I/ONX to Mentor NSF Center for Space, High-Performance and Resilient Computing (SHREC) Graduate Students on next-generation approaches to efficient, resilient AI systems

Las Vegas, April 30, 2026 – I/ONX High Performance Compute (HPC), a leading provider of heterogeneous AI systems, today announced a collaborative research initiative with the NSF Center for Space, High-Performance and Resilient Computing (SHREC) at the University of Florida. The effort aims to redefine high-performance AI inference while I/ONX also mentors SHREC graduate students on next-generation approaches to efficient AI systems.

The collaboration brings together I/ONX and SHREC researchers to explore how extreme model compression can transform AI inference. Leveraging the I/ONX Symphony platform, together the teams are developing new methods to decompose and execute compressed neural networks across heterogeneous hardware environments—including GPUs, ASICs, FPGAs, and emerging dataflow-based accelerators.

“Our collaboration with SHREC represents a meaningful step toward rethinking how AI inference is delivered at scale,” said Steven Eliuk, CEO of I/ONX. “By combining advanced compression techniques with Symphony’s orchestration, we’re unlocking new levels of performance, efficiency, and resilience—while helping train the next generation of innovators.”

The research focuses on eliminating system bottlenecks and enabling more scalable AI deployment through ultra-low-precision representations that reduce inter-chip communication without sacrificing performance. Additional areas of study include cross-architecture kernel optimization, performance-per-watt in low-power settings, and system resilience under real-world conditions like hardware degradation and failover.

“The SHREC center is dedicated to advancing resilient, high-performance computing for demanding environments,” said Dr. Lam, University of Florida Site Director, SHREC. “Partnering with I/ONX allows us to push the boundaries of AI inference through innovative compression and system design.”

Ultimately, the collaboration aims to establish a unified framework for deploying compressed AI models that improve throughput, energy efficiency, and reliability—positioning both organizations at the forefront of next-generation AI systems.

About I/ONX:

I/ONX High Performance Compute (HPC) is the pioneer of heterogeneous AI infrastructure. Headquartered in Las Vegas, Nevada, I/ONX is redefining the AI lifecycle by eliminating the "Host Tax" of legacy architectures. The I/ONX flagship Symphony SixtyFour platform consolidates up to 64 accelerators into a single node, reducing rack-scale TCO by 50% or more. By dramatically lowering power consumption and maximizing hardware utilization, I/ONX enables enterprises to achieve production-scale AI with unprecedented efficiency and faster ROI.