Federated multicluster operating system that governs and observes an operator's entire GPU cluster fleet.
Aranya's multicluster operating system sits above the per-cluster engine and federates an operator's clusters so the fleet is addressed and reasoned about as one system. Its published capability set covers plain-language fleet direction, role-based permissions applied at the operating-system level, and observability spanning a single GPU temperature through federation-wide traffic.
The offering is positioned as automating operational work a platform team would otherwise perform by hand, including recovering cost by recycling idle compute across the fleet. It is in private alpha behind a waitlist, is sold as software the customer runs, and its pricing has not been announced; support is described as a shared Slack channel plus managed observability dashboards.
