GPU cloud
Combine filters →Three ways to put your own model on someone else’s GPU, and each record says which it sells. GPU instances are servers with a card in them, virtual or dedicated, and everything above the driver is yours, including the bill while the machine sits idle. Serverless GPU runs your container only while requests arrive, at the price of a cold start. Managed inference takes the weights, yours or from Hugging Face, and hands back an endpoint. A token API over the provider’s own models does not count here: the model there is theirs. An application that only calls a model belongs under AI apps.
58 providers rent out GPUs to run the customer's own models, listed alphabetically. Across 51 countries.
By regions
Every regions page for GPU cloudAustralia, Brazil, Canada, France, Germany, India, Indonesia, Japan, Netherlands, Poland, Singapore, Spain, Sweden, Switzerland, United Kingdom and United States.
By entry price
Every entry price page for GPU cloudA
Alibaba CloudAlibaba Group's cloud platform, selling virtual machines, containers, serverless functions and managed services.
Amazon Web ServicesAmazon Web Services is the largest cloud infrastructure provider, spanning 123 Availability Zones in 39 geographic Regions.
ArsysSpanish hosting provider offering shared hosting and VPS from ISO-certified datacenters in Logroño, part of the IONOS Group since 2013.
Aruba.itAn Italian provider that owns its data centers and sells registrar services, hosting, servers and a public cloud.
AZDIGIVietnamese host in Ho Chi Minh City, running shared, WordPress and cloud hosting out of two Tier III data centers, with GPU instances alongside.
AzureMicrosoft's hyperscale cloud — hundreds of separately metered infrastructure services, with App Service as the managed application layer above them.
B
BasetenAI inference platform offering token-priced model APIs and dedicated GPU deployments of customers' own models, billed per minute and able to scale to zero.
BeamServerless GPU platform where Python functions and container images run as autoscaling endpoints, task queues and sandboxes, with on-demand GPU machines beside them.
C
CerebriumServerless GPU platform for real-time AI workloads, running Python apps and custom Docker containers that scale to zero and bill per second.
Cherry ServersLithuanian provider selling dedicated machines and virtual servers on its own hardware, billed by the hour, by fixed term or as spot capacity.
CivoBritish Kubernetes-first cloud that charges for worker nodes only, with control planes included and data transfer unmetered.
cloudscale.chSwiss infrastructure-as-a-service provider selling virtual servers, object storage and load balancers on server hardware it owns in Swiss data centers.
ContaboGerman provider selling virtual and dedicated servers from Munich, with large resource allocations at low fixed monthly rates.
CrusoeUS AI infrastructure company renting GPU and CPU virtual machines, managed Kubernetes and model inference from its own data centers.
D
DigitalOceanOne we likeDigitalOcean is a publicly listed cloud provider best known for Droplets, its VPS product, alongside managed databases, Kubernetes and an app platform.
E
E2E NetworksIndian cloud infrastructure provider listed on the National Stock Exchange, selling CPU and GPU compute by the minute with published hourly rates.
ElestioElestio is managed hosting for open-source software on dedicated virtual machines, provisioned into a third-party cloud or into the customer's own account.
ExoscaleExoscale is a Swiss infrastructure provider selling KVM instances, managed Kubernetes, object and block storage and managed databases from European zones.
G
GcoreGcore is an edge and cloud provider selling CDN, streaming, DDoS protection, virtual machines, bare metal, Kubernetes and serverless functions.
GleSYSGleSYS is a Swedish infrastructure provider selling KVM and VMware servers, bare metal, managed databases, object storage and colocation from Nordic data centers.
Google Cloud PlatformGoogle Cloud Platform is one of the three hyperscalers, selling compute, storage, data and machine-learning services on the infrastructure behind Google's own products.
Google Cloud RunGoogle Cloud Run runs any container image, scales it to zero when idle and bills per request and per resource-second.
H
HetznerOne we likeHetzner is an independent German hosting company offering VPS, dedicated servers, cloud and storage infrastructure from its own data centers.
Huawei CloudHuawei's public cloud, selling virtual servers, managed Kubernetes, serverless functions and GPU instances.
Hugging FaceMachine-learning hub whose Spaces product hosts running applications — Gradio, Streamlit or an arbitrary Dockerfile — on CPU or rented GPU hardware.
HyperstackGPU cloud from London's NexGen Cloud, selling GPU virtual machines, Kubernetes clusters and a token-priced model API.
I
IkoulaOne of the oldest independent French hosts, trading since 1998, running two of its own certified data centers in France.
InfomaniakA Swiss hosting and cloud provider selling shared hosting, servers and public cloud infrastructure.
IONOSLarge European hosting group selling shared webhosting, VPS, dedicated and bare-metal servers and an IaaS cloud, listed on the Frankfurt exchange.
J
JarvislabsIndian GPU cloud renting GPU virtual machines, framework containers, managed model endpoints and a token-priced model API, billed by the minute.
K
L
LambdaUS GPU cloud selling on-demand NVIDIA GPU instances, reserved multi-node clusters and private cloud capacity.
Latitude.shBare-metal and GPU infrastructure billed hourly with automated provisioning, an API, SDKs, a CLI and Terraform support.
LeasewebA Dutch infrastructure provider selling bare metal, virtual servers and infrastructure as a service.
LinodeLinode is a US VPS provider founded in 2003, acquired by Akamai in 2022 and now sold as Akamai Cloud.
Liquid WebLiquid Web is a US hosting provider from 1997 specialising in managed VPS, dedicated servers, cloud and WordPress hosting, with 24/7 support.
M
ModalServerless compute platform for Python, where functions are decorated in code and run in containers on CPU or GPU with per-second billing.
N
NebiusAmsterdam-based AI cloud selling GPU and CPU virtual machines, managed Kubernetes and Slurm, serverless GPU containers and a model API.
nineA Swiss provider selling managed Kubernetes, managed and root servers, databases and object storage, all from data centers in Zurich.
NorthflankBritish platform running containers, jobs and managed databases on its own cloud, in the customer's cloud account, or on a self-hosted Kubernetes cluster.
O
OVHcloudFrench hosting company selling dedicated servers, VPS, an OpenStack-based public cloud and web hosting from data centers it builds and operates itself.
Q
QoddiManaged application platform deploying Node, Java, Go, Ruby, PHP, Python and Docker workloads with per-second billing and managed databases alongside.
R
RunpodUS GPU cloud renting containerized GPU Pods, serverless GPU endpoints and hosted model APIs.
S
Sakura InternetA Japanese cloud and infrastructure provider selling shared hosting, virtual servers and dedicated machines.
SaladCloudContainer platform that runs customer images on idle consumer GPUs contributed by a distributed network of home and small-business machines.
ScalewayFrench cloud provider selling instances, bare metal, managed Kubernetes, serverless containers and managed databases, with European data residency as the pitch.
SeewebItalian cloud provider founded in 1998, part of the DHH Group, offering cloud servers, VPS, and managed services.
SelectelSelectel is a Russian infrastructure company running its own data centers in Saint Petersburg and Moscow, selling dedicated servers, cloud servers and object storage.
SesterceFrench GPU cloud renting GPU virtual machines, bare-metal servers and inference endpoints by the hour from a prepaid balance.
SiteHostNew Zealand host that owns its Auckland data center and sells Cloud Containers, a product that runs prebuilt or custom Docker images.
T
Tencent CloudTencent's cloud platform, selling virtual machines, VPS bundles, GPU instances and serverless functions.
TensorDockUS marketplace selling GPU virtual machines on servers supplied by independent hosts, now owned by Voltage Park.
Thunder ComputeGPU cloud renting NVIDIA GPU instances by the minute, managed from a CLI, an editor extension or an MCP server.
U
UpCloudFinnish cloud infrastructure provider selling virtual machines, block storage and private networking, driven by a public API.
V
Vast.aiUS marketplace where independent hosts rent out GPU machines as containers or virtual machines at prices they set.
VerdaFinnish GPU cloud selling GPU virtual machines, multi-node clusters and serverless containers from data centers in Helsinki.
VultrCloud infrastructure provider selling virtual private servers, bare metal, GPU instances, object storage and managed Kubernetes, billed by the hour.
What is a neocloud?
The industry name for a cloud company built around renting GPUs, as distinct from AWS, Google Cloud and Azure, which sell GPUs beside everything else. Neoclouds are listed here like any other GPU cloud, with no filter of their own: the GPU field on each record says whether it sells instances, serverless GPU or managed inference.
Do I need a GPU cloud to use an AI model?
Only to run a model yourself. An app that calls a hosted model over an API needs no GPU, and an ordinary web app host will do. GPU clouds rent accelerators for training, fine-tuning or serving model weights, usually by the hour.
Who is the best GPU cloud provider?
We don't know. A static portfolio and a busy shop need different things from a host, and so do a freelancer and an agency with forty client sites. Nobody pays to be listed here and the order is alphabetical. The guide covers what to compare.

