Max Burn

How the max burn subnet hyperparameter caps the dynamic TAO burn required to register a neuron on a subnet.

Max burn is a per-subnet hyperparameter that sets the upper bound on the dynamic TAO burn required to register a neuron on a subnet. Official documentation describes it as the upper bound for the dynamic TAO burn required for neuron registration on the subnet, so it is the ceiling the registration cost cannot exceed.

References: Subnet Hyperparameters

What It Controls

The hyperparameter caps the registration burn rather than fixing it. Registering a neuron requires burning an amount of TAO, and max burn is the most that amount can be on the subnet. It does not set the current cost or decide who registers; it only limits how high the burn can climb.

Reference: Subnet Hyperparameters

A Dynamic Cost Within Bounds

The registration burn is dynamic. The documentation describes it as moving between a lower bound and this upper bound: the cost rises as registrations occur and eases back during quieter periods, while staying inside that band. Max burn fixes the top of the range, so even at peak demand the registration burn is held at or below it.

Reference: Subnet Hyperparameters

Default and Setting

The documentation lists a default of 100 TAO and marks the parameter as owner-settable, so a subnet owner can adjust the ceiling for their subnet. The value in force is per-subnet chain state and can differ from the default, so the real ceiling should be read for the subnet in question.

Reference: Subnet Hyperparameters

Subnet Context

Max burn is one of the subnet hyperparameters, the on-chain state variables that configure a single subnet, for example netuid 1. Because each subnet carries its own hyperparameters, the registration burn ceiling is defined per subnet and applies only to registration on that subnet.

References: Subnet Hyperparameters, Register

Development Stage Context

The Introduction to Bittensor describes subnet development as moving from localnet to testnet and then mainnet. For max burn, that sequence changes how readers should interpret registration-cost ceiling examples.

In localnet, max burn hyperparameters can be tested in an isolated environment. Localnet burn ceilings do not represent production registration costs.

On testnet, registration burn rules can be exercised in a shared non-production network. Testnet max burn values are separate from mainnet subnet state.

On mainnet, max burn is a live per-subnet hyperparameter on production subnets. Observed registration ceilings depend on the selected subnet’s on-chain hyperparameter state (Subnet Hyperparameters).

The Bittensor Networks reference separates mainnet, testnet, and localnet. A max-burn example from one environment should not be read as representing production registration costs in another environment.

Relationship to Yuma Consensus

Max Burn and Yuma Consensus describe related parts of Bittensor’s incentive system. Yuma Consensus is the on-chain process that aggregates validator weight signals within a subnet into miner incentives and validator dividends, applying consensus clipping, bonding, and emission calculation (Yuma Consensus).

For readers, max burn names a specific part of that incentive picture, while Yuma Consensus names the consensus process that turns validator weights into the resulting incentives and dividends.

Reader Boundary

This page defines the concept at a high level. It does not report the max burn value set on any particular subnet or the current registration cost. Those are live chain state and should be checked for the relevant netuid. The default of 100 TAO is the documented value, and the parameter is owner-settable per subnet.

Reference: Subnet Hyperparameters

BurnHalfLife Eases the Burn During Quiet Periods

The Glossary: Register describes the dynamic registration burn as decaying over time according to BurnHalfLife while staying inside the documented MinBurn and MaxBurn band. Max burn still names only the ceiling; half-life names how quickly the live burn can fall when registration demand quiets down.

That decay keeps max burn from acting like a fixed price. During slow periods the burn can drift downward toward the subnet floor, but the documented upper bound remains the point it cannot exceed even when registrations surge again (Subnet Hyperparameters).

References: Glossary: Register, Subnet Hyperparameters

BurnIncreaseMult Steps the Burn After Each Registration

Official registration vocabulary also states that each successful registration can increase the dynamic burn through BurnIncreaseMult. That multiplier responds to demand inside the same min/max band: more registrations push the live burn upward, while half-life decay pulls it back down when activity slows.

Max burn therefore caps how far that upward stepping can climb. A subnet owner can raise the ceiling for their subnet, for example netuid 1, but the multiplier still operates only inside the documented hyperparameter bounds for that subnet.

References: Glossary: Register, Subnet Hyperparameters

Full Subnets Use Displacement Instead of Rejection

Official Mining documentation notes that when a subnet is already full, registration can displace the lowest-ranked existing participant through immunity and deregistration rules rather than simply failing. Max burn still bounds the TAO burn for that entry attempt, but it does not by itself decide which UID slot changes hands.

That keeps economic bounds separate from slot competition. Max burn limits how expensive the attempt can become; displacement rules decide whether a new registrant can take a slot from an existing neuron when the subnet has no empty positions.

References: Mining: Miner registration, Glossary: UID Slot

Further Reading

Topics SubnetsRegistration