MTL-3 Specifications
In March 2025, Cerebras selected WhiteFiber to design and deliver a Tier III-compliant retrofit data center on an aggressive six-month timeline. Requirements included direct liquid cooling, redundancy, resiliency, and precision environmental control.
Data Center Layout
203,800 SF built
Direct-to-chip liquid cooling designed facility
Purpose-built and designed for a sole tenant
Diverse meet-me rooms
24x7x365 remote hands availability
Energy Distribution
5MW ready
N+1 UPS and N+1 generators
Fuel supply contracts in place
Generator testing: weekly
Network Infrastructure
Neutral meet-me rooms
Diverse fiber entrances into the building
3 carriers servicing the data center
Connectivity to the Montreal Largest Carrier Hotel
Connectivity to CanIX (Canadian Internet Exchanges)
Connectivity to Enovum MTL-1
Environmental Controls
Raised floor configuration
Custom containment
N+1 redundant cooling design
Advanced monitoring
Safety and Security
VESDA (very early smoke detection)
Double-interlocking pre-action sprinkler
24x7x365 monitored surveillance
Keycard and biometric access system
Biometric temperature monitoring at entry
Approach
While typical greenfield builds require 18+ months to deliver, the Montreal 3 site was delivered in six months. Leveraging over 20 years of experience in retrofits and large-scale deployments for hyperscalers, advanced supply chain partnerships, and focused execution, Montreal 3 was delivered on-time while meeting all build-out requirements.
In six months WhiteFiber turned a former mattress factory into a custom-built, world-class, AI-optimized data center.
Complex Build - Record Time
Modular data center build approach (replicable DTC deployments) and a specialized procurement ecosystem enabled delivery in six months.
Direct-to-Chip Liquid Cooling
Custom-engineered loops provided precision thermal management, enabling rack densities unachievable with conventional air cooling.
Tier III Concurrently Maintainable Design
Achieved 3→2 redundancy across critical systems, ensuring uptime guarantees with efficient capital deployment.
TIMELINE
Build AI Infrastructure That Operates with Precision
Whether it’s dense colocation, turnkey private AI environments, or optimized GPU reservations, WhiteFiber delivers infrastructure engineered for innovation and reliable production workloads
.svg)
Preferred Partner