Big Tech might be betting your retirement savings on AI

Last week, two numbers landed regarding a single AI data center campus in Saline Township, Michigan — numbers that could turn out to be a much larger story.

The project is being built by developer Related Digital to power applications for OpenAI, with Oracle $ORCL as its anchor tenant. The price tag for it all has leaped to $16 billion, up from $10 billion estimated just last fall. Also in the mix: PIMCO, one of the world's largest bond fund managers, which is in talks to provide $14 billion of that $16 billion as debt.

But the ballooning cost isn’t the most noteworthy aspect of the deal. Nor is the "twisty" history of the financing, which, according to Bloomberg reporting, has involved months of "stop-and-start" negotiations, amid increased scrutiny of Oracle’s credit worthiness.

What’s truly noteworthy is the light the deal shines on AI data-center debt and how it can find its way into retail investors’ portfolios — specifically, their retirement savings — without their realizing it. This can also happen via bond funds, which most everyday people regard as the “safe” end of the market, a world away from risky tech bets or direct AI plays like, say, Nvidia $NVDA.

However they’re perceived, bond funds represent a huge market and a significant chunk of millions of Americans’ retirement saving. Per the Investment Company Institute, Americans held some $8 trillion in 401k accounts as of 2023, roughly $420 billion of which was held in bond funds, which make up a little over 5% of 401k assets in general, while the percentage skews higher in older workers’ portfolios. Crucially, however, the $420 billion figure doesn’t take into account target-date funds, which make up over 40% of total 401k holdings and have substantial bond holdings that increase as the target date approaches.

What the Michigan data-center deal highlights is a little-discussed aspect of the larger AI buildout itself: that some of this data-center debt can find its way into incredibly widely held bond funds. In other words, ordinary Americans are likely already financing it, in some part, and through the most conservative corners of their retirement savings, without their ever knowing it.

The AI debt in your 401k

PIMCO’s presence in the deal makes for a good example. PIMCO, which is the opposite of a niche player in the bond world, manages some $2 trillion in assets for central banks, sovereign wealth funds, pension funds, insurance companies, and individual investors — ranking it among the world’s largest bond-fund managers. Its funds appear on major 401k platforms across the country, while the PIMCO Income Fund alone manages approximately $225 billion in assets and is among the more widely held bond funds in American retirement accounts.

The Michigan deal would be PIMCO’s second major data center financing in less than a year. Last year, the firm helped anchor an $18 billion debt package for Meta $META's Hyperion data center in Louisiana — and booked $2 billion in paper gains when prices on that debt rose after closing. The Michigan deal could follow a similar playbook, with PIMCO purchasing the bonds, which would be structured as 144A securities — more on such bonds in a moment — and possibly syndicating them to other institutional buyers.

Nor has the firm been shy about this corner of the debt market in public statements. PIMCO's own 2026 investment outlook makes clear the firm is actively seeking exactly this kind of exposure. While expressing caution about private credit broadly, the firm carved out a specific category it finds attractive: “project finance, or lending secured by data centers that are being built with leases in place to investment grade tenants.” The Michigan campus, with Oracle as tenant, the lease in place, and the deal structured as project finance, fits that bill.

The channel through which such debt reaches retail investors is complicated and structural, but not accidental. The Bloomberg Global Aggregate Index — the benchmark for the PIMCO Income Fund — explicitly includes USD investment grade 144A securities as a component. 144A securities are privately placed debt instruments that cannot be purchased directly by retail investors, but can be held by the mutual funds that retail investors hold in their 401ks. By design, any fund benchmarked against the Bloomberg Global Aggregate will very likely hold them. PIMCO's own prospectus confirms this.

In this light, it becomes reasonable to assume that some bonds being used to finance AI data centers are also being held in funds that cater to retail investors. Whether the specific securities generated by the Michigan deal ever appear in a given retail fund, the deal illustrates the channel through which AI infrastructure debt could reach ordinary people — through the part of their portfolio they likely chose precisely because they thought it was safe and low-risk.

About that risk: What has to go right for this debt to be repaid?

Another wrinkle: Oracle is the anchor tenant for the Michigan data center, but Oracle isn't directly borrowing this money. The debt is held by Related Digital, the developer, through a special purpose vehicle secured against the data center asset itself. Oracle leases the facility once built, and those lease payments are what service the debt.

The chain runs like this: OpenAI pays Oracle for compute. Oracle makes lease payments to Related Digital. Related Digital uses those payments to service the bonds. The bondholders get paid — at which point, were the 144As held in a bond fund, the funds’ holders would receive a small, proportionate interest payment.

But this means there are two links that have to hold, not one. OpenAI has to keep paying Oracle, and Oracle has to keep paying Related Digital. If OpenAI slows its compute consumption, renegotiates, or hits financial trouble, Oracle's revenue shrinks, but Oracle still owes the lease. If Oracle stops paying, Related Digital defaults on the bonds.

The SPV structure is specifically designed to keep this debt off Oracle's balance sheet. Which raises an obvious question: Why would that be necessary?

Oracle's current debt-to-equity ratio appears to be approximately 400%, perhaps higher — and that's only the debt actually sitting on its books. The company, founded by Larry Ellison, spent decades as one of enterprise tech's most reliably profitable businesses, selling database software and related services to corporations that had little practical alternative. It was never a sexy, consumer-tech brand, never a household name, and yet it was an extraordinarily sticky business; once a company built its operations on Oracle's database, switching was expensive enough that most never switched.

Still, in more recent years, as that legacy business has come under pressure, Oracle has moved aggressively to reinvent itself as AI infrastructure — and piled into building the data centers that AI needs to run. That pivot requires debt on a scale the old Oracle never carried. As the obligations have grown and the financing structures have grown more complex (across the industry, not just at Oracle), some of that debt has migrated off the balance sheet entirely, making Oracle’s true total obligations genuinely difficult to gauge from the outside.

Which brings the central questions here into focus: Will the revenues materialize to service all of it? And, down the line, will the bondholders get paid?

To all appearances, the financing only works if several aggressive assumptions hold simultaneously.

OpenAI needs this much compute long term — and has or will have the revenue to pay for it, from a company that has never turned a profit. Oracle can service tens of billions in debt while its cash flows are, arguably at least, already strained. And the underlying hardware doesn't become obsolete before the debt is paid down, which is a relevant risk given that AI chips turn over roughly every two to three years, even as the bonds carrying this debt have maturities measured over much longer periods.

None of these assumptions are necessarily unreasonable. But all of them have to be right, at the same time, for the math to work — just like many other 144A bonds. The kind that could be finding their way into ordinary Americans’ retirement savings.

And whether this one deal becomes a success story or a cautionary tale, it’s very difficult to think individual investors allocating their savings to bond funds actually want to be holding this kind of debt. Which makes the details revealing and eyebrow-raising, just not for the reasons the players involved might hope.

Put simply: If the most conservative part of your portfolio turns out to hold potentially risky AI debt, then it's arguably not the most conservative part of your portfolio.