SSP, Strategic Supply Partners Live inventory
The desk
GPU procurementNodes and clusters, landed to your rack ● Live inventoryVerified lines, indicative pricing
By platform
B300 and GB300Current-generation Blackwell H200 and H100New stock and fleet rotation All platformsThe full catalogue
Rest of the stack
Memory, servers and storageThe kit around the accelerators Networking and opticsTransceivers, switching, cabling
Capacity without capex
GPU-as-a-serviceReserved, committed spend, on demand Bare-metal leasingDedicated clusters on a term
Decide first
Cloud vs bare metalWhere each one actually wins Reserved vs on demandThe break-even, calculated Cluster cost calculatorThe whole bill, not the GPU line
Space and power
Colocation and powerSingle racks to 100 MW Above 40 kW per rackWhere the shortlist collapses
Where
United States Europe Asia Pacific Middle East and Africa All regions →
Supply side
Have capacity to fill?Tell us what you hold
Analysis
InsightsWhat things cost, with the numbers Sizing guidesMemory, nodes, power, cost ComparisonsBuy, lease or rent
Start here
How an engagement worksThe first week, held to dates Find your fitThree questions, the right desk Glossary
Talk to the desk
The stack · Networking & optics

The network nobody prices until it hurts.

A GPU cluster is a network with computers attached. Fabric choice, transceivers and cabling are routinely the most under-planned line in an AI budget, and third-party optics carry some of the steepest markups in the industry. We source the whole layer buyer-side, priced with the cluster rather than discovered after it.

Scope this layer Find your fitAcknowledged within one hour, first sourcing pass within one business day.
What the layer includes

Fabric to fibre, one sheet.

Scale-out fabric first: InfiniBand or high-speed Ethernet, switching, and the topology the workload actually needs. Then the physical layer everyone forgets to budget: transceivers and cables in quantities that surprise first-time builders, because a serious cluster consumes optics by the hundreds.

The optics markup is the quiet scandal of cluster building: identical-spec transceivers vary in price several-fold depending on the channel. Sourcing them with the cluster, buyer-side, is one of the easiest large savings in the whole build.

Between sites, the same desk covers data-centre interconnect: private low-latency links between facilities, clouds and storage when the workload spans locations.

Fibre optic patch panels and transceivers inside an AI cluster fabric
Where to next

Start from what is verified.

InfiniBandHardware sourcingLive inventory
Straight answers

Asked first, answered straight.

InfiniBand or Ethernet for AI clusters?

Distributed training at scale still favours InfiniBand; modern high-speed Ethernet closes the gap for many workloads at better economics and simpler operations. The honest answer depends on job sizes and your team, and we put both configurations side by side.

How big is the optics line really?

Larger than almost anyone budgets: a serious multi-node fabric consumes transceivers and cables by the hundreds, and at third-party markups the line can rival the switching itself. Quoted with the cluster it is dramatically cheaper than bought in a panic later.

Can you connect two sites into one fabric?

Between buildings we provision DCI: private, low-latency, appropriately sized links. A single scale-out fabric across distant sites is not how these systems are built, and anyone promising it deserves sharp questions.

Do you supply compatible third-party optics?

Yes, verified for the platform and warranted, at the price the channel should charge. Where OEM-branded optics are mandated by support terms, we say so rather than quietly substituting.

Talk to the desk

Working through this on a real requirement?

Twenty minutes with the desk, no pitch and no quote at the end of it. Tell us roughly what you need and we will come back within one business day.

Acknowledged within one hour, first sourcing pass within one business day.
✓

Sent. We are on it.

Your enquiry has landed with the desk. Acknowledged within one hour.