Skip to main content

Neocloud

Neoclouds are companies that purchase GPU hardware, host it in a datacenter, and rent the capacity to multiple tenants. In the NVIDIA ecosystem, these operators are often referred to as NVIDIA Cloud Partners (NCPs).

Neoclouds invest significant capital to acquire GPU infrastructure and therefore focus heavily on maximizing the return on that investment. Financial performance is often evaluated using metrics such as internal rate of return (IRR) and multiple on invested capital (MOIC).

To maximize returns, a Neocloud must achieve:

  • Maximum GPU utilization – the percentage of GPUs rented out to customers
  • Maximum average selling price – revenue generated per GPU
  • Maximum uptime – ensuring GPU infrastructure remains operational
  • Minimum operational expenses (OPEX) – reducing the cost of operating GPU infrastructure

Problem​

Neoclouds prefer to rent GPU capacity to long-term reserved customers under multi-year contracts. Reserved instances provide predictable revenue and allow the operator to raise debt financing at favorable interest rates, which can then be used to acquire additional GPU hardware.

However, not all capacity can be sold through long-term contracts. The challenge for Neocloud operators is how to monetize unused GPU capacity.

Common strategies include:

  • Providing on-demand elastic IaaS services such as bare-metal, virtual machines, or containers
  • Offering managed Kubernetes or other PaaS platforms
  • Delivering AI platforms such as managed SLURM, Kubeflow, Jupyter notebooks, developer GPU environments, or MLOps frameworks
  • Offering LLM-as-a-Service or other model serving platforms
  • Deploying and sharing a single multi-tenant LLM service across multiple users
  • Registering excess capacity on GPU marketplaces
  • Combining several of the above approaches

Each of these strategies requires complex infrastructure automation, multi-tenancy, and operational tooling.

Solution​

Armada Bridge is a GPU management platform that enables operators to deliver GPU infrastructure as Infrastructure-as-a-Service (IaaS), Platform-as-a-Service (PaaS), and AI-as-a-Service (AIaaS). Bridge provides secure multi-tenancy, lifecycle management, and operational automation across GPU clusters, allowing organizations to safely share expensive GPU infrastructure among multiple tenants while maximizing utilization and simplifying operations.

Bridge enables Neoclouds to rapidly transform GPU infrastructure into a monetizable cloud platform.

Bridge allows operators to offer multiple service models simultaneously:

  • GPU Infrastructure-as-a-Service
  • Managed Kubernetes and platform services
  • AI developer platforms and MLOps environments
  • Model-as-a-Service or inference platforms
  • Shared LLM services across tenants
  • Integration with GPU marketplaces

Bridge provides secure multi-tenancy across CPU, GPU, network, Infiniband, NVLink, storage, and WAN resources, ensuring tenants can safely share infrastructure without impacting each other's performance or security.

Operational capabilities include:

  • Observability and monitoring
  • Alerts and fault management
  • Centralized logging
  • Lifecycle management for clusters and software

Bridge also offers flexible commercial models including GPU/hour pricing, ensuring operators only pay for the platform when GPUs are actively generating revenue.

Neoclouds can deploy Bridge within weeks. Armada has extensive experience working with topologies deployed by NVIDIA Infrastructure Specialists (NVIS) and supporting both greenfield and brownfield GPU environments.

Last updated on