Vendor directory

GPU Cloud Providers for AI Infrastructure

GPU cloud providers rent graphics processors and the networking and storage around them for training and running AI models. These vendors offer on-demand instances, reserved capacity, dedicated clusters or managed inference services.

10 vendors Listed A to Z Free listings, no ratings

Writing a GPU cloud RFP?

Start from a free, editable template, then send it to the vendors on your shortlist so every proposal answers the same requirements.

GPU cloud vendors, A to Z

CoreWeave runs a cloud built for AI that rents NVIDIA GPU compute, including bare metal, along with Kubernetes, storage and networking services. Its Forge product set adds tools for model training, inference and AI agents.

coreweave.com ↗Profile →

Crusoe builds AI infrastructure and runs Crusoe Cloud, which rents NVIDIA and AMD GPUs for model training and inference. It also offers managed inference and fine-tuning, and describes an energy-first approach to its infrastructure.

crusoe.ai ↗Profile →

DigitalOcean sells cloud infrastructure, including compute, managed databases, storage and networking. Its GPU Droplets provide AMD and NVIDIA GPUs for AI, machine learning and inference workloads, with serverless inference available on its AI platform.

digitalocean.com ↗Profile →

Lambda rents NVIDIA GPU capacity for AI training and inference, from on-demand instances to reserved clusters and dedicated supercomputers. It also provides orchestration tools and the Lambda Stack software bundle for machine learning teams.

lambda.com ↗Profile →

Nebius runs an AI cloud that provides GPU clusters with InfiniBand networking for model training, along with managed and serverless inference services. It includes MLOps tooling and self-service access for AI developers.

nebius.com ↗Profile →

Oracle offers AI infrastructure on Oracle Cloud Infrastructure, with NVIDIA and AMD GPU instances and OCI Supercluster deployments for model training and inference. Oracle also sells databases, enterprise applications and other cloud services.

oracle.com ↗Profile →

Runpod offers a cloud for AI developers with on-demand GPU pods, serverless GPU endpoints and multi-node GPU clusters. Developers use it for training, fine-tuning, inference and batch jobs, with a hub of open source model templates.

runpod.io ↗Profile →

TensorWave runs an AI cloud built on AMD Instinct accelerators, offering dedicated bare metal servers for model training, fine-tuning and inference. It pairs the hardware with hands-on engineering support for AI and HPC workloads.

tensorwave.com ↗Profile →

Together AI operates a cloud for AI that offers serverless and batch inference for open source models, fine-tuning and self-serve NVIDIA GPU clusters. It targets developers and companies building generative AI applications.

together.ai ↗Profile →

Vultr runs a public cloud with compute, storage and other infrastructure services. Its Cloud GPU offering provides AMD and NVIDIA GPUs on demand as virtual machines, bare metal or self-service clusters for AI, machine learning and high-performance computing.

vultr.com ↗Profile →

Questions to ask GPU cloud vendors

  • Which GPU generations are available to us now, and in which regions?
  • What capacity can you reserve for us, for how long, and on what terms?
  • How are the cluster network and storage sized for multi-node training?
  • What does the contract say about data handling, security certifications and support?

Related vendor categories

Data platform vendors All vendor categories

How vendors are listed: each listing describes what the vendor's own website says it sells, checked by RFPhub. Vendors are listed A to Z. Listings are free, RFPhub doesn't rank or rate vendors, and nobody pays for placement or order. A listing isn't an endorsement, and RFPhub isn't affiliated with the vendors. Product and company names are trademarks of their owners. Missing a vendor, or found an error? Tell us.