Juno Innovations
Juno Innovations provides Orion, a unified compute plane that orchestrates GPUs, VMs, containers, and bare metal under a single API, enabling 60-second provisioning and typically 2–4× more capacity from existing hardware. The platform eliminates the need for separate infrastructure tools, allowing teams to run diverse workloads—from VFX rendering to research batch jobs—without cloud dependency or Kubernetes expertise.
- Artificial Intelligence
- Defense Tech
- Developer Tools
- Enterprise Software
- Software Only
Funding
Founders
Product
Problem
Organizations running GPU and CPU workloads face a trade-off between hyperscaler GPU rental, which is fast but expensive and raises data sovereignty concerns, and DIY Kubernetes and VMware setups that require significant engineering overhead and often result in hardware utilization as low as 10–15%. This fragmentation forces teams to manage multiple systems, billing structures, and failure points, diverting focus from actual work.
Solution
Orion provides a unified compute plane that orchestrates GPU workloads, containers, VMs, and bare metal from a single API, eliminating the need for multiple infrastructure tools. The platform provisions resources in 60 seconds without IT involvement, supports per-request autoscaling to eliminate idle spend, and works with existing Kubernetes clusters without requiring changes to workload definitions. Orion's custom render operator treats rendering as a containerized, Kubernetes-driven workload, enabling GPU slicing to maximize performance from each instance. The platform supports both cloud and on-premises deployments from one system, with workload templates that let users request environments without writing YAML or submitting tickets.
Target Audience
Primary customers are VFX studios, research labs, and enterprise IT teams that run mixed GPU/CPU workloads and need to scale across cloud and on-premises environments without infrastructure complexity.
Features
- Unified API for GPUs, VMs, containers, and bare metal with 60-second provisioning
- Per-request autoscaling that eliminates idle spend and stair-step capacity planning
- KubeVirt integration for running Windows VM workloads alongside Linux containers
- GPU time slicing to consolidate multiple jobs on a single machine
- Workload templates that define environments once, enabling self-service access without Kubernetes expertise
- Customer-hosted deployment option for data sovereignty, with no data leaving the environment
- 200+ integrations with industry-standard tools including Foundry Nuke, Blender, and AWS Thinkbox Deadline
- Spot Instance support for aggressive scaling and cost optimization