AI infrastructure for
private equity portfolios.
Portfolios already hold much of what AI infrastructure needs: power, land, buildings and data centers. We're the team that turns them into running GPU clusters, or into a neocloud of their own, without the portfolio company hiring one.
- Starting point
- Power, a site or a platform plan
- Owner
- The portfolio company
- Operator
- Lightyard, to an SLA
- Brand
- The portfolio company's
From powered land to a neocloud.
Each step up can earn more from the same megawatts, and asks for more capital and more capability. Most portfolio companies can deliver the first step. We take on the next three.
- Dates for the planEvery step is dated in weeks from signing, so it can go into the value creation plan.
- No team to carryDesign, build and operations come under contract, not as hires the company keeps after the build.
- Stop at any stepLease an AI-ready building, sell a cluster's capacity or run a neocloud. The company chooses how far to go.
GPU capacity on sites the portfolio already holds.
Data centers, power plants and mining sites in a portfolio often have the power a GPU cluster needs. We assess the site, upgrade it or add a modular data center, then build and run the cluster on it.
- Starting from the powerWe work from the power and cooling the site has today, and what reaching GPU density would take.
- Retrofit or modularAn existing building upgraded for liquid-cooled racks, or a prefabricated module on the land beside it.
- Run under contractThe portfolio company owns the cluster. We operate it to the SLA in our contract with it.
A neocloud as a new portfolio company.
The company puts a management team and a brand in front of customers. Everything behind it, from the sites to the software customers log in to, comes from us under contract, with the company's name on it.
- A small team to startThe platform hires for sales and product. Engineering and operations come from us, under contract.
- Under the platform's brandIts customers see its console, docs and invoices, with no mention of us.
- Capacity added cluster by clusterEach new cluster joins the platform on a dated plan, with its own build scope.
What you can underwrite, and what stays with you.
Builds like these are often financed partly with debt secured on the hardware. Lenders ask for fixed terms and complete records, and we provide both. Utilization stays with the company.
- 01Fixed in the contract
- Build scope and price
- Dates in weeks from signing
- Acceptance tests per node
- The operations SLA
- A reporting schedule
You get Terms you can model
- 02Records for lenders
- Design and bill of materials
- Acceptance results per node
- Inventory by serial number
- SLA and repair reports
You get The file a lender asks for
- 03Not promised
- Utilization
- Customers for the capacity
- Revenue or returns
What a buyer will diligence.
A GPU business built around an in-house team is harder to sell than one whose records, contracts and operations are in order from the first rack.
| An in-house build | Built and run with us | |
|---|---|---|
| The team | Engineers hired for the build, and the key-person risk that comes with them | An operations contract the company holds |
| The records | Whatever the team kept | Design, bill of materials and acceptance per node, from the first rack |
| Operating history | Tickets, spreadsheets and memory | Health, failures and repairs, node by node, on one system |
| The asset | In the company's name | In the company's name, with its build file |
Compared by structure, not by firm.
Sites in the portfolio, or new ones.
A retrofit or a modular data center where a portfolio company has power, a Tier III colocation facility where it doesn't.
Questions we get on the first call.
Something missing? Ask us directly. We usually reply within a few hours.
Yes. We review its power, cooling, building and network, and set out what a GPU cluster there would take before anything is signed.
Data centers with spare power, power generation sites and mining sites are the usual ones. The design stage shows what each needs: a retrofit, a modular data center or neither.
The portfolio company, or a vehicle the fund sets up. We work under build and operations contracts and don't take title.
Yes. Its customers see its console, docs and invoices, and we operate what's behind them.
The design and bill of materials, acceptance results for every node, the inventory by serial number, and SLA and repair reports on the schedule the contract sets.
No. Selling the capacity is the portfolio company's job. We design, build and operate, and don't promise utilization, tenants or returns.
Each build is quoted as a project, and operations as a term contract with an SLA.
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.