Nova Array
The Rapid-Deploy
Data Center.
Turn contracted power into active power in 180 days.
Nova Array is a liquid-cooled rapid-deploy data center for GPU compute. Compute, cooling, power, and distribution, delivered as one integrated set of data center modules. Built to deploy the latest silicon at scale, fast, on the sites you already have.
About Nova Array
180-Day Deployment
From purchase order to delivery. Versus 18 to 24 months for a traditional build. Prefabricated and factory-tested before leaving our manufacturing facilities. Deployment-agnostic across greenfield, brownfield, or alongside existing facilities.
Engineered with you
Our engineering team works alongside yours, accelerated by AI-assisted CFD (computational fluid dynamics) and design exploration. Every Nova Array is designed around your requirements: your power envelope, density, and footprint, without the delay penalty.
Designed using AI, built for AI
Our in-house AI models are used at the atomic level to discover the materials and coolants that make next-generation cooling possible, and to accelerate the engineering decisions that shape every Nova Array. See how we work →
Up to 250 kW per rack today
Single-phase direct-to-chip cooling, built for NVIDIA, AMD, Super Micro, and the latest silicon. Two-phase architecture engineered for 650 kW+ as next-generation TDPs (thermal design power) arrive. Future-proofed thermal architecture across the GPU roadmap.
PFAS-free and built for heat reuse
Two-phase architecture engineered around a PFAS-free dielectric, ahead of tightening EU REACH and US state regulations. High thermal recovery rates across single-phase and two-phase, designed to meet European 2030 sustainability mandates.
Built for the GPU roadmap
Learn more about our coolingNova Array is engineered for the latest GPU generations from NVIDIA, AMD, Super Micro, and others, supporting GB200, GB300, Vera Rubin, MI400-class, and the silicon roadmap ahead.
Single-phase direct-to-chip cooling is in production today. Two-phase direct-to-chip is engineered for the next generation, as Rubin Ultra, Feynman, and equivalent-class chips push beyond what single-phase can cool.
How we deliver
Built for neoclouds, colocation operators, sovereign AI programs, hyperscalers, and frontier AI labs.
Site Survey &
Load Modelling
We assess your site, cooling, and power requirements. AI-assisted CFD against your specific power profile and footprint. Module count and topology lock.
Factory Build &
Acceptance Test
Modules assembled and pre-commissioned in our manufacturing facilities, in parallel with your site preparation. Full thermal and electrical validation before anything leaves the factory.
Ship, Bolt & Connect
Modules arrive on site and connect to power and network. Ready for commissioning and live compute.
Talk to our team
Tell us your deployment timeline, site constraints, and target rack density. We'll come back with a technical scoping session and a proposal designed for your deployment.