A data center in a 40-ft container
One product, engineered to do one thing exceptionally well: drop a complete, GPU-ready AI data center at your location and have it running in twelve weeks. Pre-racked, pre-cabled, pre-validated.
40-ft ISO · 1 MW · 256 GPUs
A complete data center in a 40-foot ISO frame. Drag to orbit the unit, then tap a hotspot to inspect the cooling, power, compute, and network subsystems. Delivered to any site with power and a pad, powered on within 72 hours of arrival.
Stage 1 configuration
Every container is scoped to your workload — GPU type, power, cooling, connectivity, and any sovereign or air-gap requirements. Below is the baseline Stage 1 build that ships in our first wave.
- Form Factor
- 40-ft ISO container
- GPU Compute
- 256 H200-class GPUs (Stage 1)
- Power Envelope
- Up to 1 MW per container
- Power Source
- Behind-the-meter gas or grid
- Backup
- 48–72 hr LNG buffer for continuous uptime
- Cooling
- Direct liquid-to-chip, sealed loop
- Connectivity
- Fiber + MPLS or fully air-gapped
- Security
- Customer-held HSM keys · your data only
- Site-Ready to Live
- 12 weeks
One container, many missions
Enterprise AI Teams
Your AI workloads are too sensitive for public cloud. You need dedicated GPU capacity at your own site — without a two-year build cycle and a procurement gauntlet.
Hyperscalers & Neoclouds
Your GPU cluster is at capacity and demand is spiking faster than your permitting timeline. Data45 deploys burst capacity at an existing site in 12 weeks.
Regulated & Sovereign Buyers
Your jurisdiction requires data to stay in-country, or your security posture requires a fully air-gapped environment. Your HSM keys, your control.
Energy & Remote Operations
Your operation is in the field — a refinery, a remote facility, a defense installation. Data45 containers run on behind-the-meter gas, turning stranded energy into AI.
Government & Defense
Sensitive government AI requires dedicated hardware, physical security, and independence from commercial cloud. Data45 is purpose-built for this requirement.
Research & HPC
Universities and research orgs need serious GPU capacity for training runs and scientific AI. We deploy at your campus — not in a shared cloud where jobs compete for allocation.
Want the spec tailored to your site?
Our team will scope GPU type, power, cooling, and compliance around your exact requirements.