A couple of months ago, I wrote an article called “The Layer War,” where I postulated that big tech and the AI Frontier Labs were splitting into two camps: the Builders and the Layerers. The Builders were the ones who were spending hundreds of billions of dollars constructing actual physical infrastructure of AI: the data centers and the servers, and, in some cases, even the electrical generation needed to power the servers in those data centers.
The Layerers were the ones who were betting that AI becomes a cheap commodity, and the money will settle into the software layering on top of whatever intelligence you want to use. The thing is, right now it sort of looks like the Builders might be right, and they’re building to win. They are willing to dump as much money is needed to build out the physical infrastructure for their models. And they’re no longer paying in cash—they’re borrowing.
The scale of the amount of money that they’re borrowing is hard to conceptualize. Morgan Stanley said it expects the six largest hyperscalers to spend $785 billion on capital projects this year and close to $1 trillion next year, with more than $1.2 trillion in data center leases already committed.
The Federal Reserve Bank of Dallas expects $3 trillion to $5 trillion of buildout over the next 3 to 5 years, and of that, it thinks about $300 billion of it is going to be AI-related investment-grade bonds this year alone. Investment-grade AI-related bonds are competition for investment against the Fed’s own Treasury notes.
In essence, the builders’ AI data center buildouts are so incomprehensibly expensive that they require incredible amounts of borrowing, some of which will be funded through sales of their own bonds and others of which are completely agnostic about the current interest rate.
What does that mean? It means that the frontier labs building the data centers don’t care what the interest rate is. They can’t afford to care about the interest rate. They are price takers. They will take whatever price is offered to them.
Can you take any price offered? Of course not.
Imagine if this were true of people. I live in northeast Indiana—a red corner of a very red state. This is the kind of place where a $250,000 home is still a reality. The kind of place with three bedrooms and a garage, and a furnace, where you could happily raise a family, if you didn’t mind the local politics.
This is the kind of place where people pay attention to the interest rate because it affects everything from credit cards, personal loans, car loans, boat loans, and, of course, mortgages.
Last week, the 10-year Treasury note closed the week near 5.28%. That’s just a few basis points below the 24-year high it set on Wednesday. The 30-year note sits at about 5.61%. Oil spent the week at or above $100 a barrel. And to top everything off all of that rolled up to lending rates across the board. Freddie Mac reported that the average 30-year mortgage had jumped a quarter point in a single week to 7.28%.
If you take out a mortgage on that $250,000 home, you probably put down about 10%, and so your 30-year loan is for $225,000. In the last several months, the interest rate for that loan rose from 6.34% to 7.28%, which is $140 more a month for your payment, or more than $50,000 over the life of the load. For many people, that’s a decision not to buy the house.
Here’s a thing that sounds disconnected, but it isn’t. The week before, CNBC ran a story about what rising yields mean for the AI Frontier Labs. A strategist named Mark Malak of Seabert Financial said something that I haven’t been able to stop thinking about. According to him, the AI builders are “basically price insensitive to that raise, which means they’re price takers.”
This is where that phrase “price takers” comes in. When it comes to financial markets, a “price taker” isn’t the kind of thing you want to be. It means that you don’t have any other choice, so you take the price that’s offered, even when somebody else walks away. It’s a strange thing to call the richest companies on Earth, and it’s also a thread that runs through all those numbers above.
Which is the point. Walking away is what higher interest rates are for. It’s what they’re supposed to make people do.
How the brakes are supposed to work
For better or for worse, when the Fed raises interest rates because of inflation, what it’s trying to do is put the brakes on the economy.
Cliff’s Note:
Lawrence has a way of taking a story that sounds like it belongs on the business page and showing you why it actually matters to your life. This is one of those pieces.
If you want to read the rest, please consider becoming a paid subscriber. Your support is what lets us keep publishing sharp, independent work like this.
—Cliff
