How it works

What happens
after you sign.

Capacity arrives in waves. Each one is deployed, burned in and handed over before the next.

Delivery schedule · Illustrative
Dates are set per wave at signing
Six steps

From first call to the end of the term.

  1. Step 1: Scope

    We size the cluster, choose the GPU and region, and agree the wave plan with your team. The smallest cluster is 8 nodes or one NVL72 rack.

    You getSizing and a wave plan
  2. Step 2: Contract

    Proposal, then MSA. Prepayment is required. It is paid into escrow with an independent escrow agent, which releases it in stages as supply, installation and acceptance milestones are met. Delivery dates run from signing.

    You getA proposal and an MSA
  3. Step 3: Deploy

    Capacity is deployed for you, one wave at a time. HGX B300 takes about 10 to 14 weeks from signing, and larger clusters arrive in monthly waves of 32 nodes.

    You getA date for each wave
  4. Step 4: Burn-in and handover

    Each wave is stress-tested before you get access. You take it as bare metal with root on every node, or as managed capacity where we also run the software layer.

    You getAccess to each wave
  5. Step 5: Operate

    We monitor the cluster, replace failed parts and support it for the full 60 months, to the SLA in your contract.

    You getSupport for the full term
  6. Step 6: Later in the term: continue, roll over or exit

    The contract includes an exit option, on terms set out in your proposal. Part of the commitment can move into a newer GPU generation when it ships. Or continue to the end of the term.

    You getOptions during the term
Two ways to take it

Bare metal or managed.

Pick at scoping. The hardware is the same either way.

  • 01

    Bare metal

    Root on every node, no hypervisor, your OS image and your scheduler. Firmware and BMC stay with us.

  • 02

    Managed

    We also run the software layer on top, such as a scheduler for containers and jobs, or inference endpoints.

Either way, it is operated

Same single-tenant hardware either way, and either way we monitor it, replace failed parts and run the cluster to an SLA. Support runs for the full 60 months.

Get started

Start with one rack.

Tell us the size and timing you have in mind. We usually reply within a few hours and come back with a written proposal.

sales@lightyard.ai