Why Compute and Data Orchestration Can No Longer Be Solved Separately

Defense agencies, national labs, and research orgs now run workloads across on-prem HPC, hyperscalers, and GPU neo clouds — but the practitioner experience hasn’t kept up. Research shows the cost: 55.4% of government tech orgs spend 26–50% of their app-dev budget on legacy technical debt, 56% cite missing shared security context across on-prem and cloud as the top barrier to a unified posture, and 48% name compliance documentation as developers’ biggest cognitive load.

In this fireside chat, analyst Paul Nashawaty joins David Flynn of Hammerspace and Matthew Shaxted of Parallel Works to unpack managing compute and data across distributed environments. They’ll share lessons from real deployments — including a defense research project that onboarded 100 users to Kubernetes and Slurm in three days and hit 95% cluster utilization in two weeks — and how to unify heterogeneous infrastructure without giving up security or control.

What you’ll learn:

• Why the shift from building AI to running AI has made unified orchestration a priority
• Why compute and data orchestration must be solved together, and what breaks when they aren’t
• How to build security and compliance guardrails into the platform instead of bolting them on
• What makes defense and government environments uniquely demanding, and how those lessons carry over to commercial and regulated industries
• How to turn neo cloud and HPC investments into fully utilized resources, fast
• Where most organizations really sit on the maturity curve, and a practical first step toward a single plane of orchestration