...

Affordable GPU servers for everything AI.

Affordable GPU servers for everything AI.

H100s from $2.80/hr. No quotas, no commitments.

TensorDock brings a global fleet of GPU servers to your fingertips, for 80% less than other clouds.

Start with just $5 and launch a server in 30 seconds.

Leading AI startups choose TensorDock

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

On demand, no nonsense

No quotas, hidden fees, or price gouging. We mean it.

Focus on building your product, not stressing over the infrastructure behind it. Deploy a GPU

...
...

Industry's widest GPU selection

The perfect GPU for every use case and budget, from RTX 4090 to HGX H100 SXM5 80GB. Find your niche among our 45 available GPU models.

Take advantage of our consumer GPUs from $0.12/hr for
up to 5x inference value. Switch now

KVM Virtualization

Root access and dedicated GPUs. Get full OS control, manage your own drivers and never run into compatibility issues again. Plus, Windows 10 support.

Multithreaded and optimized end-to-end for uncompromising VM deployment speed. Docker included on all VM templates.

See for yourself

...

Our most popular GPUs

Enterprise-grade servers hosted in tier 3/4 data centers

...

H100 SXM5

From $2.80/hr

For builders needing training and inference with no compromises.

Deploy an H100 SXM
...

A100 SXM4

From $1.80/hr

The best balance between price and performance for AI inference.

Deploy an A100 SXM
...

RTX 4090

From $0.35/hr

Truly unbeatable value for gaming, image processing, and rendering.

Deploy a RTX 4090

Engineered for excellence, priced for scale

The hyperscaler experience without the hyperscaler price.

Scale globally

Up to 30,000 GPUs available through our close partners. Hundreds ready to deploy now on our dashboard, distributed across 100+ locations in over 20 countries.

We're the last cloud you'll ever need. Reach your customers where they are. Scale now

...

An API you can depend on

Well-documented, well-maintained, well-everything.

Built from scratch, with metadata, availability, and everything you need to manage your servers on TensorDock.

Try it out

Bulletproof reliability

All hosts are vetted by TensorDock for quality hardware, technical knowledge, and communication skills, before their GPUs are listed.

TensorDock holds hosts to a 99.99% uptime standard and requires maintenance to be scheduled at least two weeks in advance. Hosts who don't meet continue to meet our standards are removed from our platform.

Learn more

...

The cloud for AI. And everything else.

Servers for any niche, for up to 80% less than other clouds. All available on demand.

GPU Cloud

Flagship

AI workloads requiring shell access

Our globally distributed GPU fleet. 99.99% uptime standard and no ingress/egress fees.

Deploy a GPU

Instant VMs

Beta

Pre-configured, deploys in seconds

Multithreaded stack optimized end-to-end for uncompromising VM deployment speed.

Deploy an Instant VM

CPU Cloud

New

Transcoding and batch processing

The latest Xeon and EPYC CPUs, available from $0.012/hr in secure, multihomed data centers.

Deploy a CPU

Frequently Asked Questions

?

My enterprise needs...

Let's chat! If you don't see what you need, we can work it out. We'll get back to you within 24 hours.

Book a meeting Message us
?

How do you guarantee the security of customer data?

Most hardware on our platform is now hosted in certified data centers. But as an additional layer of security, we revoke SSH access from our hosts so they cannot access customer data without our knowledge.

TensorDock background check all employees and restricts hostnode access to only those who need it. We have an agent monitoring hostnodes for logins and suspicious activity, from any source. Learn more.

?

How does billing work?

We operate on a pay as you go model. Users deposit funds, and we deduct balance continuously after a server is deployed. When your balance reaches $0, your servers are automatically deleted. If you need to rent servers long-term, reach out to discuss our reserved pricing.

?

Why does the same GPU/CPU have multiple prices?

We're a marketplace of independent hosts who compete and set their own pricing. As the customer, this ensures you always have access to the market's best pricing.

Additionally, hosts have different locations and redundancy, leading to varying costs. Our marketplace offers you options to pay based on your preferences.

The world's most build on TensorDock.

airgpu relies on TensorDock's API to deploy Windows virtual machines for cloud gamers. TensorDock's abundant GPU stock allows 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 generate art. TensorDock's highly cost-effective servers run their workloads faster for less than the big clouds.

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

Professor Skyler Liang from Florida State University researches GAN networks with TensorDock GPUs. TensorDock's superior economics allow researchers to do more with their limited university budgets.

Florida State University researchers receive reimbursements from using TensorDock servers.

Creavite combines TensorDock's Windows VMs with Adobe software to render logo animations. TensorDock's CPU-only instances allow Creavite to fully integrate their workflows and stay on one cloud.

...

We're democratizing access
to high-end cloud infrastructure.

We connect customers to the best available compute. Cutting-edge hardware in Tier 3/4 Data Centers, for maximum security and reliability. Converted mining rigs, for maximum price-to-performance. We've got both, for whatever suits your needs best.

Our marketplace approach guarantees the best deal in the industry. Get the availability and location distribution of a hyperscaler, without the quotas or nonsense.

100+ Locations

Available

0+ GPUs

In stock

0+ GPUs

Through partners

Get Started
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...