JPM Daily Credit Strategy Update: Quantifying the significant hyperscaler skew on 10s30s spread curves; Curveball and CD Player published

研报英文原文证据摘录

JPM Daily Credit Strategy Update: Quantifying the significant hyperscaler skew on 10s30s spread curves; Curveball and CD Player published

May 2025 at 27bp (+6.5bp YTD), but here the hyperscaler impact is almost nil with the 5s10s ex-

hyperscaler 5s10s curve also +5.1bp YTD. Supply in the intermediate part of the curve has been much more broad-based across

sectors so hyperscalers are simply not as big of a driver of the belly as they are at the long end.

Figure 1: Decomposing 10s30s spread curve between the overall Non- Figure 2: Hyperscaler 10s30s spread curve 2.0x as steep as overall

Fin market and ex-hyperscalers market

10s30s spread curve, bp 4 45 10s30s Hyperscaler spread curve, bp

22 Hyperscaler impact, rhs

Non-Fins

20 Non-Fins ex-Hyperscalers 40

16 2

12 1 25

Jan-26 Feb-26 Mar-26 Apr-26 May-26 Jun-26 Jul-26 Jan-26 Feb-26 Mar-26 Apr-26 May-26 Jun-26 Jul-26

Source: J.P. Morgan. Source: J.P. Morgan.

We ran a multi-variable regression on the 10s30s spread curve level against UST yields and overall spreads YTD (RSq 60%) and

the predicted overall 10s30s spread curve is 2.2bp flatter than the actual. However, the ex-hyperscaler curve is 3.5bp flatter than

the overall curve (the gray ‘hyperscaler impact’ in chart above) so this means curves are closer to fair value, e.g. this same

regression (Rsq 73%) predicts this curve to be 1.3bp flatter than actual. Conversely, this same analysis on the hyperscaler 10s30s

spread curve predicts it to be 7.5bp flatter, though with just a 29% R-squared, which is telling as it conveys the notion of the

extent to which these curves had disassociated themselves from a yield/spread valuation framework and are rather trading almost

entirely on supply/demand technicals at this juncture, in our view.