Skip to content
CAPACITY & COMPUTE

Capacity for owners. Compute for AI teams.

Two services for the facility after it is energised: finding the tenants, and leasing the GPUs.

GPU COMPUTE LEASING

NVIDIA B100 / B200 clusters by the node, rack or cluster

Blackwell-class GPU capacity for training and inference. No facility, no power contract, no capital budget. Reserve a few nodes for a fine-tuning run or a cluster for a multi-year programme.

Capacity is hosted in facilities in our project network, data centres we built, converted or equipped, so we know the power, cooling and network behind every rack. Configuration, availability and pricing are confirmed per request.

  • Training, fine-tuning and inference
  • Bare-metal nodes or managed Kubernetes on request
  • Reserved capacity with defined start dates
  • Monthly, annual and multi-year terms
Request GPU capacity
GPUs
NVIDIA B100 and B200 (Blackwell), HGX configuration
Node
8 GPUs per node, NVLink; CPU, memory and NVMe sized per workload
Fabric
InfiniBand or RoCE for multi-node training; separate storage network
Scale
Single node to multi-rack clusters; expansion within the same facility
Terms
Monthly reserved capacity to multi-year contracts; pricing per request
Hosting
Facilities in our project network; location and compliance matched per tenant

Specifications are indicative. GPU model, node configuration, fabric and location are confirmed in a written proposal.

TENANT SOURCING

Capacity to fill? We find the tenants.

An empty hall costs money every month. We match owners with capacity to operators, AI companies and enterprises that need it, and structure a lease both sides can sign.

The service covers facilities we built and ones we did not. Because we design and equip data centres, we can tell quickly whether a tenant's density, cooling and redundancy requirements fit your building, and what it would take to close the gap.

  • Tenant sourcing and qualification against your capacity, density, cooling and term
  • Colocation, wholesale and powered-shell lease structures
  • Technical fit review, with upgrade options costed
  • AI teams on our compute side can anchor a new or converted facility
Fill your capacity

What we need from an owner

  1. 01Location and available power (MW), with utility and redundancy details
  2. 02Hall layout, rack count and design density (kW per rack)
  3. 03Cooling type and capacity: air, liquid, or upgrade path
  4. 04Target lease structure and term
  5. 05Ready-for-service date, or current status if under construction
HOW IT WORKS

From requirement to signed term

The same four steps whether you offer capacity or need it.

  1. 01

    Requirements

    A short call to capture capacity, density, location, compliance and term.

  2. 02

    Match

    We qualify counterparties from our owner and tenant network and confirm technical fit before any term sheet.

  3. 03

    Proposal

    Configuration, location, start date, term and pricing in writing. Site visits and due diligence arranged.

  4. 04

    Contract and onboarding

    Lease or compute agreement signed. Onboarding, access and support defined. Field service and spares stay available.

Have capacity, or need it?

Tell us which side you are on and the numbers you are working with. We reply within one business day.

Start the conversation