Infrastructure models

Powered land.
Powered shell.
Different starting points.

Before comparing a price per megawatt, establish what is delivered and what you still need to build.

Discuss a project

What do the terms mean?

Powered land generally describes a site marketed with access to electrical capacity. The phrase alone does not establish an energized connection, transferable service rights or a particular readiness date.

Powered shell generally adds a building or modular enclosure and an agreed infrastructure scope. Cooling, redundancy, fit-out and commissioning boundaries vary by contract.

GPU cloud capacity gives the customer access to computing resources under a service model. It addresses a different purchase decision from owning or leasing physical infrastructure. For QIT’s compute services, see API and GPU pricing.

Compare the handover boundary

QuestionPowered landPowered shellGPU cloud
What is contracted?Land rights and specified power access.Space or modules plus defined supporting infrastructure.Compute access and service terms.
Who supplies GPUs?Define separately.Often a tenant or operator; verify the agreement.The service provider.
What proves readiness?Documented usable service and site conditions.Accepted infrastructure against a defined load and test scope.Available compute meeting the service specification.
What needs attention?Remaining construction and electrical works.Cooling, redundancy, maintenance and responsibility boundaries.GPU configuration, network, availability and usage charges.

The contract matters more than the label. Request an itemized boundary from the utility connection through the racks and heat-rejection equipment.

A published example: Energy Vault and Crusoe

Energy Vault’s July 2026 Snyder announcement describes an initial 8 MW powered-shell project with potential expansion to 25 MW. Energy Vault said it would buy ten Spark units and lease them back to Crusoe. The announced site scope includes electrical infrastructure, N+1 redundancy and battery storage.

This illustrates how the infrastructure owner and compute operator can be different companies. The initial capacity and Q1 2027 operating target were project announcements, not evidence of completed commissioning.

Source: Energy Vault’s Snyder project announcement. This is an industry example, not a QIT project or partnership.

Questions for a site owner or operator

  • What power capacity is available at handover, at which boundary and under what service conditions?
  • Which party supplies transformers, switchgear, UPS or batteries and cooling equipment?
  • Who owns the GPU hardware, and who integrates the network?
  • Which tests establish that the facility can sustain the contracted load?
  • Who maintains the shared infrastructure and responds to an outage?
  • Which charges cover electricity, capacity, maintenance and expansion?
  • What happens if a utility milestone or equipment delivery moves?

For a modular development, record these answers at both module and site level. A working container does not demonstrate that the shared infrastructure can support every planned expansion phase.

Start with the site and workload

QIT scopes modular GPU infrastructure around the site, usable power and intended workload. Discuss the preferred ownership and operating arrangement at the start; it changes the equipment scope, acceptance criteria and handover plan.

Read the Texas site selection checklist and Crusoe Spark comparison, or discuss a project.

Prepared by QIT · Updated November 2, 2026

Explore all field notes

Let’s put your
power to work.

Start with your site, your workload and the compute you need.

Discuss a project
hello@qit.energy