GPU Colocation and AI Compute Infrastructure
An energy-first campus is only half the answer. Greco brings the other half inside: turnkey, NVIDIA-accelerated AI compute built on the Dell AI Factory — racked, liquid-cooled and powered by the campus it sits in. Capacity that is already energized and already populated with current-generation accelerators.
Power and Compute as One Asset
Most operators hand you empty racks and a multi-year wait for power. Greco delivers both. We build the campus, generate the power behind the meter, engineer the liquid cooling, and bring NVIDIA-accelerated compute in on top — integrated as a single asset.
We procure and deploy through the Dell AI Factory. That means the same validated Dell PowerEdge XE-series and NVIDIA platforms the hyperscalers run, specified to the workload and installed in a building that already has its own power and cooling — instead of waiting in an interconnection queue for a substation that hasn’t been built.
AI infrastructure and managed IT services
GRECO Energy delivers next-generation AI infrastructure and managed IT solutions for enterprise and high-performance computing environments. Our services integrate advanced compute, servers, networking, data centre infrastructure and energy-efficient technologies through leading technology partners. From system integration and deployment to managed infrastructure and ongoing support, GRECO provides scalable solutions designed for the future of AI.
GRECO Energy is an AI infrastructure developer, managed service provider and systems integrator developing high-density AI data centre infrastructure in Canada and the United States. GRECO integrates advanced compute infrastructure with power generation, liquid cooling, energy storage and thermal waste-heat recovery for district energy and other beneficial uses.
The Platform We Deploy
Built on the Dell AI Factory with NVIDIA — the proven, validated toolchain, specified per campus.
Dell PowerEdge XE9785 / XE9785L
Dense GPU servers in air-cooled (XE9785) and direct-to-chip liquid-cooled (XE9785L) configurations — the workhorse of the Dell AI Factory.
Dell Integrated Rack IR7000
The liquid-cooled systems arrive as a rack-scale solution — power, cooling manifolds and compute integrated and factory-validated.
NVIDIA HGX B300 — Blackwell Ultra
Eight-GPU NVL8 nodes built on NVIDIA’s current-generation accelerators, paired with high-core AMD EPYC CPUs, terabyte-class DDR5 and NVMe.
NVIDIA ConnectX-7 400G fabric
Four-hundred-gigabit networking so clusters scale as one machine — the east-west fabric AI training and inference demand.
Configurations track Dell’s published PowerEdge XE line and NVIDIA’s reference designs; each deployment is specified to the tenant’s workload and density.
Browse the full Dell PowerEdge line we supplyEngineered for Liquid Cooling — and Heat You Can Sell
The liquid-cooled XE9785L / IR7000 line is where Greco’s energy-first model compounds. The heat these accelerators produce is captured at 50 °C and exported to a neighbourhood energy utility. Most data centres pay to vent that heat; a Greco campus turns it into a revenue stream and a community benefit.
See how we model the heat-export loopGPU colocation, without the power problem
If you are shopping for GPU colocation, you already know the constraint is not rack space.
Conventional colocation was built for 5–10 kW racks. An eight-GPU Blackwell Ultra node draws roughly 10 kW on its own, so a single modern AI rack can want more power than a traditional facility allocated to an entire row. That is why operators shopping for GPU colocation keep hearing the same answer: space is available, power is not, and the facility's own utility connection is the thing standing in the way.
GRECO campuses generate behind the meter, so the capacity on offer is not contingent on a substation upgrade or a place in an interconnection queue. Halls are engineered for direct-to-chip liquid cooling from the outset rather than retrofitted around air-cooled assumptions, at the 600 kW to 1 MW rack class.
The commercial difference is worth stating plainly: most colocation providers resell grid power and pass through demand charges. Here the power is generated on site and the waste heat is sold into district energy, which is a materially different cost structure — and the reason capacity can be offered against a schedule rather than a queue position.
Deployments run from single racks of Dell PowerEdge XE-series hardware through to multi-megawatt halls, either as colocation you populate yourself or as fully managed capacity where GRECO specifies, racks, commissions and operates the fleet. If you would rather take the building and fit out the interior yourself, that is a powered shell.
Can a standard colocation rack support 50 kW of GPUs?
Not every facility can. A rack's physical space does not establish its usable power or cooling capacity. A 50 kW GPU deployment needs an assessment of the specific hall, electrical allocation and server cooling requirements; a facility-wide megawatt figure is not a per-rack commitment.
- Power: confirm the continuous capacity allocated to the rack, compatible distribution and the capacity available under the agreed redundancy design.
- Cooling: check the server manufacturer's airflow or liquid-cooling requirements against the hall's heat-removal capacity and, where needed, its coolant distribution infrastructure.
- Space and access: verify rack dimensions, equipment weight, floor loading and the delivery path before planning the installation.
- Connectivity: include the network fabric, storage access and cabling layout in the deployment plan.
Planning single-rack GPU colocation
Start with the exact server configuration, workload, rack power budget and cooling method. Then compare facility readiness and the scope of operations: hardware ownership, installation, monitoring and support. Capacity, location and delivery dates need site-specific confirmation.
NVIDIA's data centre planning guide explains how power, cooling and layout affect deployment density. Compare powered shell and managed capacity.
From Grid to Token, One Owner
Every other provider owns a slice. Greco owns the whole column.
Behind-the-meter power
On-site generation + storage energize the campus years before a grid connection would.
Direct-to-chip liquid cooling
The liquid-cooled IR7000 racks are engineered into the energy plant, not bolted on after.
NVIDIA-accelerated compute
Dell PowerEdge XE-series with NVIDIA HGX B300 — racked, networked and ready to run.
Heat exported, not vented
The heat those GPUs produce is captured at 50 °C and sold into district energy.
De-risked in the twin before it’s racked
We model the rack layout, power distribution and cooling for these exact systems in a physics-based digital twin before anything is installed — so density, thermal behaviour and floor loading are validated at the design stage, not discovered on site.
Compare GPU colocation facilities on the same requirements
For a single AI rack or a larger cluster, compare the exact hall and service scope. A facility's total capacity or cooling description is not a commitment for your deployment.
Location and delivery state
Identify the deployment location, access requirements and commissioning date. Distinguish available capacity from a planned build or expansion.
Minimum commitment and rack power
Compare minimum rack or power commitments, contracted usable kW, redundancy, metering and expansion rights. Use the same term and power assumptions.
Cooling responsibilities
Document who supplies and maintains the CDU, manifolds, facility water and heat rejection. Validate operating conditions against the selected server design.
Network and remote hands
Confirm carriers, cross-connects, bandwidth, response windows and remote-hands scope. Separate included service from metered or additional charges.
Choose the hardware and facility together: Dell PowerEdge · Lenovo ThinkSystem · Powered shell planning.
Technology partners
GRECO specifies, procures, integrates and manages infrastructure from the vendors below. Hardware is chosen for the workload and for the power and cooling envelope of the site, not the other way round.
Servers and AI systems
The full PowerEdge line, from rack-scale AI systems to edge nodes — specified, procured, racked and operated as a managed service.
See the rangeServers and AI infrastructure
ThinkSystem and ThinkAgile infrastructure for AI, HPC, cloud and storage, specified to the workload and the power envelope of the site.
See the range