Vast Data, an AI operating system and storage provider, has launched a new multi-cloud and hybrid AI orchestration offering.

Dubbed Polaris, the platform enables users to provision, operate, and orchestrate distributed AI infrastructure across the public cloud, neocloud platforms, and on-premises data centers. It can be used for training and inference workloads.

Training, inference, and data collection are not always conducted within the same geography or platform, and Polaris enables this distributed infrastructure to act under a single operational platform.

“AI infrastructure has outgrown the idea of a single deployment in a single location,” said Jonsi Stefansson, general manager, cloud solutions at Vast Data. “Polaris establishes a global control plane that makes distributed AI infrastructure operationally coherent. It allows enterprises to deploy, scale, and govern Vast clusters anywhere, all while managing them as one system.”

Polaris builds on the existing Vast offering known as DataSpace. Vast DataSpace enables Vast to connect clusters across significant distances – having previously demonstrated the capability as far as 10,000km (6,213 miles), linking one in the US with another in Japan. In addition, the US cluster was based on TPUs, and the Japan cluster on GPUs.

Google Cloud has partnered with Vast for access to DataSpace previously, enabling users to connect clusters on Google Cloud and on-premises.

According to Vast, DataSpace “unifies data across clusters through a global namespace and distributed fabric,” while Polaris “governs how those clusters are deployed and lifecycle-managed across cloud and hybrid environments.”

Organizations will be able, with Polaris, to move AI workloads from isolated deployments to “continuously operating systems,” and to align placement easily with GPU availability, cost, and compliance requirements.

Vast Data is best known for its AI Operating System, which it describes as a platform that manages storage, database, and compute runtime, removing the need to “stitch together dozens of tools.” Several major neoclouds are known adopters of the platform, including but not limited to CoreWeave, Nscale, Buzz HPC, Fluidstack, GMI Cloud, Akamai Technologies, and Crusoe.

In 2025, the company was reportedly seeking funding at a $30bn valuation.