These come directly out of building our own rate grid across 56 subtypes. Both correct an assumption we, and most of the market, carried until the research said otherwise.
Cold storage now trades 25 to 50 bps tighter than generic dry warehouse. The tenant base, Lineage, Americold and USCS, is institutional-grade and supply is genuinely constrained, so the market has priced it as the more defensible box, not the riskier one. The cold-over-dry cap rate premium compressed from 30-75 bps down to 0-50 bps per JLL’s Q1 2026 read. If your desk is still quoting cold storage at a specialty markup over warehouse, that quote is stale.
Industrial outdoor storage used to price as a specialty niche. It doesn’t anymore. Institutional capital has entered the space, life companies are quoting 150-250 bps over treasuries, and spreads as low as 165 bps have been reported. IOS now prices in line with mainstream industrial warehouse, which is a genuinely different underwrite than the one most lender lists still reflect.
Data center transitional debt is the counter-case. It stayed wide, SOFR + 400 to 700, because debt funds are taking transitional risk banks won’t underwrite: the tenant credit question and utility access, not the real estate itself. A build-to-suit with a pre-committed hyperscale lease prices entirely differently, SOFR + 250-400 at 60-70% LTC, than a speculative or multi-tenant colo facility at 50-60% LTC and SOFR + 350-550.
"Specialty is not uniformly riskier. Cold storage and IOS institutionalized and tightened. Data center transitional stayed wide. An analogy can’t tell you which, and that’s exactly why an inherited rate should never display like a calibrated one."
Don’t assume a premium by asset name. Ask what actually happened to that subtype’s pricing this cycle, not what it used to cost, and ask the tenant question first on anything data center. Getting this wrong in either direction, assuming cold storage is expensive or assuming data center transitional is cheap, costs real basis points on the term sheet.