Skip to main content
Staking is integral to the Arcium Network, activating Arx nodes and unlocking their computational resources. Without sufficient stake, an Arx node cannot activate or join Clusters, so computational work is allocated to operators with committed stake.

​
Operation requirements

Three separate requirements govern a Node’s participation:
  1. Registration stores the Node’s metadata, including its declared CU capacity (cu_capacity_claim). The claim is informational; it does not gate stake, scheduling, or rewards.
  2. Activation requires the Node Operator to self-delegate stake meeting the program’s MIN_SELF_DELEGATION requirement of 1,000 ARX. Additional self and third-party delegations grow the Node’s stake and its stake-weighted reward share.
  3. Cluster eligibility requires the Node’s primary stake to meet a floor set by the Cluster’s mempool size:
Because activation already requires 1,000 ARX of self-delegation, the general minimum is the effective baseline for tiny, small, and medium Clusters; only large Clusters raise eligibility, to 10,000 ARX. The floor is enforced when a node is invited to a Cluster, and a node bound to a Cluster cannot reduce its primary stake below the applicable floor.

​
Delegation flow

Once activated, Arx nodes can receive additional stake delegations from both Node Operators and third-party delegators. Delegations grow the Node’s stake-weighted share of rewards.

​
Delegation mechanics

  • Lockups: Delegated stake has a minimum lockup of four epochs; a Node’s own primary stake has a minimum lockup of eight epochs.
  • Maturation: The lockup applies immediately to a stake change, but the delta joins the Node’s reward and leader totals after two epoch-settlement rotations through the pending-stake queue.
  • Rewards: Computation rewards accrue into a per-Cluster pot and are paid stake-weighted at epoch finalization. Each Node’s share is pot × node_stake / cluster_total_stake, where node_stake = primary_stake + delegated_stake.
  • Recovery Peers: Recovery Peers attached to the MXE receive a separate slice on the same stake-weighted basis.
  • Delegator payout: Delegators earn a pro-rata share of the rewards their Node receives, minus the Node Operator’s percentage fee.
After each epoch is finalized, Delegators claim their accrued rewards, which are paid out to a destination account. A Delegator can undelegate only after the lockup ends. The removal then matures after two epoch-settlement rotations, gated on the Node’s primary stake being finalized, after which the stake can be withdrawn.

​
Epoch-based rewards

Rewards are allocated at the end of each epoch. Withdrawing stake or switching a delegation to a different Node (redelegation) follows the same maturation: the stake stays locked until the removal matures, after which it is reassigned or withdrawn. For details on how computation fees are structured, see Pricing and incentives.

​
Fee allocation

Each computation fee on a Cluster’s MXE splits as:
  • 70% Cluster pot: paid stake-weighted to the Cluster’s Arx nodes at epoch close.
  • 20% Recovery Peer pot: paid stake-weighted to that MXE’s Recovery Peers.
  • 10% protocol-retained: the remainder after the Cluster and Recovery Peer shares, including integer rounding, held in the protocol fee pool.
A failed computation accrues a reduced fee, which is then allocated with the same percentage split.

​
Leader weight

On each Cluster, computations are pulled from the mempool one at a time in dispatch cycles. Once per dispatch cycle, the current leader Node may pre-empt the priority-fee queue and select one computation from the mempool via set_leader_choice. Leader weights are refreshed at epoch finalization; the leader for each dispatch is selected using the D’Hondt method from those weights, where each Node’s weight is primary_stake + delegated_stake. Slots not consumed by the leader fall back to priority-fee ordering and lock-conflict checks. This gives Nodes a stake-proportional opportunity to influence the order of execution beyond what priority fees alone determine, while preserving the priority-fee market for the remaining slots.

​
Example workflow

To illustrate how the delegation process works:
  1. A Node Operator self-delegates stake to activate the Arx node.
  2. Third-party delegators contribute additional stake, growing the Node’s stake-weighted share.
  3. Once active, the Node executes computations. The Cluster’s reward pot at epoch close is distributed across Nodes proportional to each Node’s total stake, then each Node’s slice is distributed across its Delegators pro-rata, minus a fee collected by the Node Operator.
  4. After the epoch is finalized, Delegators claim their accrued rewards, which are paid out to a destination account.
Key points:
  • Arx nodes with insufficient stake cannot activate or join Clusters.
  • Rewards are distributed stake-weighted across a Cluster’s Arx nodes and the attached MXE’s Recovery Peers.
  • Stake changes are subject to a minimum lockup and take effect after two epoch-settlement rotations.