AI Debt: Is the $200 Billion Number Real?

Jim Rickards says AI companies raised $200 billion in debt markets in 2025.

That’s the headline claim in his June 22, 2026 press release promoting the AI Black Paper. It’s also the kind of number that makes you grab your reading glasses and ask: “Where did that come from?”

Fair question. Rickards has a history of seeing systemic risk where others see smooth sailing. He called the 2008 crisis. He’s been warning about market fragility for years. So when he drops a $200 billion debt figure, you want to know if it holds up.

I went looking. Here’s what I found.

The Claim

Rickards’ position is straightforward: AI companies — the hyperscalers, the data center builders, the private credit borrowers — raised over $200 billion through debt markets in 2025. Investment-grade bonds. High-yield. Private credit. All of it.

The implied argument: this is a lot of debt, it’s piling up fast, and it creates systemic risk if credit markets turn.

The Market Data

The actual numbers are striking.

Global technology companies issued $428.3 billion in bonds in 2025 through the first week of December alone, according to Dealogic data reported by Reuters. U.S. firms accounted for $341.8 billion of that.

The five largest hyperscalers — Amazon, Alphabet, Meta, Microsoft, and Oracle — issued $121 billion in U.S. corporate bonds in 2025. Their historical average? $28 billion per year between 2020 and 2024. That’s a 4x increase.

Bank of America expects them to issue $140 billion per year for the next three years, with an upper estimate of $300 billion annually.

Tech debt hit 18% of total investment-grade supply — the highest share on record. That’s per Garrett Baldwin’s Substack analysis, confirmed by Goldman Sachs and Morgan Stanley research. By October 2025, Bloomberg reported that $1.2 trillion of high-grade debt was tied to AI, making it the largest single segment in the U.S. investment-grade market — bigger than the banks.

Meta’s $30 billion October 2025 bond sale was the largest non-M&A high-grade corporate deal in history. Oracle took on $18 billion in September 2025 alone. Then JPMorgan and MUFG spent six months placing a $38 billion debt package for Oracle data center projects in Texas and Wisconsin — one of the largest AI infrastructure financings ever.

This is real. It’s documented. It’s coming from Bloomberg, Reuters, Dealogic, and Bank of America — not from a Substack newsletter with a chart drawn in MS Paint.

The Nuance

The $200 billion figure is defensible. But the definition matters.

If “AI debt” means everything the five hyperscalers borrowed in 2025, you’re at $121 billion in public bonds. Add another $38 billion for the Oracle data center deal. Throw in private credit, project finance, and smaller tech company issuance — and you’re pushing past $200 billion easily.

If you take the broader view — $428 billion in global tech bonds with a significant share tied to AI infrastructure — then $200 billion is actually conservative.

The debate isn’t about whether the number is fabricated. It’s not. Rickards’ team clearly did their homework. The question is: what counts as “AI debt”? All corporate borrowing by tech companies? Only data-center-specific financing? Does private credit count?

Depending on how you draw the circle, the answer ranges from “plausible” to “understated.”

What This Means

The claim passes.

Rickards isn’t making up numbers. The $200 billion figure is in the right ballpark based on available market data from authoritative sources. If anything, the broader trends suggest the number will be larger in 2026.

But here’s the part that matters more than the number itself: the framing.

Rickards compares this to Enron. That’s a separate argument — and we cover it in our standalone Enron piece. The debt is real. Whether it’s a time bomb is a different question.

What I can tell you: credit markets are paying attention. JPMorgan launched a credit default swap basket on AI company debt in 2026 — Alphabet, Amazon, Meta, Microsoft, Oracle — giving institutional investors a structured way to hedge against exactly this risk. Morgan Stanley warned in its “Thoughts on the Market” podcast that “if credit markets lock these companies out, the AI supercycle ends.” HTX Research called AI a “liquidity black hole” for global capital markets.

Wall Street doesn’t build hedging products for fairy tales. When JPMorgan starts packaging CDS baskets on hyperscaler debt, it means the smart money sees the risk.

The Bottom Line

The $200 billion number is real. The data supports it. Rickards isn’t crying wolf on this one.

But numbers don’t tell the whole story. The real question — the one that separates a good promo from a good investment — is whether all that borrowed money will earn a return before the debt comes due.

That’s a question the bond market hasn’t answered yet.