Run:ai

Run:ai was founded in 2018 by Omri Shapira and Ran Lahav. The company builds a GPU resource orchestration and scheduling platform for AI and machine learning workloads. In 2024, NVIDIA acquired Run:ai.

Run:ai's primary product is a platform for managing GPU infrastructure in Kubernetes environments, enabling AI research and development teams to share and efficiently utilize organizational GPU clusters. The platform provides dynamic resource scheduling so that GPU resources are not left idle while other teams are waiting. Typical buyers are large enterprises with on-premise GPU clusters — technology companies, research institutions, and financial services firms — managed by MLOps and AI infrastructure teams.

The core product, the Run:ai Platform, is built on Kubernetes as its base layer. It includes a dedicated AI workload scheduler that replaces Kubernetes' standard scheduler, and supports fractional GPU allocation, priority queuing, and resource sharing across projects. The platform supports on-premise, cloud, and hybrid environments.

Run:ai's direct competitors include open-source tools such as MLflow and Kubeflow, as well as Determined AI (acquired by HPE). NVIDIA, which acquired Run:ai, positions the platform as a management layer that complements its GPU hardware. The 2024 acquisition by NVIDIA significantly strengthens Run:ai's position as part of NVIDIA's broader AI stack.

Run:ai was founded in Israel, with its core R&D center located in Tel Aviv. Both founders are Israeli, and the majority of R&D activity has been based in Israel. The company also maintained a US presence for sales and marketing functions.