CoreWeave and Parallel Works deploy managed AI/HPC for DARPA NODES
CoreWeave, Inc. (Nasdaq: CRWV) and Parallel Works, Inc. announced September 9, 2026, deployment of a managed AI and high-performance-computing platform for the Defense Advanced Research Projects Agency Network of Optimal Dynamic Energy Signatures (NODES) program, which studies dynamics of biological systems. Parallel Works’ ACTIVATE control plane runs on CoreWeave’s AI cloud, with a dedicated reservation on NVIDIA HGX H100 systems, enterprise storage, and NVIDIA Quantum InfiniBand networking; researchers get single sign-on access to CoreWeave’s Slurm-on-Kubernetes (SUNK) offering and CoreWeave Kubernetes Service under ACTIVATE allocation policies. The release frames the stack as end-to-end operated by Parallel Works (onboarding, identity, provisioning, scheduling, reporting, 24/7 support) with CoreWeave observability on the underlying fleet. No contract value, total GPU count, megawatts, term, or campus locations were disclosed. No accompanying Form 8-K appeared in EDGAR for this item in this check.
Why it matters
It is a named federal research workload on CoreWeave capacity via a managed partner—useful customer-mix color after yesterday’s Goldman Communacopia fireside, but without disclosed economics.
Companies
- CoreWeave
- Parallel Works
Risks
- No disclosed MW, GPU count, contract value, term, or revenue contribution.
- Federal program work can be intermittent and is not equivalent to multi-year commercial offtake.
- No Form 8-K; newsroom release is not an SEC operating metric.