Getting Started with PodStack

PodStack is a global GPU cloud for developers, built on a proprietary stack (PodVirt) rather than OpenStack or vanilla Kubernetes. You get fractional and full NVIDIA GPUs, per-second billing, zero egress fees, instant provisioning, and an ISO 27001-certified platform — available both as a managed cloud and as a licensable platform.

This section takes you from zero to a running GPU workload.

The four PodStack products

ProductUse it to…
QuickPodsDeploy production-ready AI stacks in one click — 1-click templates, built-in MLOps, fractional GPUs.
TrainPodsGet on-demand NVIDIA GPUs at low prices — provision, SSH, and move data from the podstack CLI.
InferenceServe open-source models on low-latency, OpenAI-compatible endpoints with autoscaling.
CLIDo it all from your terminal — an AI coding agent, live app previews, GPUs, fine-tuning, and inference.
DC SuiteOperate your own GPU datacenter — orchestration, BillOps and FinOps, licensable for self-hosting.

New here? Follow this path

  1. Create your account — passwordless sign-in with an email code or a social login.
  2. Create a project — every resource lives inside a project.
  3. Add an SSH key — how you’ll connect to pods and VMs.
  4. Launch your first pod — the end-to-end Quick Start.

Prerequisites

Before you begin, have ready:

  • A valid, permanent email address (temporary/disposable addresses are rejected).
  • A payment method to fund your wallet once you’re ready to launch paid resources.
  • Optionally, an existing SSH public key — or generate one in the portal.

What you can run

ResourceBest for
PodsContainerized GPU workloads with persistent storage — training, Jupyter, apps.
VMsFull virtual machines with GPU passthrough.
BaremetalDedicated physical GPU servers for maximum performance.
SandboxesFast, ephemeral compute environments.
NotebooksGPU Jupyter notebooks with checkpoint rewind (Time-Travel).
Inference endpointsOpenAI-compatible model serving with autoscaling.

Next step

Start with Creating Your Account.