PricingContact us

Compare

Fuzzball vs. NVIDIA Run:ai

Keep your orchestration layer independent

Run:ai is a GPU optimization layer that extends Kubernetes, centered on getting more utilization out of hardware you already own, and NVIDIA owns it. CIQ sells no accelerators. Fuzzball owns the whole workflow, from build through train, validate, and serve, executes it close to the hardware through its own runtime, and schedules NVIDIA and AMD from one control plane.

Get the Fuzzball solution brief

The short answer

NVIDIA Run:ai is a GPU scheduling and optimization layer that extends Kubernetes, best known for fractional GPU sharing. NVIDIA acquired it in 2024 and open-sourced its scheduling core as the KAI Scheduler. Fuzzball is built by CIQ, a company that sells no accelerators. It runs the full AI lifecycle, executes on-node without the Kubernetes pod model in the path, and schedules NVIDIA and AMD hardware from one control plane.

From the team that created Apptainer and Singularity at Lawrence Berkeley National Laboratory, and the founding sponsor of Rocky Linux.

Compare what each one is built for

NVIDIA Run:ai

  • Fractional GPU sharing at the CUDA level, a genuinely sophisticated capability
  • Quota, fairshare, and gang scheduling across a shared Kubernetes GPU cluster
  • Interactive Jupyter, batch training, and autoscaling inference in a familiar Kubernetes ecosystem
  • An optimization layer on a stack you still assemble and operate, sold by NVIDIA

Fuzzball

  • Build, train, validate, serve, and iterate in one portable workflow definition
  • Its own per-node runtime executing directly on the hardware, with gang scheduling built in
  • NVIDIA and AMD ROCm scheduled from one control plane, on any major cloud and on-prem
  • Layers on existing Slurm and PBS, and on the Kubernetes you already run

Give Run:ai credit where it is due

GPU efficiency is a real problem worth solving, and Run:ai solves it well. Fractional GPU sharing, which time-slices a physical GPU at the CUDA level so several workloads share it, meaningfully raises throughput on hardware you already own. The AI-aware scheduler handles multi-tenant quotas, priorities, fairshare over time, and gang-scheduled distributed training. If the goal is squeezing more utilization out of an NVIDIA GPU cluster running on Kubernetes, that is a legitimate reason to choose it, and CIQ says so plainly.

The distinction worth weighing is what each product is at its core: an optimization layer that extends Kubernetes, or a platform that owns execution from the hardware up.

Prove the neutrality, do not claim it

CIQ sells no accelerators. There is no GPU line, no hardware revenue, and no incentive anywhere in the business for Fuzzball to schedule one vendor’s silicon better than another’s. That is a fact about how CIQ makes money, and it does not change with a roadmap decision. Every claim below follows from it.

Two accelerator ecosystems, one control plane

AMD ROCm support shipped in Fuzzball 4.2, alongside existing NVIDIA support. Both are discovered and scheduled by the same platform, and neither is a bolt-on for the other.

Five clouds and your own racks, one workflow definition

Fuzzball deploys to AWS, Google Cloud, Oracle Cloud, Azure, and CoreWeave, and to on-premises clusters built with Warewulf, VMware, or bare metal. The same workflow definition runs unchanged in all of them, so the platform never becomes the reason a hardware or cloud decision is hard to reverse.

The scheduler you already run keeps running

Fuzzball’s provisioners integrate with existing Slurm and PBS, and Orchestrate deploys on any conformant Kubernetes distribution. Adoption does not require displacing what already works.

A track record of giving control away

CIQ’s founder created Singularity at Lawrence Berkeley National Laboratory and moved it to the Linux Foundation as Apptainer, vendor-neutral and community-governed. Rocky Linux is stewarded by the Rocky Enterprise Software Foundation rather than by CIQ. This is a team that has repeatedly handed its foundational projects to neutral bodies, which is behavior rather than positioning.

Neutral does not mean generic

Independence is not the same as lowest-common-denominator support. InfiniBand is auto-detected, GPUs are allocated with topology awareness, and MPI, Shmem, and NCCL run unchanged inside containers. Fuzzball gets close to whatever hardware it is placed on, and it stays uncommitted about whose hardware that is.

Weigh the differences that matter

Two of the big schedulers now sit inside a hardware vendor

NVIDIA acquired Run:ai in 2024 and SchedMD, Slurm’s maintainer, in December 2025. Both products remain open source, and both pledges are credible. The pattern is still worth naming, because two of the most widely used ways to schedule accelerated work now belong to the company that sells the accelerators. That is not an accusation. It is a concentration of dependency worth pricing into a platform decision.

Execution close to the hardware

Run:ai extends Kubernetes, so the unit of work is a pod and the platform’s job is to schedule and share GPUs among pods. That is a strong model for its purpose, and it is a layer above the container runtime. Fuzzball’s Substrate agent executes directly on the compute node, with direct access to GPUs, high-speed interconnects, and parallel filesystems. MPI, Shmem, and NCCL run unchanged, InfiniBand is auto-detected, and GPUs are allocated with topology awareness.

A platform in place of a scheduler

Run:ai raises utilization on a Kubernetes cluster you stood up and continue to operate. The scheduler is one sophisticated piece of a stack you still assemble. Fuzzball ships the workflow engine, data movement, container preparation, autoscaling inference, federation, and a browser interface as one system, with a web UI, CLI, and SDK at full parity.

The full lifecycle in one definition

Training, fine-tuning, validation, and inference are steps in the same Fuzzball workflow through Service Endpoints, at the same close-to-hardware performance. Model weights arrive from Hugging Face or Ollama as declared inputs, and data ingress, caching, and egress are declared as volume steps rather than scripted around the job.

Sovereign AI on infrastructure you govern

The whole lifecycle runs on hardware the organization owns, on premises or in any supported cloud, with the same workflow definition in both places and every action auditable. Train on cloud GPUs and keep sensitive inference in house, using one workflow either way.

See what Fuzzball 4.2 adds

Schedule NVIDIA and AMD from one control plane

AMD ROCm support shipped in 4.2, so Fuzzball discovers and schedules workloads across both NVIDIA and AMD infrastructure. This is the concrete version of the neutrality argument: not a roadmap promise, a shipped capability in the current release.

Give AI agents a safe way to run the workload

A new MCP server lets an AI agent inspect a Fuzzball environment and draft, submit, and monitor workflows, with explicit permission required for writes, execution, and destructive actions. Agent loops and pipelines drive real cluster work inside boundaries an operator sets.

Let a workflow submit its own work

Every job and service container receives a workflow-scoped API credential and the cluster’s API address automatically, so a running job can submit, track, and stop additional workflows with no separate login and no long-lived shared secrets.

Schedule around unhealthy hardware

Fuzzball scores every node’s reliability from host and GPU health signals and factors that score into placement. Jobs restart elsewhere when a fault is detected, and degraded hardware is cordoned, drained, and replaced automatically. Utilization only counts when the work on the GPU finishes.

Account for what each workload consumed

Per-workflow accounting covers compute, storage, and network egress, with organization-specific pricing adjustments, alongside organization-level storage isolation and compute-policy grants. Utilization tells you the GPUs were busy. Accounting tells you what the training run cost and who owes for it.

Compare Fuzzball and Run:ai at a glance

NVIDIA Run:ai

Fuzzball

Design center
CategoryGPU optimization layer built on KubernetesPurpose-built, vendor-neutral HPC and AI platform
Core valueMaximize GPU utilization through fractioning and quotasRun the whole AI workflow close to the hardware
ArchitectureScheduler extending Kubernetes, unit of work is a podOn-node runtime executing directly on the compute node
Workload unitKubernetes workloads and jobs on a shared GPU clusterA workflow: batch job, inference service, or both
Scheduling and hardware
GPU sharingFractional GPU and CUDA time-slicing, a genuine strengthHardware-aware scheduling with bare-metal execution
Interconnect awarenessGPU-focused, through the Kubernetes device modelInfiniBand auto-detected, GPUs allocated with topology awareness
Node healthGPU utilization visibility and policy enforcementReliability scoring feeds placement, jobs restart on healthy nodes
Accelerator supportNVIDIA, as part of the NVIDIA AI Enterprise stackNVIDIA and AMD ROCm from one control plane
Lifecycle and data
Lifecycle coverageJupyter, training, and autoscaling inference on KubernetesTrain, validate, and serve in one workflow, with RAG and agents
Data managementHandled by the surrounding stack and the applicationIngress, caching, and egress declared as workflow volumes
Model ingressHandled by the ML tooling you assemblePull from Hugging Face or Ollama as a declared input
AccountingGPU utilization visibility and cost controlCompute, storage, and network egress per workflow
Independence and operation
OwnershipAn NVIDIA product, with the KAI Scheduler core open sourcedBuilt by CIQ, which sells no accelerators
Hardware incentiveThe vendor of the platform also sells the siliconNo GPU line and no hardware revenue anywhere in the business
Agent interfaceKubernetes and ML ecosystem toolingMCP server for workflows, with per-action permissions
You still operateYour Kubernetes cluster and the surrounding stackOne platform, and it layers on Slurm and PBS if present
Relationship
To each otherRequires a Kubernetes GPU cluster to install ontoOrchestrate deploys on Kubernetes, Substrate runs on-node

Run Fuzzball alongside Kubernetes

The comparison is not always mutually exclusive.

  1. On your Kubernetes

    Orchestrate is a Kubernetes-native microservice stack, so it deploys on the conformant cluster you already operate.

  2. On your nodes

    Substrate executes jobs directly on the compute node, which keeps bare-metal performance for both training and serving.

  3. On your schedulers

    Fuzzball coexists with existing Slurm and PBS deployments and with parallel filesystems, and adds orchestration without a migration.

Match this to your situation

  • The orchestration layer comes from the same vendor that sells the accelerators
  • AMD or custom silicon is on the roadmap, and the platform would make switching costly
  • A Kubernetes cluster has to exist and be operated before the AI platform can start
  • Utilization looks healthy, and training runs still fail on bad nodes and restart from scratch
  • Training runs on one stack and inference on another, with a manual handoff between them
  • A sovereignty or data-residency requirement needs an answer a vendor-owned layer does not give

Zero

Accelerators CIQ sells

No GPU line and no hardware revenue, so nothing in the business favors one vendor’s silicon.

Full

Lifecycle in one portable workflow

Autoscaling inference, RAG, and agents defined alongside training rather than bolted on.

Sovereign

Models and data stay on infrastructure you control

Serve on-prem, or train on cloud GPUs and keep sensitive inference in house.

Per workflow

Compute, storage, and egress accounted for

Utilization says the GPUs were busy. Accounting says what the run cost.

Define the terms

What is NVIDIA Run:ai?
NVIDIA Run:ai is a GPU orchestration and optimization platform that extends Kubernetes with an AI-aware scheduler. It is best known for fractional GPU sharing, which time-slices a physical GPU at the CUDA level so several workloads can share it, alongside quota, fairshare, and gang scheduling. NVIDIA acquired Run:ai in 2024, and it is part of the NVIDIA AI Enterprise stack.
What is the KAI Scheduler?
The KAI Scheduler is the open-source scheduling core of NVIDIA Run:ai. NVIDIA released it under the Apache 2.0 license in April 2025, and it is now a CNCF Sandbox project. It is a Kubernetes-native scheduler for GPU and CPU AI workloads, and it continues to ship as part of the commercial Run:ai platform.
What is fractional GPU sharing?
Fractional GPU sharing is a technique that divides a single physical GPU among several workloads, so a job that cannot use a whole accelerator does not hold one idle. It raises utilization on existing hardware, and it is distinct from scheduling a tightly coupled job that needs whole GPUs allocated together with interconnect topology in mind.
What is Fuzzball?
Fuzzball is CIQ’s orchestration platform for HPC, AI training, and AI inference. It treats every workload as a containerized workflow, executes it through an on-node agent with direct access to GPUs and high-speed interconnects, and schedules NVIDIA and AMD ROCm hardware from one control plane across on-premises clusters, cloud, and edge.

Run Fuzzball on your own workflow

Bring one workflow and the environments you have to run it in. A CIQ engineer will walk your team through a run on your own hardware, then the same run in a cloud, unchanged.

Get a demo

Frequently asked questions

Yes. Run:ai is a strong GPU optimization layer, and GPU efficiency is a real problem worth solving. Fractional GPU sharing and AI-aware quota scheduling meaningfully raise throughput on hardware you already own. The distinction worth weighing is what each product is at its core: an optimization layer that extends Kubernetes, or a platform that owns execution from the hardware up.

Partly. NVIDIA released Run:ai’s scheduling core as the KAI Scheduler under the Apache 2.0 license in April 2025, and it is now a CNCF Sandbox project. The commercial Run:ai platform, including its management interface, policy controls, and fractional GPU capabilities, remains an NVIDIA product within the NVIDIA AI Enterprise stack.

Fractional sharing at the CUDA level is Run:ai’s signature capability. Fuzzball’s focus is different: hardware-aware scheduling with direct GPU and interconnect access, gang scheduling for tightly coupled training, and bare-metal execution for both training and inference. Teams choosing between them should weigh which problem is actually costing them more.

Yes. AMD ROCm support shipped in Fuzzball 4.2, so Fuzzball discovers and schedules workloads across both NVIDIA and AMD infrastructure from one control plane. That keeps the platform independent of any single accelerator supplier.

Fuzzball Orchestrate is a Kubernetes-native microservice stack and deploys on any conformant Kubernetes cluster. Substrate, the compute agent, runs directly on the compute node rather than as a Kubernetes-scheduled pod, which is what keeps the job execution path close to the hardware.

It is structural. CIQ sells no accelerators, so there is no GPU line and no hardware revenue anywhere in the business that would favor one vendor’s silicon. Fuzzball schedules NVIDIA and AMD ROCm from one control plane, deploys to AWS, Google Cloud, Oracle Cloud, Azure, CoreWeave, and on-premises hardware with the same workflow definition, and layers on existing Slurm, PBS, and Kubernetes rather than replacing them.

Leverage. NVIDIA acquired Run:ai in 2024 and SchedMD, Slurm’s maintainer, in December 2025, so two of the most widely used ways to schedule accelerated work now sit inside the company that sells the accelerators. Both remain open source and both pledges are credible. A neutral orchestration layer still means the platform never becomes the reason a hardware decision is hard to reverse.

Both. Service Endpoints define persistent, scalable inference and interactive endpoints alongside batch training steps in one portable workflow, and they preserve the same close-to-hardware performance while scaling with demand as part of the orchestrated job.

Fuzzball deploys to AWS, Google Cloud, Oracle Cloud, Azure, and CoreWeave, to on-premises clusters built with Warewulf, VMware, or bare metal, and onto existing Kubernetes distributions. The same workflow definition runs unchanged in all of them, and Fuzzball Federate unifies them and routes work on cost, performance, and data locality.

Still have questions?