Hyperstack

Spot VMs

Run high-performance GPU workloads at a fraction of the cost with Spot Virtual Machines, ideal for AI, batch jobs and fault-tolerant tasks that don’t need guaranteed uptime. Launch today on Hyperstack.

Launch Spot VM

Ideal Use Cases of Spot VMs

Launch Spot VM

Experiment with AI

Train and experiment with workflows where you can manually save checkpoints and resume later if interrupted.

Run Batch Jobs Efficiently

Execute repeatable jobs like data processing pipelines or rendering, tasks that benefit from affordable and scalable compute.

Test Environments with Flexibility

Use Spot instances for dev/test environments, background processing or staging pipelines where occasional interruptions are acceptable.

Scale Compute On-Demand

Mix Spot and On-Demand Virtual Machines to expand your compute power while managing your budget.

Spot VM Pricing

Spot VM pricing offers fixed percentage discounts over on-demand pricing. Get the same GPU performance at a 20% lower cost.

GPU

OnDemand

Spot VM

H100 SXM

$2.4

$1.92

H100 NVLink

$1.95

$1.56

H100

$1.9

$1.52

A100 NVLink

$1.4

$1.12

A100

$1.35

$1.08

L40

$1.0

$0.80

A6000

$0.5

$0.40

A4000

$0.15

$0.12

H200

$3.5

$2.80

What to Know Before Launch

Know these considerations before launching a Spot VM:

No Termination Notice

Spot VMs can be reclaimed at any time if capacity is needed.

No Hibernation or Snapshots

Spot VMs are ephemeral and can’t be paused or saved as custom images.

No Boot-from-volume

All Spot VMs must launch from a base image only.

No Data Persistence

Any unsaved data will be lost if the VM is terminated.

Public IP Pricing

Spot pricing discounts don’t extend to public IP usage.

Try Hyperstack Spot VMs Today

Get started with GPU compute for AI, batch jobs and large-scale tasks without the high cost. Spot VMs are your go-to solution for scalable, cost-effective cloud computing.

Try Spot VMs

Frequently Asked Questions

Our product support and development go hand in hand to deliver you the best solutions available.

What are Spot VMs on Hyperstack?

Spot VMs let you access unused GPU capacity at discounted rates compared to on-demand VMs. They're ideal for cost-effective GPU computing when your workload can tolerate interruptions.

How does a Spot VM differ from an On-Demand VM?

Spot VMs offer a lower price but no guaranteed uptime and can be terminated at any time without notice. On-demand VMs provide stable, persistent resources but at a standard price.

What are the use cases for Spot VMs?

Spot VMs are ideal for fault-tolerant and flexible tasks such as machine learning model training, batch processing, dev/test environments and scaling compute-heavy workloads affordably.

Can I save data on a Hyperstack Spot VM?

No. Data stored on Spot VMs is ephemeral and data is lost upon termination. Always save important data to external storage or use checkpointing to preserve progress.

How do I launch a Spot VM on Hyperstack?

Select a VM flavor that includes "spot" in the name (e.g., A100-80G-PCIe-spot) when deploying your VM on Hyperstack. For full steps, refer to our deployment guide here.

Are there any limitations to using Spot VMs?

Yes. Spot VMs don’t support hibernation, snapshotting, boot-from-volume or data persistence. They may also be terminated without warning based on capacity needs. For the official policies, conditions, and disclaimers regarding Spot instances, please refer to section 4.5 of our Terms and Conditions..

What happens to attached volumes if a Spot VM is terminated?

Attached volumes may enter an error state or become corrupted. It is your responsibility to delete or recover them manually after termination.

What flavors are available for Spot VMs in Hyperstack?

The availability depends on current capacity and can vary by region or GPU type.

Can I use Spot VMs for machine learning or AI workloads?

Yes. Spot VMs are ideal for training and experimenting with workflows where you can manually save checkpoints and resume later if interrupted.