Pacific

Pacific builds factory-produced, modular micro data centers that combine compute, cooling, and power distribution in a single unit, designed to deploy on sites that already have grid connections. The approach addresses delays in traditional data center construction by placing standardized pods on prepared pads, with factory work and site preparation proceeding in parallel.

The reference module, Pod 32, integrates 32 Supermicro HGX B300 nodes with 256 NVIDIA B300 GPUs, rail-optimized 400G Ethernet, 983 TB raw NVMe storage per module, and direct-to-chip liquid cooling with redundant CDUs. Capacity is single-tenant with bare-metal access. Pacific offers four deployment paths: reserved GPU compute for training and inference, managed on-premises pods, colocation for owned hardware, and bridge capacity for sites awaiting a permanent build. Deployments are operated remotely with contracted regional field service.

Pacific targets organizations with sensitive data and compliance requirements, including aerospace, defense, and financial services workloads involving CUI and CMMC considerations. Capacity is offered under short or long-term contracts, with deployment timing confirmed per site and targeted to roughly three months from pre-payment.