Saturn Cloud, the AI token factory platform, has entered into a partnership with OpenNebula Systems, the developer of OpenNebula, an open enterprise cloud and virtualisation platform widely used for AI factories and neocloud environments.
The integration combines OpenNebula’s GPU virtualisation, secure multi-tenant orchestration, elastic Kubernetes integration, and bare-metal lifecycle management with Saturn Cloud’s fine-tuning, model serving, distributed training, and managed environment capabilities, creating a unified token factory stack for AI infrastructure operators.
OpenNebula is used by organisations building sovereign AI Factories across neoclouds, HPC centers, telecom operators, and enterprise data centres. Its open, vendor-neutral architecture gives operators full control over their GPU infrastructure, with support for NVIDIA GPUs, NVLink, GPU passthrough, Multi-Instance GPU (MIG) partitioning, BlueField DPU offload, and Spectrum-X Ethernet networking. What operators still need is the application layer, where fine-tuning, model serving, and inference actually happen.
Saturn Cloud fills that gap with its token factory platform. AI teams on OpenNebula-managed infrastructure can fine-tune open models (full-weight or LoRA), deploy them to OpenAI-compatible inference endpoints, and meter usage per token, all within the operator’s branded environment. Training jobs automatically configure multi-GPU DeepSpeed, with Saturn Cloud handling orchestration, retry, and checkpoint output across frameworks including Axolotl, vLLM, Unsloth, TRL, PEFT, and DeepSpeed. The platform also provides managed environments for development and experimentation, as well as distributed training.
“OpenNebula provides the sovereign, vendor-neutral infrastructure layer for building and operating AI environments with full control over GPU resources and governance. While it delivers the foundation for infrastructure and orchestration, organisations still need a consistent application layer to enable AI development. This partnership closes that gap, enabling seamless, end-to-end AI workload execution,” said Ignacio M. Llorente, Managing Director, OpenNebula Systems.
This is especially relevant for organisations that have moved away from proprietary virtualisation stacks and adopted OpenNebula as an open alternative. Many of these operators have solved the infrastructure problem, but still lack a standardised way to deliver AI workloads to their users.
The partnership addresses that directly: OpenNebula handles everything from hardware provisioning and GPU scheduling to tenant isolation and governance, while Saturn Cloud handles the platform layer.
“GPU cloud operators have invested heavily in infrastructure, but their tenants and teams still need a real platform on top of it for fine-tuning, serving, and inference. Saturn Cloud and OpenNebula together turn that infrastructure into a self-service token factory, giving operators a ready-made platform layer and giving engineers the tools to start producing value on day one,” said Sebastian Metti, Founder, Saturn Cloud.
The result is a consistent workflow across sovereign European data centres, neocloud facilities, enterprise on-premises environments and major cloud providers.
