What does “whole-server” actually mean?
You get the entire physical machine — bare metal, full root access, no hypervisor and no other tenants. Every GPU, NIC, and PCIe lane is yours.
How is it billed?
A flat price for the whole server. There's no per-instance metering, no utilization tracking, and no surprise overage — you know the number up front.
Which GPUs do you offer?
NVIDIA H100, H200, B200, and B300 nodes, typically eight GPUs per node. Blackwell nodes include NVIDIA ConnectX-8 SuperNICs on board.
Can I run multi-node training clusters?
Yes. Single nodes run on internal NVLink, and multi-node clusters can use 400 Gb/s InfiniBand or 800 Gb/s NVIDIA SN5610 Spectrum-X so large jobs scale across racks as one fabric.
Why does bare metal help enterprise security teams?
It gives security teams a clear physical asset boundary, root control over the OS and driver baseline, and a known place to enforce IAM, encryption, logging, network policy, patching, and incident-response controls. Bare metal does not replace those controls; it makes the infrastructure boundary easier to govern.
How should we compare GlassPearl against AWS, Azure, or Google Cloud?
Compare the same GPU class, GPU count, expected uptime, and production workload. For steady production use, GlassPearl can be more than 60% lower in published H100 and B200 cloud comparisons, with B300 still far below published AWS pricing.
What does the quote include?
We spell out the node type, GPU count, topology, facility assumptions, and pricing so the full-server handoff is clear before you move forward.
What if I already own the hardware?
We also broker power-dense colocation — sourcing the cabinets, power, and scale-out fabric-ready space your cluster needs, and matching you to the right facility.
Where are the servers hosted?
In power-dense colocation engineered for sustained GPU draw, with up to 80 kW per cabinet and 415 V 3-phase power. Nodes are racked, cabled, and burned in before handoff.
How do I get started?
Send your workload details through the contact page — GPU count, node type, topology, and timeline. We come back with hardware, topology, and pricing.