Data Centers & Neoclouds: contextualizing returns on colocation: US Communications Infrastructure & Global Digital Assets

Research evidence excerpt

Data Centers & Neoclouds: contextualizing returns on colocation: US Communications Infrastructure & Global Digital Assets

15 July 2026

US Communications Infrastructure & Global Digital Assets

Data Centers & Neoclouds: contextualizing returns on colocation

Today, we are leveraging the same framework as in our “Contextualizing REIT-urns” note, but Madison Rezaei

+1 917 344 8622 broadening the aperture to include emerging AI infrastructure companies and bitcoin miners,

madison.rezaei@bernsteinsg.com i.e. WULF, CIFR, CORZ, CLSK and RIOT. As we noted in our prior piece, data center REITs

often screen poorly on traditional metrics such as ROIC or earnings-based multiples, largely

Gautam Chhugani due to their capital intensity, long development timelines, and accounting treatments. We

+91 226 842 1416

gautam.chhugani@bernsteinsg.com take a stabilized returns approach and compare traditional AI colocation providers with their

more nascent peers in emerging AI infrastructure to arrive at industry-wide metrics.

Nancy Wu

+1 917 344 8545 Within colocation REITs, EQIX’s retail-centric platform (+strong interconnection)

nancy.wu@bernsteinsg.com supports structurally higher stabilized returns. We use two approaches to observe

stabilized returns—ROA (total stabilized NOI divided by gross PP&E) and yield on cost

Mahika Sapra (estimated stabilized net income divided by average net PP&E), and observe a consistent +91 226 842 1408

mahika.sapra@bernsteinsg.com return premium at EQIX. We believe that differences in pricing model and customer mix

are a primary driver of this divergence, as EQIX is more heavily weighted towards retail

Sanskar Chindalia colocation, characterized by smaller deployments, higher unit pricing, and structurally