An existing site,
upgraded for GPUs.
A building or a mining site with power, upgraded for GPU density: power distribution, liquid cooling and network. Then we build the cluster inside it.
- Site
- Building or mining site
- Upgrade
- Power, cooling, network
- Owner
- You
- Operator
- Your team or us
Power you already have, put to work.
Many sites have the power but not the density: a hall built for lighter racks, or a mining site built for other hardware. A retrofit closes that gap and keeps the site yours.
- Start from a live connectionA new power connection can be the longest item in a new build. A retrofit starts from one that exists.
- Keep the siteThe building or the land stays yours. The upgrades and the cluster are added to it.
- Building and cluster in one designThe upgrades are sized to the hardware that goes in, not to a guess.
What changes inside the building.
The assessment sets the exact scope for a given site. These are the four areas a retrofit usually touches.
| As built | After the retrofit | |
|---|---|---|
| Power distribution | Sized for lighter loads per rack | From the site's supply to every rack position, sized for GPU density |
| Cooling | Air, built for lower heat per rack | Liquid to the racks that need it, with heat rejection outside |
| Structure | Rated for the original equipment | Checked for GPU racks, and strengthened where the layout needs it |
| Network | Uplinks for the original use | The cluster fabric (RoCE v2 Ethernet, spine-leaf) and uplinks sized for it |
Illustrative of the work. The assessment sets what a given site needs.
Upgrades that follow the cluster.
The building doesn't have to be fitted out all at once. Where the site allows it, each area is upgraded ahead of the wave of nodes that goes into it.
- Scoped before you commitThe assessment and the proposal list every upgrade, with the cluster it serves.
- Owned by youYou own the upgraded site and the cluster. We design and build both.
- Run by us if you wantOnce the cluster is accepted, we can operate it to an SLA, or your team takes over.
What the upgraded space has to take.
The platform sets the scope. HGX B300 nodes run on air; GB300 NVL72 racks need liquid cooling and carry far more weight and power per rack.
- Rack scaleNVIDIA GB300 NVL72
- Next generationNVIDIA Vera Rubin
From an assessment to a running cluster.
Step 1: Assess the site
Power, structure, floor loading, cooling and connectivity, written up for you.
You getA site assessmentStep 2: Design both together
The retrofit and the cluster, so the upgrades match the hardware that goes in.
You getOne designStep 3: Upgrade the site
Power distribution, liquid cooling and network, area by area.
You getSpace ready for GPU racksStep 4: Build and run the cluster
Racked, cabled and burned in, then run by us or handed to your team.
You getA running cluster
Who owns and runs a retrofitted cluster.
The site stays yours. The cluster can be yours, run by your team or by us, and a running site can take our software on its own.
| Way to work with us | Owner | Design | Source | Build | Operate | Software |
|---|---|---|---|---|---|---|
| Cluster deploymentWe design, source and build it, then hand it over burned in. Your team runs it. | You | |||||
| Managed clusterYou own it, we run it. We can build it first, or take over a cluster that's already live. | You | |||||
| SoftwareYour team runs every stage, on our stack. | You | |||||
| Reserved clusterWe own and run it. You reserve it for 60 months, with rollover into the next generation. | Lightyard |
Owners of powered buildings and mining sites.
01Neoclouds
Capacity on a site you've secured.
You bring the site and the customers. We fit it out and build the cluster.
GPU clusters for neoclouds05Asset owners
A powered site turned into a GPU site.
You bring the building or the mining site. We upgrade it and build the cluster.
Clusters for asset owners06Investors
Capital put to work on an existing site.
You fund the upgrades and the cluster. We design, build and run them.
GPU infrastructure for investors
When a retrofit isn't the answer.
If the building can't take the load, a module on the same land or a colocation hall in the same region can carry the cluster instead.
- 03 Retrofit · this page
A building or mining site upgraded for GPU density.
Questions we get on the first call.
Something missing? Ask us directly. We usually reply within a few hours.
Buildings and mining sites that already have power. We assess structure, floor loading, cooling and connectivity before we propose anything.
It can, if the assessment shows the structure, cooling and network can be brought up to what GPU racks need. The site already has the power, which is often the hardest part to find.
When the building can't take the load, or the upgrades would cost more than a new module on the same land. The assessment shows which.
You keep the site and own the cluster. We design and build the upgrades and the cluster, and can operate the cluster.
It depends on the site and the upgrades it needs. The proposal states the schedule.
Permits depend on the site and where it is. The assessment lists what the upgrades need, and the proposal states who handles each one.
Solutions, sites and who we build for.
- AI factoryEvery stage, from the site to a running cluster.
- Cluster deploymentDesigned, sourced, built and handed over burned in.
- Managed clustersYou own the cluster. We build it or take it over, and run it.
- Reserved clustersWe own and run it. You reserve it for the term.
- SoftwareRun every stage yourself, on our stack.
- White-label GPU cloudYour GPU cloud under your brand, built and run by us.
- ColocationA Tier III facility in the region you choose.
- On-premYour data center, with the cluster designed around it.
- RetrofitA building or mining site upgraded for GPU density.
- ModularA prefabricated site on land with power.
- NeocloudsCapacity built to your standard, ready to sell.
- Venture capitalOne cluster, shared by your portfolio companies.
- Enterprises & sovereignYour own cluster in your region.
- Private equityGPU capacity on portfolio sites, or a neocloud as a platform.
- Asset ownersHardware, a hall or a power site, turned into a running cluster.
- InvestorsAn operated GPU asset, without building an operator.