How AI is Reshaping Pension Risk
The Challenge
AI borrowing is reshaping the bond benchmark that sets pension discount rates.
The AI boom is driving a historic mobilization of capital. Tech companies, semiconductor manufacturers, utilities and data center operators are committing hundreds of billions of dollars to the computing and power infrastructure needed to fuel the AI buildout. Much of that borrowing is taking place in the investment-grade corporate bond market.
As AI-related companies take up a larger share of the AA-rated bond universe, they’re gaining greater influence over the yields used to set pension discount rates. Those rates determine the present value of future pension payments. When they move, liabilities — as well as funded status — can move with them, even if a plan’s benefits and investments remain the same.
These dynamics create an unexpected exposure. A shift in the credit quality or borrowing costs of AI-related companies reshapes pension funding, and sponsors have no direct control over any of it.
“AI concentration in AA credits has become an embedded risk,” says Jeffrey Passmore, CFA, Pensions & LDI Solutions Strategist at MetLife Investment Management. “And it’s one that I don’t think is at the forefront of anyone’s mind.”
The Impact
Pension liabilities can shift on credit-market moves no plan sponsor controls.
The scale of the AI boom is unprecedented, but pension plans have felt the effects of credit-market upheaval before.
During the 2012 and 2013 downgrades of major financial institutions, six large banks fell out of the AA-rated bond universe. Their higher-yielding bonds were removed from the calculations used to set pension discount rates. With those yields excluded, discount rates fell and liabilities rose, even though plan benefits, Treasury yields and investment performance hadn’t changed.
For Passmore, the episode is a reminder that pension funding can be affected by shifts in the bond market, even when nothing changes within the plan itself.
A similar dynamic is emerging as AI companies borrow to finance the buildout. Passmore notes that some bond managers are already nearing limits on how much they can hold from individual issuers, pushing spreads higher as the market absorbs the debt. Those wider spreads can lift discount rates and lower pension liabilities. But if issuers are later downgraded out of the AA universe, their bonds drop out of the calculations, potentially reversing the effect.
The risk is that AI companies don’t have to falter for pensions to feel the strain. If they thrive and their credit spreads tighten, discount rates could fall, increasing pension liabilities even as the companies grow stronger.
The scale of the AI buildout — and the borrowing behind it — continues to grow, while the industry itself remains uncharted territory. “We’re really at a frontier with AI,” Passmore says. “Because of that, I think there are greater unknowns associated with AI that could ultimately affect how investors view AI-related bonds.”
The Takeaway
Knowing which AI issuers sit in the discount curve is the first step to managing the risk.
The task for pension sponsors isn’t to predict the next turn in the AI boom, but to understand how AI companies’ borrowing and credit profiles feed into pension liabilities. As AI-related issuers take up more space in the AA bond universe, changes in their spreads or ratings will have an even greater impact on discount rates and funded status.
Passmore recommends starting with a list from the plan’s actuary of the AI-related issuers included in its discount-rate calculations, ideally before year-end measurement. Sponsors concerned about the concentration can also consider yield curves that exclude outlier bonds.
“The more practical thing is just being aware and monitoring the potential exposure,” Passmore says.
The aim is not to predict where the AI wave will break, but to understand the currents it sets in motion, and how they may carry pension funding with them.