get in touch

t3ra

decentralised private infrastructure

data center you control.

talk to us → see the command center ↓

/what we do

we deliver your own data center, and it runs like a cloud.

hardware, operating system and operations, installed where you decide. you get the simplicity of the public cloud on infrastructure you control: a lower bill, your data under your jurisdiction, and no ops team to hire.

/what we deliver

the whole data center, from the rack up.

one team ships the nodes, the os and the support. on your premises, in your colocation, or at a hosting provider.

  • /hardware

    the nodes.

    pre-configured, sold or leased. or the hardware you already own.

  • /software

    an operating system.

    a decentralised mesh pooling cpu · gpu · storage · network. nothing central to fail.

  • /operations

    it runs itself.

    nodes find each other and self-heal. install and support come from us.

controlled by you, not us.

/how it works

from bare hardware to one system in under 10 minutes.

  1. 015-min os install per node.

    one deployment per unit. no master node.

  2. 02nodes find each other.

    self-healing peer-to-peer mesh. no config.

  3. 03compute · storage · network

    unified. ai-compatible apis.

start with any workload that is easy to move. zero migration risk. then scale up to ai inference, on gpus you control.

survives >60% node loss.

  • any hardware

    x64 · arm64 · risc-v · gpu

    • our os or your linux distro
    • new or existing hardware
    • under 10 min setup

  • p2p mesh

    self-healing cluster

    • no dedicated control node
    • nodes join and leave gracefully
    • survives >60% node loss

  • unified apis

    cpu · gpu · storage · network

    • declarative, iac and ai compatible
    • works with existing tooling
    • air-gap ready

/command center

the operator ui, for people who run things.

the dashboard opens on the live mesh. when everything is healthy, the topology recedes. when something breaks, the view moves to where it broke: the node, the event, and every entity it touches.

click around: a preview of the command center with demo data. subject to changes.
  • the page is the mesh

    an adaptive surface that follows the mesh state in real time. no stale overviews.

  • self-healing, in place

    watch migrations, failovers and rebalancing unfold where they happen.

  • the path to act

    event context, entity properties and deep links side by side. no hunting, no modal detours.

/start small

land where it hurts. grow from there.

nobody moves production off the cloud for an unknown startup, so we don't ask you to.

  1. 01

    day 1

    ci runners

    first cluster live. first win.

  2. 02

    first month

    dev environments

    no production risk. the cloud bill starts to drop.

  3. 03

    first quarter

    production

    the whole stack moves over.

  4. 04

    year 1

    ai inference

    same mesh, on gpus you control. no new project.

/pricing

pricing per node.

however you come in, you pay one subscription per node. the longer you commit, the lower the rate.

three ways in

  1. 01 / your hardware

    subscription only.

  2. 02 / our nodes, sold

    hardware + subscription.

  3. 03 / our nodes, leased

    lease + subscription.

    no upfront capex.

commitment period

period discount
monthly base rate
1 year -10%
2 years -20%
5 years -30%

edition

standard

core platform

unified compute · storage · network · iac and ai-compatible apis

premium

advanced features

priority support · sla guarantees · extended observability and policies

+ we sell or lease pre-configured hardware. install included. live in under 10 minutes.

ask for a quote

/roadmap

the setlist. what's shipped, what's playing, what's next.

  1. A1core command plane

    versioned commands, replicated mesh state, an append-only state log.

    shipped
  2. A2compute, network and storage

    vm lifecycle, volume attach and detach, snapshots. first live mesh, 5 nodes.

    shipped
  3. A3command center v1

    the operator ui. networks, firewall, full audit.

    now
  4. B1design-partner release

    designing the platform catalog services with you.

    planned
  5. B2general availability

    self service registration.

    planned

/contact

run your own cloud. let's talk.

we're onboarding design partners. if you operate infrastructure, are betting on sovereignty, or invest in deep tech, we want to hear from you. or just drop your email for the occasional update.