GPUs for ML teams

Three engineers, three sets of provider accounts, three copies of the same setup script — and nobody sure what's still running on Monday morning. Aquanode puts the team on one balance, one deployment list, and environments that survive a stop.

8
Providers, from one team account
89
Regions
724
Live GPU offers

What breaks when a team shares GPUs

Everyone rebuilds the same environment

Custom nodes, weights, CUDA versions, a dozen pip installs — and it all evaporates the moment someone terminates a box. With storage connected, you save the environment as a snapshot and restore it later, on the same GPU or a different one. To hand it to a teammate, share the snapshot with them explicitly; they restore from it instead of rebuilding from your README.

Nobody knows what's running

Deployments belong to the team, not to whoever happened to launch them. Every member opens the same list, sees every deployment the team has running and who started it, and can stop one a teammate left up over the weekend. Live GPU utilisation, VRAM, temperature and power are on each running deployment, visible to any member.

The bill is a surprise

One shared balance for the team, drawn down by the second while a deployment runs — not per-engineer invoices you reconcile at month end. The team page breaks usage down per member, and pausing a deployment stops the billing while keeping the environment intact, so idle time between runs isn't charged.

What a team account gives you

Capability
What it does today
Roles
Owner, admin and member. Admins invite and remove people and rename the team; only the owner can delete it. The checks run on the server, not just in the UI.
Invites
An owner or admin generates a shareable invite link that expires after seven days. Anyone who opens it and signs in joins as a member.
Shared deployments
Every deployment belongs to the team. Any member can see it, watch its metrics, and stop it — including ones a teammate started.
Shared balance
One balance for the whole team, metered per second while deployments run, with a per-member breakdown of credits used and deployments started.
Live metrics
GPU utilisation, VRAM, temperature, power draw and clocks, plus CPU and system memory, refreshed every few seconds on any running deployment.
API keys
Keys belong to the team rather than to one person, so automation keeps working when someone leaves. Admins create and revoke them, and revoking is immediate.
Snapshots
Save a configured environment and restore it later on the same GPU or a different one. Share a snapshot with a teammate via a revocable link. Needs storage connected — the deploy screen tells you before you start.
Pause and resume
Stop paying for a deployment without losing it. Billing stops while it's paused and the environment comes back as it was. Pausing depends on the same storage connection snapshots use.

What isn't there yet

No spend caps or budgets. You can watch the shared balance and per-member usage, but you can't yet set a limit that stops deployments at a threshold.

SSH keys are per-person. Deployments, volumes and balance are shared across the team; SSH keys are not. Each member adds their own key and uses it on the deployments they create.

Snapshots aren't shared by default. A snapshot belongs to the member who took it. Sharing one with a teammate is a deliberate action — you create a share link, and you can revoke it.

If one of these is the thing standing between you and moving a team over, tell us — it moves up the list.

FAQ

How do I add someone to my team?

An owner or admin generates an invite link and sends it over. Whoever opens it and signs in joins as a member. The link expires after seven days and isn't tied to an email address, so send it only to the person you mean to add.

Does everyone share one balance?

Yes. Credits belong to the team, so every member's deployments draw down the same balance. The team page shows it alongside per-member usage totals.

Can I see what my teammates are running?

Yes. The deployments list is team-scoped: every member sees every running deployment, who started it, and its live metrics — and can stop one that was left up.

Can I set a spending limit?

Not yet. There's no per-team budget or cap today. What you get is a shared balance you can watch, per-second billing, per-member usage, and pause-to-stop-billing.

Can a teammate reuse an environment I set up?

Yes, by sharing it explicitly — save it as a snapshot and create a share link. Snapshots aren't visible team-wide by default, and a share link can be revoked.

Ready when you are

Stop paying for
idle GPUs.

Sign up in 60 seconds. Pay only for the GPU minutes you actually use.

Aquanode LogoAquanode

Your GPU environment, preserved. Pause it, move it, come back to it.

© 2026 Aquanode. All rights reserved.

All trademarks, logos and brand names are the property of their respective owners.