Skip to main content

Platform

Compute

Status: Waitlist

Dedicated GPU capacity for training, fine-tuning and inference, planned on NVIDIA HGX B300. Phase 1 is being built, and nothing is in service yet.

What it is

Single-tenant GPU servers, planned on NVIDIA HGX B300, that you will use by the hour or reserve for a term. Phase 1 is planned to draw on Nepal’s national grid, which the IEA records as almost entirely hydropower.

Where it stands today

Phase 1 hardware is on order, and nothing is racked. The build, step by step:

Phase 1 build status
StepStatus
Phase 1 hardware: on orderStatus: In progress
Phase 1 data centre: a third-party facility in Nepal, under NDAStatus: In progress
Network bandwidthStatus: In progress
GPU deploymentStatus: Planned
General availabilityStatus: Waitlist

Rates for the planned capacity are on the pricing page. They are Indicative: nothing can be ordered yet.

Indicative pricing

What comes next

In order, without dates: a date is published only once it is certain.

  1. Deploying and testing the Phase 1 hardware in the data centre.
  2. Opening capacity to developers from the waitlist, with its status updated on this page.
  3. Publishing measured energy use once the hardware is in service and metered.

Build with CloudFARM

Developer accounts open from the waitlist once capacity is connected. Until then, tell us what you want to build.