Skip to content
GEOGRAPHYUSCanadaMexicoEuropeAsiaPLATFORMSH100H200B200B300GB300 NVL72MI300XSERVICESBare MetalManaged SlurmManaged KubernetesO&MSTANDARDTier IIISingle-tenant by defaultSPEEDDeployments in weeks
GEOGRAPHYUSCanadaMexicoEuropeAsiaPLATFORMSH100H200B200B300GB300 NVL72MI300XSERVICESBare MetalManaged SlurmManaged KubernetesO&MSTANDARDTier IIISingle-tenant by defaultSPEEDDeployments in weeks
BARE METAL GPU / SERVICE

Real hardware.
Root access.

Dedicated GPU nodes on a dedicated fabric, provisioned for one customer at a time. No hypervisor between you and the silicon, no shared interconnect, no throughput that quietly disappears into someone else's job.

At a glance
Tenancy
Single-tenant
Access
Root, bare metal
Fabric
InfiniBand / RoCEv2
Scale
8 GPUs to multi-MW
Facilities
Tier III
A01 / WHAT'S INCLUDED

What you get.

01

Dedicated nodes, provisioned to your image

You choose the OS, the CUDA and driver track, and the base image. We provision it, keep it on a tested version track, and hand you root. Nothing is virtualized and nothing is oversubscribed.

02

Non-blocking GPU fabric

InfiniBand NDR/HDR or RoCEv2 in a rail-optimized, non-blocking topology sized to the cluster. Full-cluster NCCL all-reduce results are part of the acceptance test you sign off on.

03

Storage that keeps the GPUs fed

Local NVMe scratch on every node plus a shared high-throughput parallel filesystem sized to your dataset and checkpoint pattern, on a separate storage network.

04

Private networking and isolation

Dedicated VLANs, private interconnect between your nodes, controlled egress, and optional direct connectivity back to your own network or cloud accounts.

05

Burn-in and acceptance testing

Sustained thermal and load soak, per-link error counters, storage throughput, and GPU health screening before handover. Weak components fail here, not in your first long run.

A02 / SPECIFICATION

The technical sheet.

Configured per deployment. These are the ranges we build within; your quote names the exact parts.

Compute
NVIDIA Blackwell
GB300 NVL72 / B300 / B200 HGX
NVIDIA Hopper
H200 / H100 SXM and PCIe
AMD
MI300X
Node form
8-GPU HGX or rack-scale NVL72
CPU / RAM
Dual socket, 1-2 TB typical
Fabric
Compute network
InfiniBand NDR/HDR or RoCEv2
Topology
Rail-optimized, non-blocking
Per-GPU bandwidth
Up to 400 Gb/s
In-node
NVLink / NVSwitch
Management
Separate out-of-band network
Storage
Local scratch
NVMe, per node
Shared
Parallel filesystem, sized per dataset
Object
S3-compatible, optional
Checkpointing
Throughput sized to run cadence
Facility and power
Standard
Tier III certified or equivalent
Density
Up to 130 kW per rack
Cooling
Air or direct-to-chip liquid
Redundancy
N+1 power and cooling
Geography
US / Canada / Mexico / Europe / Asia
A03 / WHO IT'S FOR

Where this fits.

Frontier and foundation model teams

Long, tightly-coupled training runs where a single degraded link costs days of wall-clock time.

Enterprises with data constraints

Teams that cannot run on shared multi-tenant infrastructure for security, residency, or audit reasons.

Platform teams building their own stack

Groups that want the metal and the fabric, and intend to run their own scheduler and tooling on top.

A04 / HOW IT'S OPERATED

Who carries the pager.

We do. Every item below is our responsibility for the life of the engagement.

  • 24/7 monitoring of GPU health, fabric errors, power draw, and thermals.
  • On-site sparing and RMA handling, so a failed node is swapped rather than ticketed.
  • Firmware, BMC, and driver lifecycle managed on a tested version track.
  • Fabric health checks and link-flap remediation before they show up as job failures.
  • A named engineer who knows your cluster, reachable directly, not through a queue.
A05 / REQUEST A QUOTE

Tell us the shape of the cluster.

An engineer replies with a configuration and a deployment window. No pricing on the website, no sales sequence.

contact@crystalcloud.ai / Los Angeles, CA

Request a cluster quote
~2 MINUTES

Tell us the shape of the cluster. An engineer replies with a configuration and a deployment window, not a sales sequence.

Single-tenant bare metal, managed end to end. Configuration and availability confirmed in writing before you commit.