The below commentary has been provided by PGIM
MACRO
The severity of the recent strain across the software sector underscores AI’s potential creative destruction, including the winners and losers that will emerge from the sector’s historic CapEx and infrastructure buildout. In 2026 alone, hyperscaler CapEx is expected to increase by 60% to $650-$700B. In a highly simplified analysis, this CapEx could amount to 2.0-2.5% of GDP (see the accompanying chart), or a 1 percentage point contribution to real GDP, which could be enough to drive above-trend real GDP growth this year, possibly to the tune of 2.5%.
However, events in the software sector also prompt us to consider a scenario where AI CapEx falls short of expectations, potentially moving the economic trajectory to “muddle through” growth of about 1.2%. For example, if one were to assume $485B in CapEx this year for a 14% growth rate and a lower share of GDP, that outcome could amount to a 0.16pp contribution to real GDP growth, or about 80 bps lower than the scenario with a 1pp contribution.
In terms of categories, the speculative tier of AI firms, with potential 2026 CapEx of $100-$120B, could be the most at risk given their reliance on the debt and equity markets and where a financing shortfall is tantamount to cancelled GPU orders.
Considering the massive CapEx amongst the eight major AI players, it is also logical to examine scenarios for return on invested capital. For example, the potential revenue required to achieve a 10% return on two-year forecast investment amounts to nearly $750B annually, assuming a 30% gross margin and a 10% depreciation rate. In approximate terms, the amount of revenue would essentially amount to the addition of another Amazon each year.
DEVELOPED MARKET RATES
As market volatility increased last week, U.S. two-year yields rallied, but remain constrained by the number of Fed cuts priced in. Meanwhile, 10-year yields consolidated in a tight range for several months, moving lower as equity volatility increased last week.
With Kevin Warsh nominated as the next Fed Chair and, if confirmed, likely to consider balance sheet contraction, a significant reduction at this point would likely require bond sales, as opposed to a balance sheet roll off process. Indeed, after the prior QT initiative, much of the balance sheet assets are long maturity bonds that will not roll off over the short term.
Last week, mortgage spreads tracked the broader risk curves, generally widening during most sessions, with a brief tightening at the end of the week. Liquidity remained healthy, with limited concerns about market functioning given the broad volatility.
