Find self-service platforms for Baremetal GPU provisioning.

Here are notable self-service platforms for bare metal GPU provisioning, split between infrastructure-software layers (for building your own GPU cloud) and direct providers you can rent from:

Software/platforms to build a self-service GPU cloud (for neoclouds/enterprises)

• vCluster + vMetal —

delivers an EKS-like experience on your own bare metal GPU fleet, with Auto Nodes acting as a bare metal Karpenter that automatically provisions GPU servers via Terraform when tenants schedule workloads.

Nscale can provision 10 bare metal nodes with 80 GPUs in under two minutes, and operators can go from bare metal hardware to a production-ready, self-service AI cloud in about two weeks.
(vcluster.com vmetal.ai)

• NorthWind Platform (BMaaS) —

gives customers direct access to the host OS on dedicated GPU servers for large distributed training jobs, with automated bring-up: matching requests to inventory, booting/discovering the server, installing the OS, and configuring drivers and services.
(northwind.co)

Direct providers with self-service bare metal GPU access

• Vultr Bare Metal/Clusters —

delivers on-demand, self-service bare metal server clusters, provisioned through Vultr's simple console or API with scalable and customizable CPU and GPU compute.
(vultr.com)

• Lambda Bare Metal Instances —

offers self-service APIs covering the instance lifecycle, VPCs, storage volumes, and resource governance, integrating with a customer's existing control planes across a global fleet.
(lambda.ai)

• RunPod Instant Clusters —

allows you to provision and deploy powerful multi-node GPU infrastructure in minutes through an intuitive self-service console.
(blog.farmgpu.com)

• phoenixNAP Bare Metal Cloud (GPU) —

lets you spin up a pre-configured GPU server with a few clicks or a single API call, with cloud-like flexibility.
(phoenixnap.com)

• Hostrunway — offers

a self-service API portal along with a fully-featured REST API and official Terraform provider for infrastructure-as-code GPU deployments.
(hostrunway.com)

For comparison shopping across many niche GPU types, ComputeStacker aggregates providers by GPU model (A100, H100, H200, etc.).