Academic & Institutional GPU & AI Infrastructure

Dedicated Canadian GPU infrastructure for AI, machine learning, simulation, visualization and GPU-accelerated research.

myresearchcloud.ca provides sovereign GPU infrastructure for laboratories, departments, colleges, universities, hospitals, research institutes and grant-funded projects.

Reserve a dedicated GPU for a sustained research workload, establish a multi-GPU allocation for a laboratory or institution, or schedule high-performance H100 capacity for intensive AI and computational projects.

Paid GPU resources use dedicated physical GPU allocation with dedicated CPU and memory resources. There is no GPU time-slicing, no CPU overcommitment, no burst-credit throttling and no data-egress billing.

GPU Infrastructure Matched to the Workload

GPU research requirements vary considerably. Some projects require persistent GPU capacity for months or years, while large training and computational workloads need much more powerful infrastructure for a shorter period.

Persistent GPU Pricing

Persistent GPU allocations include the GPU, dedicated CPU resources, RAM and replicated enterprise SSD storage in one predictable monthly price.

All prices are in Canadian dollars. Persistent GPU allocations use dedicated physical GPUs and are subject to capacity availability. Additional CPU, RAM and storage can be added to match workload requirements.

Institutional GPU Capacity Blocks

Institutional GPU Capacity Blocks allow laboratories, research programs and institutions to reserve multiple dedicated GPUs under a single agreement.

NVIDIA A2 Institutional Capacity

Designed for development, inference, model experimentation, GPU-accelerated applications and other sustained workloads that do not require large GPU memory.

NVIDIA L4 Institutional Capacity

Designed for larger inference workloads, model development, visualization, image processing, accelerated research applications and moderate AI workloads.

Need a larger persistent GPU environment? Larger allocations and mixed CPU/GPU research environments can be designed under an institutional capacity agreement.

Multi-GPU allocations are subject to GPU availability and may require advance capacity planning.

High-Performance H100 Project Capacity

H100 infrastructure is intended for large AI training, fine-tuning, inference and other computationally intensive GPU workloads.

Rather than requiring a long-term reservation, H100 resources are provided as scheduled one-week project blocks.

H100 allocations have a minimum of four GPUs and are scheduled subject to availability.

Additional one-week allocations may be requested. Larger GPU configurations and multi-node requirements can be discussed as part of an institutional or project-specific infrastructure plan.

H100 capacity is not offered as ad hoc hourly GPU infrastructure. The hourly figure above is provided only as an equivalent unit cost for budget comparison.

Dedicated GPU Allocation

Paid GPU resources are designed for sustained research performance rather than transient retail cloud workloads.

• 1:1 physical GPU assignment
• No GPU time-slicing
• Dedicated CPU allocation
• No CPU overcommitment
• Dedicated RAM allocation
• Persistent research environments
• No burst-credit throttling
• Predictable resource availability
• Canadian data residency
• No data-egress charges

Designed for GPU-Accelerated Research

GPU infrastructure can support workloads including:

• AI model development and training
• Large language model fine-tuning
• AI and machine-learning inference
• Computer vision and imaging analytics
• Medical and scientific imaging
• GPU-accelerated simulation and modelling
• Molecular modelling and computational chemistry
• Visualization and rendering
• CUDA-accelerated research applications
• Research software development
• Graduate research environments
• Research AI laboratory environments

Additional Research Storage

Additional research storage can be attached to GPU environments, training datasets and institutional capacity allocations.

Large storage allocations, including 100 TB and petabyte-scale environments, can be quoted separately based on capacity, performance and retention requirements.

Research Infrastructure Support

Platform support is included with all paid GPU infrastructure. Laboratories and institutions requiring operating-system, GPU software, researcher onboarding or ongoing technical assistance can add a Research or Institutional Support plan.

Included technical-assistance hours are monthly allocations and do not accumulate from month to month. Target response times are service objectives and are not contractual service-level guarantees unless specifically included in an agreement.

What Platform Support Includes

Platform Support covers infrastructure operated by myresearchcloud.ca, including:

• Physical GPU and compute infrastructure
• Hypervisor and virtualization platform
• Underlying storage infrastructure
• Platform networking
• GPU and VM provisioning failures
• Platform availability incidents
• Account and billing assistance

Platform Support does not include administration of the researcher's guest operating system, research software, AI models, application code or research workflows.

For short-duration H100 allocations, technical and engineering assistance can be purchased separately on an hourly or statement-of-work basis.

Professional Services

Complex engineering and project-based GPU, AI and research infrastructure work is available separately from monthly support plans.

Larger projects can be delivered using fixed-price or statement-of-work arrangements where appropriate.

Designed for Grants and Institutional Procurement

GPU requirements can be difficult to budget when infrastructure is priced entirely around variable hourly consumption.

myresearchcloud.ca uses reserved allocations so researchers and institutions can establish infrastructure costs before a project begins.

We can provide budgetary estimates for grant applications, fixed-term GPU allocations for funded projects, and shared GPU capacity for laboratories, departments and institutions.

CPU compute, GPU capacity, research storage and technical support can be combined under a single research infrastructure proposal.

How Institutional GPU Capacity Works

• Physical GPUs are reserved for the laboratory or institution for the contracted term.
• Each GPU remains dedicated rather than being time-sliced among unrelated customers.
• GPUs can be assigned among multiple approved researchers and projects within an institutional allocation.
• Institutional A2 and L4 allocations receive volume pricing in addition to commitment-term pricing.
• GPU assignments can be changed as research requirements evolve.
• Additional CPU and research storage can be added to GPU environments as required.
• H100 and other high-performance resources are scheduled separately according to workload requirements and infrastructure availability.

Pricing & Service Notes

• All prices are in Canadian dollars.
• Applicable taxes are additional.
• Persistent A2 and L4 pricing requires the corresponding one-year or three-year commitment.
• Institutional GPU Capacity Block pricing represents reserved physical GPU capacity.
• Published persistent GPU pricing includes the CPU, memory and SSD resources listed with each service.
• Additional research storage is billed at the published per-TB monthly rate.
• H100 pricing is based on scheduled one-week project allocations.
• There are no data-egress charges.
• There are no GPU time-slicing charges or burst-credit models.
• CPU resources included with paid GPU services are not oversubscribed.
• GPU resources and larger allocations remain subject to infrastructure availability and capacity scheduling.