Bite: The $750B race to build AI infrastructure

0:00You're listening to a Shersey's podcast. I want to come back to the video. These are companies that, as you say, it's got a huge future ahead of it. But a lot of that is based on not money that's coming from your eye using an application and paying anthropoclord or whatever it is to use it. It's coming from these companies, hyper-scaler companies, Meta and so on, that are putting huge bets, huge capital purchases and so on in there and saying, we'll buy this many chips, we're going to invest this much. I want to try to get a bit of a sense with the build out is going to be sustained because the adoption part doesn't seem to be quite be turning into money at the moment. As you say, the revenues in the video are linked to continued capital expenditures from these hyper-scalers, Amazon, Microsoft, Meta and the willingness of them to continue to spend that money and then actually the ability to fill them to continue to spend it by raising finance and to keep spending this.

1:00To give you some context, these companies might have been spending around about $30 billion a capex per year a couple of years ago. We've got some of them spending $200 billion per year. So Amazon was a spectator. I think it's coming into this year as expected to spend about $130 billion, $940 billion in capex. That's just to tell everyone it's going to spend $200 billion. So you can track written in the video's revenue. So nobody is expected to generate revenues. There was a high $300 billion next year. It might be $350 to $370. That number needs to be directly tied to these hyper-scalers capex. And a big five companies out there, Amazon, Meta, Google, Microsoft, and you can sort of include some unlisted ones like XAI. They spent about $400 billion last year in capex. That's expected to probably be around about $750 billion or even higher. It might even go higher, $700 billion in 2026. That's a massive jump. NVIDIA is going to capture a big share of that.

2:00But then markets probably come to all. They'll probably go, yeah, we know they're going to spend the capex this year. But they need to do it again. We need to keep raising that number and raising it. And these are companies that haven't really had to go into debt very much at all in the last 10 years. You're now seeing them be free cashline negative. Meta's going to be free cashline negative this year. Amazon's going to be free cashline negative. They've got the most profitable business in the world, right? These guys generate hundreds of billions of cash flow, tipping all of that in in capex to build data centers. A big chunk of that goes to NVIDIA. And then they still don't actually have enough cash. So they're going to be raising debt. So they're tapping the debt markets now to help fuel this ride. And everything's going well. And people see AI been used more on the utilities working. I think you can get confidence that this keeps going. But what happens if you get a period where we'll sit on expectations and AI softens, then all of that. And that's when the pressure comes on. And everyone wants to see this as a return. You're going to spend 200 billion dollars a capex and investors want to see a return generated on that capex. And at the moment, there's pretty high uncertainty

3:01on what that return's going to be. Investing involves risk. You might lose the money you start with. We recommend talking to a licensed financial advisor. We also recommend reading product disclosure documents before deciding to invest.