Allianz paper says hidden debt may raise risks for hyperscaler bonds

In short: A new paper from Allianz argues that lenders may be underpricing the risk of lending to some hyperscalers because “off the books” lease commitments can make their debt burden look much larger.

What's going on

Big tech companies that run huge cloud and AI systems, often called hyperscalers, borrow money by selling bonds. Bonds are basically IOUs that pay interest. The key question for investors is whether the interest rate is high enough for the risk that the company might struggle to repay.

Financial Times columnist Toby Nangle highlights a new Allianz paper by chief investment officer Ludovic Subran and co-authors. It compares the bond pricing of firms including Alphabet, Amazon, Meta, Microsoft, Nvidia, Oracle, and SpaceX to other companies with similar credit ratings. Credit ratings are grades for how likely a borrower is to miss payments.

Using standard “on the books” debt figures, Allianz finds that markets often treat several hyperscalers as very safe borrowers, close to top rated corporate issuers. But Allianz says a big part of the risk may be easy to miss. The paper points to off-balance-sheet lease commitments, which are long-term rental promises for things like buildings and equipment (like signing a multi-year lease that still needs paying even if business slows).

Allianz estimates these lease commitments lift the average debt burden by nearly 150%, and can push the model’s implied credit quality down by 1 to 2 notches. In its model, adding these leases pulls Amazon and Microsoft down into mid investment grade, and pushes Meta toward the higher end of “junk” territory. Oracle and SpaceX look weaker in the model.

What to watch

If more investors focus on lease-like obligations, bond investors may demand higher interest rates from some hyperscalers. That could raise borrowing costs for data centers and AI build-outs, even for well known tech names.

Source: Financial Times