TensorDock Cloud:
GPUs from $0.29/hr

We make the GPU servers that cost too much on other clouds available for a low, hourly price.

Start with $5 and launch a server in 45 seconds.

Performant Cloud GPUs

From NVIDIA RTX 4000s to A100s, our GPUs have the power to fit your needs

Secure & Reliable

We partner with top tier suppliers with proven uptime and security measures


Be cost-effective. NVIDIA V100s from $0.57/hour. That's up to 70% cheaper.

... ... ... ... ... ... ... ... ... ... ... ... ... ...

Three GPU cloud products, all best-in-class

All products come with bandwidth included and are usually between 50 and 80% cheaper than competing products on the market. They're developed in-house by our 100% US-based team.

Cheap Hourly GPUs

Core Cloud

Hourly-billed, resilient, scalable, and secure cloud for burstable workloads. Up to 70% cheaper than incumbent clouds.

For workloads that require reliability and scalability.

  • Prices start at $0.29/hour
  • A stopped VM incurs less costs than a running VM
  • 3x-replicated storage
  • 10 gbps networking
  • 3 data centers in North America
  • API available
  • Instant Deployment
Learn more Deploy Cloud GPU
Cheap monthly GPUs

Subscription Servers

Servers on monthly or longer terms for continuous workloads (e.g. ML inference). Up to 50% cheaper than Core Cloud.

For long-term workloads that benefit from lower costs.

  • Prices start at $100/month
  • Stopping the VM does not reduce the cost
  • Local or replicated storage
  • 1-10 gbps networking
  • 11 data centers throughout the world
  • No API available
  • Up to 24 hours wait time
Learn more Order Subscription
Coming Soon in August

Marketplace Platform

Servers operated by independent hosts who run our software. Truly unbeatable prices.

For cost-effective workloads that don't require 100% uptime.

  • Prices start at $0.13/hour
  • Stopping the VM does not reduce the cost
  • Local storage
  • 0.1-1 gbps networking
  • Endless locations throughout the world
  • Undocumented API
  • Instant Deployment
Learn more Deploy Marketplace VM

Chat with Sales

Response within 24 hours

Need help deciding, or need something custom? From storage servers to 16-A100 NVLink rigs, we've got the compute you need and the expertise you can depend on.

Schedule a video chat Send us an email or message

Latest Blog Post: Subscription Orders are Live!

Subscription servers are servers that are rented for a longer period of time (e.g. a month) than a typical hourly server. Already, you can order TensorDock subscription servers in 11 locations, with...

Riya Thapar

A customer-first focus
for everything we do.

Engineered for excellence, priced for scale.

Built by builders, for builders

No slow licensed software, no outsourced engineering team. Built 100% by US-based engineers for scale and reliability.


Being integrated with our customers' tech stacks is a core pillar of our business. We add features to our API before we add them to our dashboard. Well-documented, well-maintained, well-everything.

1,000+ in stock GPUs

GPU stock numbers available in realtime.

5,000+ reservable GPUs

5 fleets of partner GPUs, at the same low TensorDock price, on request.

Free transfer / bandwidth

Stop worrying about bandwidth overages — eliminate your #1 cost at other clouds. Just contact us if you anticipate needing 100TB+ of monthly traffic.

Learn more about the TensorDock difference

The world's use us.

airgpu uses TensorDock's API to deploy Windows virtual machines for cloud gamers. TensorDock's abundant GPU stock enables airgpu to scale during weekend peaks without worrying about compute availability.

airgpu is a popular, renowned cloud gaming service known by gamers worldwide

ELBO uses TensorDock's reliable and secure GPU cloud to create generate art. TensorDock's highly-performant servers run their workloads faster on the same GPU types than the big clouds.

ELBO.ai creates managed AI services like art generators and logging and has been featured in The Spectator.

Researcher Skyler Liang from Florida State University uses TensorDock's A40s and A6000s to work with GAN networks. TensorDock's low pricing enables FSU researchers to do more with their limited university compute budgets.

Florida State University researchers receive reimbursements from using TensorDock servers.

Creavite uses TensorDock's Windows GPU servers with Adobe software to render logo animations as well as CPU-only servers. The presence of both types of servers enables Creavite to tightly integrate their workflows.


TensorDock's goal is to democratize access to expensive HPC hardware.

Our goal is enable you — whether you are a sole developer, funded startup, or enterprise — to compute groundbreaking innovations easier at industry-leading prices. Learn about our story.

  1. Jupyter Lab/Jupyter Notebook preinstalled
  2. Affordable prices
  3. Real human support
  4. Secure and reliable

We're a team of ex-engineers of cloud platforms, here to create the platform that we couldn't before. We're always receptive to feedback, just contact us!

0+ API


0+ GPUs

In Stock

0+ GPUs


Frequenty Asked Questions


My enterprise needs ...

Let's chat! We're happy to set up custom solutions for larger customers. Learn more.


How are we so cheap?

First off, we aren't cheap — the others are too expensive. Big clouds are getting away with charging obscene gross margins, often 80%+. By charging a more reasonable margin and running our company efficiently, we can afford to be up to 70% cheaper than others while turning a profit, which we have done every consecutive quarter since our founding.


What virtualization? Docker containers?

No Docker here. We provide full KVM virtualization with root access and dedicated GPUs passed through.


How is billing done?

We operate on a pre-paid model: you deposit money and then provision a server. Once your balance nears $0, the server is automatically deleted.

For long-term servers, we offer a monthly subscription servers, and you can also discuss custom options by contacting us.

Get started

Spin up a GPU server from $0.29/hour.

Go ahead — go build the future of tomorrow — on TensorDock. Cloud-based machine learning and rendering has never been easier and cheaper.

Deploy a GPU Server
... ... ... ... ... ... ... ... ... ... ... ... ... ...