For tenants
AI capacity, built around your workload.
Power, space and cooling for AI workloads across North America, built to your specification. Deploy your own GPUs or partner with us on compute.
Who we work with
Built for every kind of AI builder.
We serve five types of customers, each with distinct requirements and an engagement model tailored to them.
Private cloud
Dedicated, single-tenant capacity for private and enterprise cloud platforms.
- What you need
- Secure capacity under your control, built to your specification.
- How we help
- Single-tenant data halls designed to your requirements and run by our operations team.
Neoclouds
GPU cloud providers that need to add capacity quickly, in phases that track customer demand.
- What you need
- Capacity that keeps pace with the customers you sign.
- How we help
- Phased delivery on available power, so you add capacity as demand grows.
Internet and software companies
Capacity for AI features and inference at scale, without building your own data center.
- What you need
- Inference capacity close to your users, without a data center program.
- How we help
- Ready capacity, or a compute partnership with us.
Large enterprises and conglomerates
A long-term AI infrastructure partner for groups with demand across many businesses.
- What you need
- A dedicated, long-term AI infrastructure partner.
- How we help
- Dedicated sites and joint development options, from one accountable partner.
AI companies
Power and space for training and inference clusters, sized to your roadmap.
- What you need
- Power, density and cooling for the next cluster, on your timeline.
- How we help
- Build-to-suit capacity with high-density air or liquid cooling.
Something else?
If your requirements fall outside these categories, share your workload and needs, and we will structure an engagement around them.
Talk to our teamWays to work together
Four models. One accountable partner.
Choose a single model or combine them. Every agreement is structured around your workload, timeline and budget.
Build-to-suit
We design and build dedicated data halls to your specification and operate the facility. You bring and run your GPUs.
- You bring
- Your GPUs and requirements
- We deliver
- Dedicated data halls with power, cooling, network and security
- Best for
- AI companies, private cloud, neoclouds
Turnkey colocation
Move into ready capacity in our facilities, rack by rack or hall by hall, with power, cooling, network and security in place.
- You bring
- Your hardware
- We deliver
- Ready capacity, operated by our team
- Best for
- Neoclouds, internet and software companies
Compute partnership
For customers who prefer not to procure GPUs directly, we can work together on a compute arrangement suited to the workload.
- You bring
- Your workload and requirements
- We deliver
- A compute arrangement structured together with you
- Best for
- Internet and software companies, enterprises
Joint development
For long-term demand, we co-develop a dedicated site or campus with you, from power to operations.
- You bring
- Your long-term demand
- We deliver
- A dedicated site or campus, co-developed with you
- Best for
- Large enterprises and conglomerates
Compute partnerships
Flexible approaches to compute.
For customers who prefer not to procure GPUs directly, we can work together on a compute arrangement that fits your workload, timeline and budget.
How delivery works
From requirements to ready for service in 12 to 18 months.
We start from available power and land, not from a utility queue. That's how we deliver faster than new-build timelines.
- 01
Requirements
Power, density, location, timeline and how you want to work together.
- 02
Site match
We match you to available power and land in our North American portfolio.
- 03
Agreement
Capacity, structure, delivery date and terms, agreed up front.
- 04
Design and build
Our engineering team delivers turnkey, with conventional or modular construction selected for each site.
- 05
Ready for service
Commissioned, handed over and operated by our team.
Delivery timelines depend on site, scope, utility approvals and equipment lead times, and are confirmed in each agreement.
Power is already there
We build on energized sites, so projects skip the multi-year wait for a new grid connection.
Flexible construction
Conventional and modular construction options, selected site by site to shorten build and commissioning.
Funded from the start
Projects are capitalized at the outset, so construction doesn't wait on financing.
North American network
Working relationships with utilities, landowners, contractors and suppliers across North America.
What you get
Everything around the GPUs, delivered.
Each facility is designed around your workload and run by our operations team.
Request capacity- PowerEnergized capacity, with UPS, backup and room to scale
- CoolingAir or liquid cooling for high-density GPU racks
- NetworkHigh-capacity, redundant fiber
- SecurityControlled access, monitoring and staffed operations
- OperationsOn-site team and data center standards from day one
What we need from you
Five answers to get started.
Share what you have, and we will respond with options, a proposed structure and a delivery date.
- CapacityHow many MW you need now, and later
- DensityExpected kW per rack and cooling preference
- LocationPreferred regions in North America, or none
- TimelineWhen you need capacity live
- ModelCustomer-owned GPUs, a compute partnership, or a combination