How much storage does a GPUwerk DGX Spark have?

1 TB of NVMe per node, per GPUwerk's DGX Spark hardware page and pricing page. That's shared between the container filesystem and /workspace, the directory that persists across a stop.

What survives a stop versus a terminate?

Stopping keeps /workspace exactly where it is on the same node, and the SSH port and hostname stay the same, so restarting resumes the same machine. Anything installed outside /workspace lives on the container filesystem and is gone after a rebuild. Terminating deletes the workspace permanently, which is why it requires typing the instance name to confirm, per GPUwerk's pricing page FAQ.

Does GPUwerk back up my data?

Not as a backup service. GPUwerk keeps only a periodic recovery copy of /workspace, refreshed roughly every six hours, for hardware failure recovery. If credit reaches zero or a monthly budget is hit, the instance is stopped and the machine is released for someone else to rent; the workspace is saved off it for 7 days so topping up and pressing start restores it, and deleted for good after that window, per GPUwerk's pricing page.

Docs analysis

DGX Spark/DGX Spark storage: what's persistent and what to back up

DGX Spark storage: what's persistent and what to back up

By Samuel Seidel · Published September 9, 2026

One node, 1 TB of NVMe, and two very different outcomes depending on whether you stop or terminate. This page lays out exactly what survives which action, using the language already on GPUwerk's pricing and SSH docs, and nothing beyond it.

The storage itself

Every GPUwerk DGX Spark instance, on-demand or the 256 GB cluster, comes with 1 TB of NVMe, per the spec tables on the pricing page and the hardware page. You land as root in /workspace over SSH, per GPUwerk's SSH keys documentation, and that 1 TB is shared between the container's own filesystem and whatever you keep in /workspace.

The one directory that matters: /workspace

/workspace is the part of the container that survives a stop. GPUwerk's SSH documentation is explicit about this: it "stays right where it is on the same node," and starting the instance again "takes seconds." Anything installed outside /workspace, meanwhile, lives only on the container filesystem and is gone after a rebuild. The practical rule follows directly from that: keep your models, checkouts and datasets under /workspace, and treat anything installed elsewhere as disposable setup you'd need to redo.

Stop versus terminate, exactly

These two actions do different things to your data, and the pricing page's FAQ spells out both precisely:

Stop keeps /workspace in place on the same node. The Spark stays reserved for you, the container stops, and the SSH port and hostname don't change, so restarting resumes the same machine, though model presets may need several minutes to warm back up. A customer-requested stop bills 75% of the running rate to hold that reservation, a 25% discount off the $0.79/hour on-demand rate.
Terminate deletes your workspace from the node for good, which is why the console requires typing the instance name to confirm. This is the way to stop paying entirely; a stopped instance keeps billing at the reduced rate as long as it holds its reservation.

There's a third path worth naming separately, because it isn't something you choose: if your credit reaches zero or you hit a monthly budget, the instance is stopped and the machine is released for someone else to rent. Your workspace is saved off it and kept for 7 days, so topping up and pressing start picks up where you left off on whichever Spark is free; after 7 days without that, the saved workspace is deleted for good, per the pricing page. Low-balance emails warn you before this happens, and auto top-up can reduce the risk, but the pricing page notes plainly that payments can fail, so a low-balance warning is not a guarantee you'll catch it in time.

No backup service, stated plainly

GPUwerk's pricing page states this without qualification: it is not a backup service. It keeps only a periodic recovery copy of /workspace, refreshed roughly every six hours, solely to recover from hardware failure, alongside the separate 7-day saved copy kept after a zero-balance release. Neither is a substitute for your own backup, and GPUwerk doesn't guarantee either copy is current or complete. The pricing page's own callout on this point is direct: "Protect your results... Keep your own backup."

What that means in practice is that anything in /workspace you can't afford to lose needs its own copy somewhere else, before you terminate, and ideally on a schedule if the instance runs long enough that a zero-balance release is a real risk. rsync, scp, or pushing to your own object storage or git remote over the SSH tunnel documented in the SSH keys guide all work for this; GPUwerk doesn't provide or manage the backup destination, since that decision belongs to you, not to the node.

Practical checklist

Put everything worth keeping in /workspace. Nothing outside it survives a rebuild.
Stop, don't terminate, if you want the exact same machine back later at a 25% discount off the running rate.
Terminate only once you've pulled anything you need, since it deletes the workspace immediately and irreversibly.
Keep your account funded, or set up auto top-up: a zero-balance release still saves the workspace for 7 days, but past that it's deleted the same as a manual terminate, with no backup on GPUwerk's side either way.
Back up anything genuinely important off the node, on your own schedule, using your own storage.

See the full billing FAQ for the exact stop and terminate pricing, or a first engagement if you want help designing a backup and rollout plan around a specific workload.