Internet2 Expanded Research Network With New AI PODs

The deployment adds localized modular infrastructure to the national research platform to boost shared AI capacity.

Updated on Sept. 22, 2026 in Data Centers

Internet2 Expanded Research Network With New AI PODs

Live Poll

Do you believe shared national computing infrastructure improves the quality of research in your area?

Internet2 has deployed Cisco AI PODs at the University of Nebraska-Lincoln and Sacramento State to increase GPU capacity for the National Research Platform. This modular infrastructure is designed to bridge computing access gaps for researchers and students across the country.

Why it matters

By deploying pre-validated, standardized infrastructure, the project reduces the engineering barriers for institutions to contribute to a shared, high-performance computing ecosystem. This model allows for the aggregation of resources from over 70 organizations into a unified research fabric.

The new sites add a combined 8 NVIDIA H200 GPUs, 1.1 TB of GPU memory, 128 CPU cores, and 90 TB of NVMe flash storage. Each AI POD provides a maximum connectivity speed of 200 Gbps, integrating into the existing Nautilus Kubernetes-based fabric.

The players

Internet2

A non-profit community providing a high-performance network for research, education, and collaboration.

National Research Platform

A large-scale, distributed research computing fabric that aggregates specialized hardware from over 70 organizations.

Cisco

A networking and data center infrastructure company that provides the AI POD hardware for scalable machine learning workloads.

The details

The Cisco AI PODs serve as pre-validated, modular infrastructure that includes integrated compute, networking, and management software. These systems plug directly into the National Research Platform, a distributed fabric built on Kubernetes—open-source software used to automate the deployment and scaling of containerized applications. This architecture allows researchers to treat geographically dispersed GPU and CPU clusters as a single, cohesive pool of computing power.

Internet2 was founded in 1996.
Internet2 announced the expansion of the National Research Platform on September 22, 2026.

The Tech Race

This deployment marks a significant expansion of the National Research Platform, which competes with centralized hyperscale cloud services by offering a decentralized, community-owned alternative. It follows the project’s established strategy of aggregating specialized hardware to support academic research workloads that require high-speed connectivity.

Researchers and students at the University of Nebraska-Lincoln and Sacramento State gain immediate access to high-performance GPU hardware for standardized AI experimentation. The system is designed to lower the integration workload, meaning researchers can focus on computational tasks rather than configuring the underlying cluster fabric.

The takeaway

The project demonstrates that decentralized hardware pooling is a viable path for academic institutions to scale AI research without relying solely on commercial cloud providers. Observers should track the future integration of additional performance assurance service nodes to see how capacity increases across the network.