Supply side · Node operators
Run a node. Host isolated chains on hardware you already operate.
Sell capacity, not shell access. A UniNet workload arrives as a container with declared resource limits, on a machine you keep administering the way you already do.
The gap
The hardware is paid for. The idle hours are not.
Racks with headroom. A VPS fleet provisioned for peak. GPUs sitting between contracts. Utilization is never flat, and the gaps are not a scheduling failure — they are what provisioning for peak costs.
Selling the gaps is the obvious move. The offers are the problem. The usual ask from a network that wants your hardware is broad: shell access to a machine, a workload whose shape you learn after it lands, and a token you are expected to model yourself before you can price anything. That is three unpriced risks traded for one line of demand, and it is why most of these conversations end at the security review.
UniNet asks for less. A node process runs on the machine. Tenants receive containers with declared resource limits and a reservation with a declared boundary. The host stays yours.
- a containerdeclared resource limits
- reserved capacitycpu · memory · storage
- a reachable service addressdecoupled from the machine
- a metered recordwhat was reserved, for how long
- the host operating systemroot and remote access
- the hypervisor and the kernelyour patch cadence
- your other tenantsuntouched by the node
- physical and network operationsunchanged
Operator questions
Six questions, answered before you have to ask them.
These are the questions operators open with. The answers below describe the current state, not the target state — each one is graded in the table further down.
What runs on my machine?
Can it be isolated?
How much of the box does it take?
How is reliability measured, and by whom?
How do I get paid?
What happens if a tenant disappears?
Architecture
What sits on your machine.
One process, one identity, and containers bounded by what they declared. Everything the network knows about you is derived from what you publish.
YOUR MACHINE — host OS, your root, your remote access
│
│ one process, installed by you
▼
uninet node
│
├──────────────► operator identity
│ the key is yours · reliability
│ is scored against it
│
├──────────────► published capacity
│ cpu · memory · storage
│ shared or dedicated
│
├──────────────► atomic reservation
│ another suitable node when full
│
├──────────────► workload containers
│ declared resource limits
│ start · stop · remove
│
└──────────────► metered in UNE
settlement path defined
reward model not finalizedOperator path — identity, published capacity, reservation, containers, settlement
The path from operator identity through published capacity to a metered container is defined end to end. What is not settled is the last step: what a metered unit of delivered capacity is worth. That is one open question at the bottom of the diagram rather than a gap running through it.
Evidence
What is built, and how far.
Every status word below is read from the same file that renders the build status page, so this table cannot drift in the optimistic direction while that one stays honest.
| Capability | On your hardware | Status |
|---|---|---|
| Operator identity | A node operator is a distinct identity type at the network layer, alongside personal and AI-agent identities. It is a key you hold, not an account issued to you. | Ungraded |
| Reliability scoring | Delivery is scored against that identity at network level rather than inside one marketplace database, so the record is portable across whatever is scheduled to you. | Ungraded |
| Workload isolation | Hosted chains and jobs run in containers with declared resource limits. Isolation is container-level only, and it is not hardware-backed. | Ungraded |
| Shared and dedicated allocation | Published capacity can be allocated as shared or as dedicated. That choice is yours per machine, and it is the lever that decides how much of a box is committed to a single tenant. | Ungraded |
| Discovery and atomic reservation | Available CPU, memory and storage are inspectable across nodes. A reservation is atomic, with selection of another suitable node when the requested one is full. | Ungraded |
| Container lifecycle control | Start, stop and remove are operations on the node rather than requests you file with someone else’s control plane. | Ungraded |
| Payment | Signed transfers at the network layer, denominated in UNE, addressed to your operator identity, with human-readable recipients. What a delivered unit is worth is a separate and unfinished question. | Ungraded |
Ungraded means the capability is described in the architecture and has not yet been graded for production readiness by an engineer. It is not a claim that the subsystem is finished, and it is not a claim that it is broken. The full table, with every subsystem on the network rather than the seven that concern operators, is at build status.
Metering and payment
UNE is the unit of account for resources.
It is how reserved compute, memory and storage are counted, and how a delivered unit is addressed back to the operator who delivered it. On this page that is the whole of its job.
UNE. Reserved compute, memory and storage are metered in it, and transfers are denominated in it.
Reserved capacity rather than advertised capacity — what a workload holds against the pool you published, for the period it holds it.
Node operators can be paid for the capacity they deliver. Payment is addressed to the operator identity, not to a listing or an account someone else administers.
The network records a reliability score against each operator identity, kept alongside the record of what was delivered.
No rate card, no price target, no return expectation, no yield figure.
Reward parameters are not finalized, so no number we printed here would be defensible, and printing one anyway is how operators end up planning against fiction. This is not investment material. UniNet does not publish price targets or return expectations.
Limitation
Read this before you commit hardware.
Four subsystems bear directly on what an operator can plan around today.
Reward economics are not finalized, and there is no live consumer demand on the network yet — capacity discovery is a primitive, not a queue of paying jobs.
Full multi-node infrastructure orchestration is still evolving.
Validator production deployment is not complete.
Storage endpoints are still being connected.
Demand is the honest gap. There is no live consumer demand on the network today. Capacity discovery is a primitive the network exposes, not a queue of paying jobs waiting to be picked up, and publishing capacity into it does not by itself produce a tenant. Size a first deployment against a machine you can spare, not against a revenue line.
Possibility
What you can do now.
You can run capacity and hold an operator identity now.
Treat the first period as a technical pilot, not contracted revenue. If you need contracted revenue this quarter, wait.
Onboarding is a conversation with an engineer, and it is short. It covers what hardware you can commit and for how long, whether the machine can carry a node process alongside what it already runs, what isolation your existing tenants require of you, and whether you want capacity published as shared, dedicated, or both. If the answer is that UniNet is not worth your rack this quarter, that is a legitimate outcome of the call.