Robin Wigglesworth on Hyperscalers' 1.5 Trillion of Off-Balance Sheet Liabilities, Private Credit, and His Book "A Fabulous Debt" artwork

Robin Wigglesworth on Hyperscalers' 1.5 Trillion of Off-Balance Sheet Liabilities, Private Credit, and His Book "A Fabulous Debt"

Monetary Matters with Jack Farley

August 16, 2026

Robin Wigglesworth — editor of FT Alphaville and author of A Fabulous Debt: The Epic Story of How Bonds Built the Modern World — joins Jack Farley to unpack the hidden debt fueling the AI buildout.
Speakers: Jack Farley, Robin Wigglesworth

Topics: Investing, Business, News, Business News

**Jack Farley** (0:00)
I'm joined today by Robin Wigglesworth, editor of Alphaville, the Financial Times Financial Blog and author of A Fabulous Debt, The Epic Story of How Bonds Built the Modern World. Robin, we want to talk about bonds, fixed income, of course, but we've got to start with what's going on right now.
You've been doing some work on the off-balance sheet, hidden leverage of the hyperscalers, Meta, Google, Microsoft. So we're reading, oh, $100 billion of CapEx, Google's doing that, Microsoft's doing $150 billion. This is so much money, oh my God. But actually, it's looking like it's almost guaranteed to be way, way higher. You've been looking at the numbers and just from the first to the second quarter, the guarantees, the lease obligations and so forth, off-balance sheet again, went from roughly one trillion to roughly 1.5 trillion. What are you looking at?

**Robin Wigglesworth** (0:51)
It's fascinating. It is one of the biggest capital markets events of our lifetimes really.
We've seen massive CapEx booms before, like the railways in the 19th century is like the classic parallel that people draw, transformative technology, very expensive to build.
What's unusual, of course, railways back in the day used to be almost like venture capital ideas. They were very VC-ish. Today, it's like major large money machines that are doing it. Google Alphabet makes lots of money, Meta makes lots of money, Amazon makes lots of money.
For a long time, the money they were pouring into data centers, they could just fund it from their free cash flow. Google Search, Amazon, Facebook itself just prints money, so it's easy. But the scale is just becoming so massive that they've increasingly turned to the debt markets.
As we now see, there are actually some signs of indigestion, like the sheer scale of the bond sales. We're talking multiple hundreds of billions of dollars both last year and already this year. We've already smashed last year's record for the hyperscalers bond sales. And they're getting more creative. And look, creativity and finance can be a good thing. I find a lot of this stuff fascinating, but it can also be quite dangerous, as you know, Jack. And first, it was structuring some of the bonds as leases. So essentially, let's say, take a great example, Meta is building a huge data center in Louisiana called Hyperion. And rather than do some pay squillions to build it, they're only investing 20 percent, they're buying 20 percent of it. But they are guaranteeing that they will lease that data center for 20 years. So and the lease payments are essentially will cover the cost of that company itself, like a JV with Blue Owl, and they'll sell those bonds to other investors. But you know, since off balance sheet, it doesn't come up as a bond or a debt or a loan for Meta. But of course, it's on the hook for paying this lease for 20 years. And this has inspired a lot of the other hyperscalers. So we've seen massive amounts of these kind of lease structures happen. So that's what's gone to $1.5 trillion.
I mean, less than a trillion last year, and nothing de minimis a few years ago. And crucially, 500 billion or so of that, you can see as the leases have started, so you can see them in the financial accounts. They won't appear as debt, but you'll see the payment obligations on the balance sheets. But a trillion dollars of that is for leases that haven't even started yet. And that doesn't appear except as a footnote. So Goldman Sachs, that's where I got the numbers from. They did their god's work in going through all the filings to find their stuff. What I did then was I started looking at the purchase commitments, because these companies have also promised to buy, obviously chips and equipment, cooling, stuff like that. But obviously power as well. These data centers need electricity, and they need quite often guarantees that they will get that power. And that has gone from, again, also roughly $1 trillion earlier this year to $1.5 trillion. And these are quite often payment obligations they can't squirrel out of. So they kind of walk, talk and quack a bit like debt, but they don't actually pair as debt. And I think it's fascinating. Maybe this is, you know, they're betting the house on AI, and I hope this all works out, but I'd feel maybe slightly more comfortable if they structured some of this as more plain vanilla debt and let the debt markets do the talking.

**Jack Farley** (4:40)
If they structured it as plain vanilla debt, we, you know, Goldman and you wouldn't have to do all of this work to figure out. It is really interesting. So it's really a transformation from a pure, complete AAAA investment grade counterparty like Microsoft. And it is using that, but through a much less investment grade data center developer or a Neo Cloud to then they're the ones who are actually spending the money and they report to investors this giant backlog, which Core just reported. And that giant backlog is basically these off balance sheet commitments that the Microsofts and the hyperscalers have made.

53 more minutes of transcript below

Thousands of transcripts fetched by people building searchable podcast archives

Feed this to your agent

Try it now — copy, paste, done:

curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Works with Claude, ChatGPT, Cursor, and any agent that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.

Using your own key:

curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/YOUR_EPISODE_ID