Designed for Scalable Tenant Operations

A proposed technical basis, not installed infrastructure

The Rev B design brief starts with separately maintainable blocks, overhead distribution and equipment-specific cooling. The final interior, reliability topology, operating responsibilities and recovery services must be agreed with the tenant and licensed designers.

Source basis: September 9, 2026 planning package · Not independently re-underwritten.

A / B

Proposed critical-path approach

1.25

Planning annual PUE target

Separate

Off-site recovery scope

Planning data & boundaries

System	Working design direction	Validation needed
Cooling	Direct-to-chip capability for AI; residual air cooling	Heat maps, coolant, CDU ownership, redundancy and leak tests
Heat rejection	Compare dry systems with hybrid / chiller trim	Weather, peak duty, water and energy performance
Electrical	Engineered A/B paths, UPS and standby	Coordination, fault domains, load steps and failure tests
Floor / overhead	Slab / hard-floor starting alternative; overhead services	Point/rolling loads, structural supports and access
Security / controls	Controlled access, monitoring and segregated facility controls	Actual scope, staff, escalation and tenant IT demarcation
Recovery	Separate data copies, links and replacement compute	Workload RPO / RTO, contracted capacity and tested restore

Interior construction

Rev B does not assume raised flooring is mandatory. It proposes a hard-floor alternative with overhead distribution, subject to engineered loads and tenant needs. Technical ceilings/support systems and tested fire/smoke/security partitions need their own designs. No final drywall, ceiling, acoustic or fire-system specification is issued here.

Reliability is an acceptance task

Test loss of utility, loss of power/cooling modules, UPS ride-through, cooling recovery and safe live-phase isolation. Separate A/B paths do not prove independent upstream utilities. PUE 1.25 is a planning target, not a measured guarantee, and low-water or closed-loop cooling is not a zero-water promise.

Remote recovery is not another full campus by default

Define acceptable data loss (RPO) and restore time (RTO) per workload. Off-site recovery needs retained data, usable network links, credentials, restore procedures and replacement compute. The source example shows why transfer time alone does not establish total application recovery time. No recovery provider or site is selected.

Energy independence remains a separate study

The working case is grid-backed with standby support. On-site solar, extended storage and continuous diesel operation are separate studies and costs. Annual solar energy matching does not demonstrate round-the-clock adequacy, and diesel islanding still depends on imported fuel, maintenance and permitted operation.

EPIC CONSTRUCTION MANAGEMENT LLC