Richard Urwin is chairman of the investment committee at Saranac Partners
The AI story is migrating from equity screens to bond desks. As the hyperscalers turn to debt markets to fund their capex build-out, a consensus is hardening that this borrowing now underpins higher-for-longer yields. The story is real, but overstated, and the decomposition matters.
Interest rates around the world have jumped in recent weeks, and plenty of explanations are doing the rounds. The question is which ones really matter – and the split between short rates, mostly set by central banks and long rates, driven by inflation and the sustainability of government borrowing, is where to start.
See also: Allspring: An investor’s guide to the AI infrastructure stack
Look closely at recent trends in bond markets and a consistent picture emerges. Short rates have risen more than long rates, with near-term inflation expectations up largely on energy prices. Indeed, expected short rates have risen faster than expected inflation: markets now price central bank hikes, a sharp reversal from the cuts priced in before tensions flared in the Middle East this spring.
Long-term inflation expectations, by contrast, have barely moved – so much of the rise in yields has come through real rates. Cross-country yield spreads have shifted but stayed within normal ranges, currencies have been calm, and hyperscaler borrowing to fund AI investment has surged.
Not a fiscal credibility story
To a certain extent this has been reported as markets losing faith in governments’ finances, with every small move an assessment of a government’s economic credibility. The evidence doesn’t support this. If worries about government debt were driving yields, we would expect to see three things: higher long-term inflation expectations, widening cross-country spreads and sharper currency moves. We have seen none of these. Higher rates do make life harder for governments that borrow heavily. But government debt is not primarily what has pushed yields up.
Where the AI borrowing actually fits
The more convincing explanation lies with central banks. With inflation picking up in many countries, policy-makers are now more likely to raise rates than cut them – and in the US, the AI boom is now a major force of that shift, in two ways. First, surging investment is boosting demand.
Second, large companies borrowing to pay for the investment boom are competing with the government for investors’ money. There are also signs the AI boom is squeezing other parts of the economy, such as construction, as higher borrowing costs make non-AI projects less profitable. This is a concentrated sector event sitting on top of a more subdued borrowing picture elsewhere, not an economy-wide leverage cycle.
Surge, not regime change
Both pressures on rates may prove temporary. The Middle East outlook is highly uncertain, but energy prices already largely reflect that. And while AI-related borrowing has yet to peak, it is unlikely to last for a multi-year period.
The opportunity in repricing
Many investors are wary of bonds after such a fast rise in yields. That caution cuts against the arithmetic. Higher yields mean higher expected returns from bonds from here – with less risk of capital losses and in the US – investors can currently lock in a return of almost 3% a year after inflation for 10years. With equities looking expensive, that may not be far short of what cautious forecasts expect from equities over the same period and with a much smoother probable ride.
