I’ve been investing for over 40 years. I lived through the dot-com bust. I watched the 2008 housing crisis unfold in real time. And right now, I’m seeing something that gives me that same familiar feeling in my gut.
Big Tech is borrowing trillions of dollars to build AI data centers. That part you probably know. What you probably don’t know is where a lot of that borrowed money is coming from.
Your retirement account.
Let me explain.
The Biggest Peacetime Construction Project in History
According to McKinsey, global data center spending could reach $7 trillion by 2030. The five biggest tech companies spent over $330 billion on AI infrastructure in 2025 alone. And they need more. Much more.
One company announced plans to raise $45 to $50 billion this year through a combination of debt and equity. Its total debt now sits around $125 billion, with a debt-to-equity ratio above 400%. Its stock is down more than 25% in 2026.
Last week, one of the world’s largest bond managers entered talks to provide roughly $14 billion in financing for a single data center campus in Michigan. That same township already has another $16 billion financing deal in the works for a separate facility.
These aren’t small numbers. And the money isn’t coming from nowhere.
How This Ends Up in Your 401(k)
Here’s the part that should get your attention. These deals are often structured as bonds sold to large institutional investors. The bond manager involved in that $14 billion deal manages roughly $2 trillion in assets for pension funds, insurance companies, sovereign wealth funds, and individual investors.
Some of the biggest, most widely held bond funds in America are actively seeking this kind of exposure. Their own 2026 investment outlooks say so. If you own a target-date retirement fund or a diversified bond fund in your 401(k), there’s a real chance you’re already financing the AI buildout without knowing it.
The GPU Debt Treadmill
Data centers are long-lived assets. A well-built facility lasts decades. But the chips inside them? The high-performance GPUs that actually do the AI work? Those have a lifecycle of about seven years.
That creates a problem. Companies are borrowing against facilities that will need their most expensive components replaced multiple times over the life of the debt. One analyst calls it the “GPU debt treadmill.” You keep running, keep borrowing, keep replacing. If revenue growth ever slows down or a new chip generation arrives faster than expected, the math stops working.
One cloud computing company just closed an $8.5 billion financing facility backed entirely by GPUs. It received an investment-grade rating. First of its kind. That’s either a breakthrough or a warning sign, depending on how you look at it.
Senators Are Paying Attention
In January, four U.S. senators sent a letter calling on the government to investigate how Big Tech is “turning to complex and opaque debt markets to borrow staggering sums of cash.” They warned that massive debt loads could cause “destabilizing losses” for financial institutions and trigger broader economic damage.
A structured finance litigation attorney who worked cases after the 2008 crash put it bluntly. He said tracking AI data center financing feels like “deja vu.” His exact words: “We’re talking about trillions of dollars, and almost going back to the same cycle where there’s almost no transparency about the financing structures.”
What This Means for You
I want to be clear. I’m not calling this a bubble. I’m not saying AI is going away. The demand is real, the revenue growth at these companies is real, and there’s a strong case that most of this debt will be serviced just fine.
But I’ve been around long enough to know that when borrowing gets this big, this fast, and this opaque, regular investors should at least understand what’s happening. Especially when it’s their retirement savings on the line.
Check your 401(k) holdings. Look at what bond funds you own. Read the fund fact sheets. If you see terms like “private credit,” “infrastructure debt,” or “structured finance,” you might be more exposed to this AI buildout than you realize.
Nobody’s going to tell you about it. So I just did.
— Tom
