ITS Private Cloud: a modernized platform, ready for what's next
ITS Private Cloud Modernization
The ITS Private Cloud is the University's sovereign, on-campus cloud hosting service for departments across U of T. Over the past several months, the platform has been refreshed end-to-end, with new compute, networking, storage, GPUs, and leadership, to give the U of T community a secure, modern foundation for today's workloads and tomorrow's AI, data, and research needs.
New leadership
Jose Lopez has moved into a new role as Senior Manager, AI Strategy & Platforms, a position created to focus ITS leadership attention on the University's growing AI needs. Sean Despres has stepped into the role of Manager, Servers, Storage and Virtualization.
What's new
A shipment of new infrastructure arrived in Q3 2025 and is now in production. The new blade servers carry 25 per cent more RAM than the previous generation and run on Intel's latest Xeon server processors, with built-in AI acceleration and dedicated accelerators for cryptography, analytics, and data movement.
The platform also now includes 48 NVIDIA L4 Tensor Core GPUs paired with NVIDIA AI Enterprise, a fully supported software stack of NVIDIA-certified frameworks, inference servers, and containers for running AI workloads. This gives the ITS Private Cloud a production-ready environment for AI inference, generative AI, accelerated video, and prototyping. The L4s replace the older P6 GPUs and give teams across U of T, including researchers exploring AI in their work, a place to run small-scale AI testing and workloads on campus. Anyone curious can try the on-premises models at menu.dev.ai.its.utoronto.ca (select "On Premise").
The networking environment also got a full data centre switching refresh, with servers now connected at 40G and 100G links between switches. Two storage clusters also received new disks.
What this means for your workloads
The ITS Private Cloud started as the University's enterprise administrative hosting platform and has grown over the past decade into a shared service used by departments across U of T; a secure alternative to running server hardware in wiring closets and small departmental server rooms. For clients looking to migrate their servers, the result is a more stable IT state with predictable OpEx costs, less local infrastructure to maintain, and a stronger security posture.
What's next
A proof-of-concept is underway to evaluate a new storage platform built on hyperconverged NVMe technology, with the goal of delivering faster, more responsive storage for the workloads we host. A managed container service is also being explored for the longer term, which would give clients a modern deployment option alongside virtual machines.
An AI-ready University needs more than access to models; it needs infrastructure under its own roof. The L4 capacity on the ITS Private Cloud, paired with NVIDIA AI Enterprise, gives U of T a sovereign foundation for rapid prototyping and secure inference. It's a strong step toward making AI a routine capability across teaching, research, and administration”
Jose Lopez, Senior Manager, AI Strategy & Platforms
This work advances several IT@UofT Strategic Plan (2025–2027) priorities, including shared infrastructure, information security, research enablement, and GenAI.
Interested in learning more or moving a workload onto the Private Cloud? Get in touch through utor.cloud/help, or visit the customer wiki for service details and documentation.