backlash in parts of the U.S. over rising energy costs tied to massive data center builds.
Some regions provide tax breaks, grid access, and cheap electricity as incentives for building data centers.
There’s discussion at the federal level (mention of the Trump administration) about allowing data centers to construct their own on-site power plants to offset grid stress.

“If OpenAI goes and builds these five new ones that could use the same amount of energy as 5 million homes, pulling that from the grid could be pretty, pretty rough. And so we need more energy production.”
— Mark Cuban (17:51)

Notable Quotes & Memorable Moments

On the scale of AI infrastructure:
“These data centers are going to be insane. This is going to be massive. A lot of energy consumption.”
— Mark Cuban (18:35)
On industry dynamics:
“We kind of have this moment where it feels like Nvidia is giving money to OpenAI to build data centers. OpenAI is giving money to Oracle to also build them data centers. And Oracle is giving money back to Nvidia and OpenAI... it's like a giant circle of everyone is giving each other money or letters of intent.”
— Mark Cuban (07:14)
On market exuberance:
“Oracle, who’s kind of been this like incredible company to watch over the last six months—they’re up 48%, which of course that’s amazing as well. But like even more than Nvidia, we’re seeing Oracle stock absolutely rip.”
— Mark Cuban (08:07)
On skepticism:
“Some people say it’s dubious, these $300 billion LOIs, but the company is exploding. Their stock price has doubled... It feels like new breath is being breathed into Oracle in an insane new way...”
— Mark Cuban (11:00)

Timeline of Major Segments

| Timestamp | Topic / Quote | |------------|--------------| | 01:01 | OpenAI’s Stargate data center announcement | | 04:00 | Fundraising and Nvidia’s $100B investment | | 06:12 | The “infinite loop” of investments and stock impacts | | 08:07 | Oracle’s stock surge and $300B LOI details | | 09:30 | Oracle leadership shakeup and new co-CEOs | | 13:00 | Bond sale and LOI financial leverage | | 14:22 | Meta’s $20B compute deal | | 16:40 | Energy demands and public policy responses | | 17:51 | OpenAI’s impact on national power grids |

Tone & Style

Mark Cuban maintains an enthusiastic, analytic tone, frequently expressing awe at the scale of investment and the complexity of the business dance among OpenAI, Oracle, Nvidia, and others. He blends industry gossip (“the gift that keeps on giving”) with practical concerns (“hopefully this isn’t something that just completely nukes the entire grid of America”), keeping the conversation both lively and grounded.

Summary

This episode presents an insider’s view on how the business of AI infrastructure is fueling unprecedented deals, stock surges, industry transformations, and even national policy debates. OpenAI’s Stargate expansion, backed by Nvidia and Oracle, embodies the new titan race to build the backbone for the next decade of artificial intelligence, for better or worse. Mark Cuban ties it all together with market logic, skepticism about “paper” money (LOIs), and a watchful eye on the real-world consequences in the energy sector.

Perfect for listeners seeking an up-to-date, business-focused perspective on the global arms race to own the future of AI.

Transcript

A (0:01)

OpenAI has said that they are building five new Stargate data centers with Oracle and SoftBank. This is massive from a company that is already doing so much with the Stargate project. Just on Tuesday they announced this. They're going to build all of these. They're going to be in the United States. And I know that they've mentioned they're going to do other Stargate projects other places. These five are just still in the United States. The new data centers are going to have a planned capacity of about 7 gigawatts. This is absolutely massive. That's enough energy to power more than 5 million homes. So today on the podcast, we're going to be breaking down everything that's going into this project, where they're getting the money from, the impact this has had on Oracle, how Nvidia is helping to fund this, and a lot more. But before we get into that, I wanted to mention if you want to build AI tools without knowing how to code, I'd love for you to check out my startup AI box AI. We've just launched a beta of our builder, which is a platform where you can describe a tool you want to build, whether that's a newsletter generator, whether that's some sort of tool that you use internally in your organization. You know how you use AI to automate your processes. You can describe what that tool might look like and it will automatically link together the AI models, put in the prompts, build the front end UI and create that tool for you so you can automate your processes using AI. You also on the platform get access to the top 40 AI tools on our playground. So if you want to chat with ChatGPT or Anthropic and mix between them, use the image generators and audio generators all in one chat thread without having to pay subscriptions to all the different platforms, you get access to that as well on the AI box playground. So go check it out. AI box AI. But let's get into the rest of the episode today. So the new sites for the current project, or that the five new that just got announced are going to all be in interesting places. So there's Shackelford, Texas, New Donna Ana, New Mexico, there's one in the Midwest, and there's two others that are being built with SoftBank in Ohio and Milham County, Texas. So we're seeing a lot of like New Mexico, Texas, Ohio and some of the Midwest kind of all over the place. Not a lot on the east or west coast specifically, which is interesting. And these new projects are part of this massive infrastructure build out that they're doing. And what's interesting is all of this is coming as a lot of people are asking like how are they going to spend all of this money, how are they or how are they going to raise all the money to fund this? Right? Like OpenAI essentially is becoming a data center creator, which is really interesting. And a lot of this comes back to Nvidia. So Nvidia said that they are planning to invest $100 billion into OpenAI. Specifically just on Monday they announced this and this is all part of the deal for this data center build out. So basically it's interesting because up until this point we've seen that Nvidia, they invest money into all AI, all the big AI startups, 11 labs, anthropic like they, they put money into all the big ones because they know they're gonna have to buy their GPUs. And this is interesting because it's moving just from okay, you need our GPUs to train your models to like we know you need to buy our GPUs to fill up an entire data center or multiple data centers. So it seems like Nvidia's investment here just is the gift that keeps on giving. Even if these data centers aren't exclusively used for OpenAI, which I think they would be, but like they're, they're creating an asset, a data center asset that is going to be filled with Nvidia GPUs. So this is a no brainer for Nvidia to invest in. Now the controversy that a lot of people have talked about is the stock implications of a lot of these deals because we have Oracle making a lot of these new deals. Oracle also just, you know, assigned announced that they were signing a $300 billion deal with OpenAI where OpenAI was going to pay them $300 billion for compute. And this of course was controversial because it wasn't like OpenAI wired them $300 billion, they don't have that much money. They gave them a letter of intent that in the next like three years or four years, I think that they're going to want to do this deal. So what's interesting to me is that we kind of have this moment where it feels like Nvidia is giving money to OpenAI to build data centers. OpenAI is giving money to Oracle to also build them data centers. And Oracle is giving money back to Nvidia and OpenAI to as investments like all of these different companies are, it's like a giant circle of everyone is Giving each other money or letters of intent. Right? So not even money, but like, hey, I'm putting this down that I want to buy it. And when this is happening, the stock prices of these companies are going crazy. If you look at Oracle stock price over the last six months, it's up 90%. Over 90% in the last six months. And I mean in Nvidia, who's kind of been this like incredible company to watch over the last six months they're up 48% which of course that's amazing as well. But like even more than Nvidia, we're seeing Oracle stock absolutely rip. Like going up 90% in six months is crazy. Coming from $150 to 100 to 293, basically the whole stock has doubled in six months. And you know, by the way, this wasn't a small company before. Like Oracle is a massive corporation. They've been around for a very long time. And so it's like basically these deals, these 300 billion dollar LOI deals are boosting Oracle stock up a ton. And then Oracle is use this inflated stock price or I don't know if you call it inflated, but this greatly increased stock price to go and invest more money into their GPUs or invest it into these other companies, which is interesting. So Oracle because of this has had a bunch of shakeups. They just announced that two of their presidents are going to basically their, their CEO is moving into a new role and they're, they, they have got two new presidents to co CEO. The company. Obviously this thing is, is growing into an entire beast of its own. I mean their stock doubled in the last six months alone. So Magic joined Oracle in 2014 from AWS and he was a founding member of Oracle's cloud engineering team. So he's gonna be one of the new CO CEOs. And then we also have Mike Silke who's also gonna be a co CEO. And Silke has served as the president of Oracle's industries division since June. He had a bunch of other different roles at the company and he actually came to Oracle through an acquisition of Primavera Systems back in 2018. So Safra Katz, who is currently the CEO and has been since 2014, is going to move into a role as the executive vice chair of Oracle's board of directors. So we're seeing this like trend where the company is just exploding and some people say it's dubious these $300 billion, Lois, but the company is exploding. Their stock price has doubled. They're moving the CE out to new position and putting new CEOs in. It feels like new breath is being breathed into Oracle in an insane new way where they're, they're basically catching up with what was only, you know, Nvidia was basically the only company that had people had said was going to win as selling, you know, picks and shovels in the AI gold rush. And it seems like now we're bringing them other players that are growing rapidly. Oracle, AMD and a bunch of other players. So with that, Oracle is now going to be raising $15 billion in corporate bond sales, which is interesting. And also like $15 billion in corporate bond sales today versus six months ago is massive. They're literally their stock price doubled. So they're able to, they're able to get a way better valuation on this. You know, they've been known as a cloud infrastructure giant for a very long time and they're going to be doing this through corporate bond sales. That's according to reports out of Bloomberg. And this could actually happen this year. This is the second biggest bond sale this year that Oracle will be, will be doing. And this could include up to seven different parts with potential for one of those to be uncommon 40 year bonds. That's according to Bloomberg. So all of this is right after OpenAI signed that massive $300 billion LOI when they got that right. So OpenAI is like, all right, here's a $300 billion LOI. Everyone was asking like where are they going to get the money to build the compute? Because that's just an loi. They have to go and build the compute. But using that loi, they can now raise these corporate bonds. Their stock price is up. Like they have a lot of, all of a sudden everything exploded. So now they have the tools because publicly there's this $300 billion LOI. They now have all the tools they need to go and raise the money and actually build this, which is interesting. And also apparently they are in talks with Meta for about a, for about $20 billion worth of compute as well, which a lot of people are like, oh, it's only $20 billion, like that's nothing. You know, OpenAI just did a 300 like $20 billion of compute. This is a massive deal. This should have really been quite positive for their stock. But with everything else going on being so much crazier. It is, you know, it didn't seem to get grab as many headlines as other things. So it seems like everything in data center land is going absolutely crazy. And I just want to bring you back to something I said at the very beginning of this episode, which is that these new data centers are going to be 7 gigawatts. So that is enough energy to power more than 5 million homes. Like these data centers are going to, are going to be insane. This is going to be massive. A lot of energy consumption. A lot of people in America, I will say, are complaining that the cost of energy is going up dramatically because of data centers. Um, and so it's sort of another point of controversy where if you look at certain jurisdictions are giving these data centers tax breaks or basically incentives to build their data centers. They're giving them cheaper electricity, they're tying them straight into the grid. I believe the current Trump administration has. I don't know if anything has happened here, but they've talked about allowing the new data centers that are being built to also build power plants on site to offset that. Because basically, like, you know, if OpenAI goes and builds these five new ones that could use the same amount of energy as 5 million homes, pulling that from the grid could be pretty, pretty rough. And so we need more energy production. So I'm not sure if, I'm not sure what OpenAI's plan is for that. I sure hope that they're planning on building out their own energy production along with the data centers. So it'll definitely be something we'll have to, to look at. But yeah, hopefully this isn't something that just completely nukes the entire grid of America. I'm sure this, that will happen in, in some small part. But it's incredible just the scale and the price of a lot of these projects. So I'll definitely keep you up to date on all of that. And as always, make sure to go check out AI Box AI to build your own tools with AI and to get access to the top 40 different AI models in one place for 20 bucks a month, including text, image, audio, and basically everything you need for using generative AI at work and in your job. Thanks so much for tuning in and I'll catch you in the next episode.