Skip to content
// FLEET
us-west-1GB300 · liquid
eu-central-1B300 · liquid
apac-sg-1GB200 · NVL72
me-uae-1VR200 · Rubin
// DOCS

Getting started

Segal Compute gives you NVIDIA GPU capacity two ways: Segal Cloud for on-demand, reserved, and spot instances, and Segal Deploy for physical AI servers and racks delivered to your datacenter or a Segal facility.

Create an organization

Sign in with an email magic link or Google, then create an organization. Organizations hold your instances, billing, quotas, and team. Invite teammates with a scoped role (owner, admin, billing, member, read-only).

Launch your first instance

  1. Open Console → Instances → Launch instance.
  2. Pick a SKU, a region, and an image (CUDA, PyTorch, JAX, Triton, vLLM, or bring your own).
  3. Choose a tier — on-demand, reserved, or spot.
  4. Launch, then watch the instance move pending → provisioning → running.

Billing accrues per GPU-second at the SKU rate times the region multiplier. The cost-so-far meter updates live.

Architecture

The Console runs on the application layer. GPUs, provisioning, and metering live in a separate Control Plane reached over a typed API — Segal hosts the control surface, the fabric runs the silicon.