CoreSite’s AI data centre operates with the curb‑side logic of a shopping mall

During a scheduled walkthrough of CoreSite’s newly commissioned AI‑focused data‑centre, journalists observed a facility whose layout, tenant arrangement and service philosophy were explicated by the company’s chief executive as being deliberately modelled on the dynamics of a shopping mall, a comparison that foregrounds the firm’s reliance on a retail‑style leasing framework rather than a purely technical optimisation of compute capacity.

In the course of the visit, the chief executive articulated that each rack and pod is treated as a retail unit, with prospective clients – identified by their publicly traded symbols – invited to select space much as shoppers choose storefronts, a process that ostensibly simplifies the procurement of AI‑grade power and cooling but simultaneously introduces a layer of administrative choreography that can obscure the underlying constraints of power density, latency and thermal management that are integral to high‑performance machine‑learning workloads.

The tour progressed from the entrance lobby, where digital signage displayed occupancy rates alongside advertising for ancillary services, to the core of the building where rows of servers hum beneath raised floors, illustrating how the mall metaphor extends to common‑area amenities such as on‑site security, cafeteria offerings and flexible lease terms, all of which are presented as value‑added features despite the fact that the primary determinant of AI workload reliability remains the robustness of the underlying infrastructure, a factor that critics argue is being downplayed in favour of a more marketable, consumer‑oriented narrative.

By framing the data centre as a commercial promenade, CoreSite implicitly acknowledges a business model that privileges tenant turnover and revenue per square foot over long‑term engineering rigor, a stance that becomes increasingly problematic as AI developers demand ever‑greater bandwidth, lower latency and uninterrupted power supplies, requirements that are often at odds with a leasing strategy predicated on short‑term occupancy and the commoditisation of space.

Ultimately, the juxtaposition of a high‑tech AI hub with the operational ethos of a shopping centre highlights a systemic tension within the colocation industry: the drive to monetise every square metre through market‑friendly analogies may mask essential infrastructural deficits, suggesting that without a decisive shift toward engineering‑first priorities the mall‑like paradigm risks delivering the illusion of readiness while leaving the most demanding AI applications to grapple with the practical limitations of a facility designed more for foot traffic than for computational throughput.

Published: April 30, 2026