A Three-Zone Spatial Framework for Modern Data Centers
White Space, Gray Space, and Black Space. Architecture, Standards, Capital, and Operational Doctrine for the AI Era
Abstract
The three-zone spatial framework organizes the modern data center into white space (IT equipment), gray space (mechanical and electrical support), and black space (perimeter and utility-interface infrastructure). The framework supports operators making architecture, capital, governance, and operating decisions that span the three zones rather than treating each zone as an independent specification. The publication is calibrated for the senior infrastructure executive, the mission-critical engineering organization leader, and the capital-committee member who must commit to a spatial architecture before the campus’s structural envelope is finalized.
This publication develops the three-zone framework in seven movements: the spatial framework itself, standards and codes by zone, engineering and operations by zone, capital and schedule considerations, AI factory implications, reference architectures, and adoption guidance. The substantive engineering and governance content is supported by figures and tables across the operating dimensions the framework integrates.
The target reader is the senior data center architect, the engineering organization lead, the capital committee member, and the operations leader committing the campus to a spatial framework that supports AI-factory density. The publication is also relevant to OEM equipment suppliers, regulatory engagement teams, and the operating-model leadership inheriting the spatial architecture across the campus lifecycle.
Methodologically, the publication combines primary source review of governing standards (TIA-942-C, NFPA 75/76/70, ASHRAE TC 9.9, IEEE C2, NEBS GR-63-CORE, Uptime Institute Tier Topology, and related documents), engineering analysis of representative reference architectures across three deployment scales, capital and schedule modeling against representative cost and lead-time data drawn from 2024–2026 industry sources, and practitioner observations from direct hyperscale and colocation operating engagement. Quantified claims drawn from practitioner experience are explicitly labeled as such throughout the body.
The principal findings are: that the legacy white-gray binary is structurally inadequate to AI-factory density profiles above approximately 60 kilowatts per rack; that the three-zone framework with formally defined black-space scope produces measurable operating, capital, and governance benefits that the binary cannot deliver; that gray-space critical-path scheduling rather than white-space density is the binding constraint at hyperscale campus build; that cross-zone failure propagation is the highest-severity event class on the resilience register and is best contained by zone-aware operating procedures; and that the framework’s benefits accrue over the operating lifecycle rather than over the construction calendar, which mismatches with the prevailing quarterly metric cadence of the industry.
The principal recommendations are: that hyperscale and colocation operators adopt the three-zone framework as the canonical coordinate system for engineering, operating, and capital decisions; that architecture and operating teams reconcile against the formal boundary definitions before commissioning rather than after first incident; that capital allocation be planned with explicit zone-by-zone signatures rather than as uniform percentages of campus budget; and that governance structures be installed with named decision authorities for each zone and each boundary, with the named authorities documented in the campus operating record.
Executive Summary
Modern data-center architecture has outgrown the binary classification of White Space and Gray Space. The contemporary AI factory is a campus-scale system in which the largest, most expensive, and longest-lead components, utility substations, medium-voltage distribution yards, bulk diesel and natural-gas generation, battery energy-storage systems, cooling towers, dry coolers, fuel storage, microgrid controls, and the perimeter security and telecommunications outside-plant, sit physically outside the building envelope and functionally outside the gray-space doctrine.
This paper proposes and develops a three-zone spatial framework consisting of White Space, Gray Space, and Black Space. White Space is the controlled interior environment in which information-technology compute, storage, and networking equipment resides; Gray Space is the internal building support infrastructure that conditions and protects the White Space; and Black Space is the external, site-level, and utility-facing infrastructure on which the entire campus depends. The three zones are distinguished not only by physical location but by code authority, qualified-person scope, capital allocation, schedule criticality, telemetry topology, and failure-mode signature.
The argument develops along seven parts. Part I traces the origin of the white/gray distinction, defines each zone, and identifies the boundaries that separate them. Part II maps each zone to the codes and standards that govern it: TIA-942-C, ASHRAE TC 9.9, NFPA 75 and 76 inside the envelope; the National Electrical Code Article 110 working-clearance and Article 225 outside-branch rules at the boundary; IEEE C2 (the National Electrical Safety Code), NFPA 30 fuel storage, NFPA 110 emergency power, EPA NSPS Subpart IIII for stationary internal-combustion engines, and the Uptime Institute Tier topology in the yard. Part III addresses the engineering and operations of each zone, including density trajectories, thermal envelopes, power-conversion topology, telemetry and control architectures, and commissioning workflows. Part IV develops the capital, schedule, and risk profile of each zone, drawing on industry capex benchmarks for 2026 hyperscale construction and treating Black Space as the dominant determinant of campus critical path.
Ten findings frame the argument. First, the binary white/gray vocabulary obscures more than half of the capital and schedule risk surface of a modern AI campus. Second, the External infrastructure that the binary obscures is governed by different codes, different qualified-person scopes, and different AHJ authorities than the building interior. Third, the dominant schedule risk on a contemporary campus is the utility interconnect and substation, both of which are Black Space. Fourth, the dominant capital exposure on a 100 MW or larger build is Black Space, frequently exceeding the combined capital of White and Gray Space.
The closing recommendation is straightforward. Organizations engaged in modern data-center design, build, or operations should adopt the three-zone framework as their default spatial vocabulary; embed it in basis-of-design templates, reference architectures, requirement documents, sales collateral, RFP responses, commissioning plans, and operating procedures; and reorganize their internal RACI for capital projects so that Black Space has clear ownership distinct from the building owner and the IT customer. The framework is not expensive to adopt. The cost of not adopting it is paid in scope churn, schedule slip, capital surprises, code disputes, and the kind of incident reports that erode an operator’s standing with its customers and its insurers.
Fifth, the telemetry topology of a modern campus does not respect the white-gray binary: data center infrastructure management, building management, electrical power monitoring, and microgrid control systems each terminate in different zones, under different owners, and at different security boundaries, and the absence of a shared spatial vocabulary leaves the integration seams undefined. Sixth, the qualified-person scope that governs who may work in each zone differs by an order of magnitude in consequence, from the rack technician in White Space to the high-voltage switching authority in Black Space, and conflating the two in a single operating doctrine is a safety exposure, not merely a semantic imprecision.
Seventh, the failure-mode signature of each zone is distinct: White Space failures are typically thermal and contained, Gray Space failures are typically power-quality and recoverable through redundancy, and Black Space failures are typically utility-scale and capable of removing an entire campus from service for days. Eighth, the commissioning sequence that actually de-risks a campus runs from Black Space inward, because energization, cooling, and life-safety systems in the outer zones gate the integrated testing of everything they support, and a commissioning plan organized around the binary consistently underestimates the outer-zone critical path.
Ninth, modular and skid-based delivery is reshaping Black Space economics, moving substantial portions of power and cooling scope from field construction into factory production, and the binary vocabulary provides no clean place to account for this shift in constructability, schedule, and capital phasing. Tenth, the governance authority over the boundaries between zones is the single most frequently unassigned responsibility on a capital project, and the framework exists in part to force that assignment before it becomes the subject of a change order, a code dispute, or an incident report.
On the forward view, the publication treats the density transition as the controlling variable. Below approximately thirty kilowatts per rack the binary remains serviceable; between thirty and sixty kilowatts the seams begin to open; above sixty kilowatts, the regime that AI-factory deployments now occupy routinely, the binary mis-allocates capital, mis-sequences schedule, and mis-assigns authority. The three-zone framework is therefore positioned not as a forecast of a possible future but as a description of the operating reality that hyperscale and AI-class campuses already present to the teams that build and run them.
The recommendations follow directly from the findings. Operators should adopt the three-zone vocabulary as the default spatial language of their basis-of-design and reference-architecture libraries; capital teams should track White, Gray, and Black Space as distinct cost and schedule centers rather than folding the outer zone into a generic site allowance; and operations leaders should align qualified-person scope, method-of-procedure libraries, telemetry coverage, and incident response to the zone in which the work and the risk actually reside. The remainder of the publication develops each of these positions in technical, capital, and governance detail.
The full white paper is attached below.
