roadmap

What we're
building next.

The next three months of Metalhost, in the open. Dates are targets, not contracts — when something ships, it flips to shipped right here.

already live

Shipped and running in DC-1.

GPU passthrough VMs

RTX 4090, real passthrough

Bare-metal leases

custom ISOs, rescue mode, BMC console

NFS file shares

shared storage across VMs

VM metrics & consoles

graphs, serial + VNC in-browser

Webhooks & audit log

every event, signed and delivered

Per-second billing

card or crypto, spend alerts, budgets

august
[ 2026.08 ]

Backups you can trust.

Snapshots exist today; August makes them automatic, complete, and governed by retention you set once.

planned

Snapshot schedules & retention

Hourly, daily, or weekly snapshots with keep-N retention — per VM or across a whole project. Set it once, stop thinking about it.

planned

Full-VM backups

Snapshots capture every attached disk — boot and data — as one consistent set, and restore brings the whole machine back.

planned

Disk snapshots

Point-in-time copies of any disk, attached or not. Spin up a fresh disk from any snapshot.

september
[ 2026.09 ]

Bring your own observability.

Your dashboards, your alerting stack, our data. September opens the platform's telemetry to your tooling.

planned

Scoped API keys & service accounts

Read-only and per-service key scopes, plus service accounts for CI and automation — credentials that don't belong to any one teammate.

planned

Prometheus metrics endpoint

A scrape endpoint for your project — CPU, memory, disk, network per VM. Point your own Prometheus or Grafana at it, locked behind a metrics-scoped API key.

planned

Infrastructure alerts

VM down, CPU pegged, disk filling up — you define the rules, delivered in-app, by email, or to your webhooks.

october
[ 2026.10 ]

Networking, completed.

October rounds out the network story: keep your addresses, balance your traffic, boot your own images, drive it all from Terraform.

planned

Reserved & floating IPv4

Reserve public IPs, move them between VMs without a restart, keep them across rebuilds.

planned

Managed load balancers

L4 TCP balancing with health checks across your VMs. High availability without the duct tape.

planned

Custom VM images

Upload your own qcow2 images, boot new VMs from them, reimage existing ones in place.

planned stretch

Official Terraform provider

VMs, disks, firewalls, keys — the whole platform as code, on the public registry.

your move

Want something bumped up the list?

The roadmap bends toward what customers actually ask for. Tell us what's missing — we answer.