§ Wiki · Wiki entry

Node Provider Documentation

How independent operators run the hardware that hosts the Internet Computer — the role, the current onboarding status, the lifecycle from proposal to operation, and the rewards.

·May 4, 2026·6 min read

A node provider is an independent operator that owns and runs node machines on the Internet Computer. The network is not hosted in any single data center or by any single company; it is a federation of these operators, deliberately spread across jurisdictions and hardware suppliers so that no one party can take it down or reshape it.

This entry summarizes what node providers do, how they get accepted, what they are paid for, and the current status of new onboarding. For the underlying network the providers run, see Internet Computer Protocol.

[!WARNING] No new node machines are being onboarded. The network reached its target topology in December 2023 and currently does not require additional capacity to meet its decentralization objectives. New node machine proposals will not be approved by DFINITY until subnet demand warrants further infrastructure. Existing providers continue to operate and earn rewards as before.

What a node provider does

A node provider supplies and operates node machines — the physical servers that host the network. Their responsibilities span the full lifecycle of those machines:

  • Procure hardware that meets the published node specification. See the Node Provider Machine Hardware Guide.
  • Place it in suitable data centers, in regions that satisfy the network’s geographic-distribution rules. See the Data Center and ISP Guide.
  • Install IC-OS, the operating-system image that turns a generic server into a node. See the Gen-2 Node Deployment Guide (current) or the Gen-1 Node Deployment Guide.
  • Maintain the machines — diagnose hardware faults, replace parts, resolve networking incidents. See the Node Provider Maintenance Guide.
  • Operate without live support from DFINITY. Onboarding and day-to-day maintenance are the provider’s own responsibility; community support is available via the public Matrix channel and the developer forum.

Diversity is not optional. The more varied the set of providers — by ownership, by jurisdiction, by hardware supplier, by network upstream — the harder the network is to coerce. Acceptance criteria favor applicants who add diversity rather than replicate it. See Decentralization and Security for the rules in detail.

Acceptance: the NNS route

Becoming a node provider is a governance action, not a contract with DFINITY. The flow is:

  1. Learn the role and the network it supports — start with the Node Provider Roadmap.
  2. Confirm fit against the published skill, hardware, and networking requirements. Operators are expected to be able to install, run, and troubleshoot their own infrastructure.
  3. Submit a proposal to the Network Nervous System (NNS) — the Internet Computer’s on-chain governance system — for community approval. Provider acceptance is decided by NNS vote. The full procedure is in Node Provider Onboarding, along with the Self-Declaration and Validation of Candidate Node Machines procedures.
  4. Procure hardware and sign data center contracts.
  5. Install IC-OS on the nodes; have them validated.
  6. Operate, including ongoing maintenance and any required migrations between hardware generations.

A provider who does not submit a reward configuration proposal will not receive rewards, even if their nodes are otherwise running correctly. Reward configuration is a separate, mandatory step — see the Reward Configuration Guide.

What “useful work” means

Node providers are paid for useful work — not for merely keeping a machine powered on. Rewards are a function of the work the network schedules onto each node and the provider’s region; the rates themselves are set by NNS proposal and adjusted over time. The formal description is in Proof of Useful Work; the rate card and regional structure are in Node Provider Remuneration.

If a node is sold from one provider to another, the receiving provider is expected to relocate it to a country that maintains or improves the network’s overall decentralization. Sales that would concentrate the network in a single jurisdiction are not approved.

New regions, where no provider currently operates, require a community discussion on the developer forum before a reward rate can be agreed.

Hardware generations

The fleet has gone through two generations:

Network requirements

Each provider is asked to deploy at least two nodes in every data center they operate in with IPv4 reachability and a registered domain name, attached to the first two nodes in their first rack at that site. Beyond that pair, additional nodes can run on IPv6 only. This rule preserves the network’s ability to reach back into a data center regardless of how the provider has configured their wider network. The full networking spec is in the Node Provider Networking Guide, and domain-name procedures are in the Node Provider Domain Name Guide.

Resources

All node-provider procedures live on this site. Use these as your starting points:

Acceptance and onboarding

Deployment

Hardware, networking, data center

Decentralization, security, legal

Operations

Economics and reference

Community support

Related entries on this site