Network profile

Avalanche: subnets, validators and app-specific chains

Avalanche sells customisation: a project can run its own chain and still borrow validation. The trade-off is complexity and a smaller shared liquidity pool.

Layer 1 · SubnetsChecked 8 Nov 2026

Avalanche sells customisation: a project can run its own chain with its own rules and still plug into a shared validator base. The price is complexity and thinner shared liquidity.

Native token
AVAX, for fees on the C-Chain and for staking on the Primary Network
Type
Layer 1 with a Primary Network of three chains, plus subnets that run their own rules
Consensus
Repeated sub-sampled voting (Snowman family), not a longest-chain rule
Fees
Dynamic gas fee in AVAX on the C-Chain; a subnet can charge in its own token
Settlement
Self-settled; subnets inherit security from their own validator set, not from Ethereum
Ecosystem
DeFi, gaming chains, institutional deployments, tokenised assets, custom VMs
Explorer
snowtrace.io
Documentation
docs.avax.network

How it works

The Primary Network is three chains with different jobs. The X-Chain handles asset creation and exchange with a UTXO model. The C-Chain runs the EVM and is where almost all activity happens. The P-Chain coordinates validators, staking and the registry of subnets. They are validated together, and AVAX is the token that pays for all three.

Consensus is not proof of work and not classic proof of stake either. Validators repeatedly sample small random subsets of the validator set and ask them about a transaction; each round increases confidence until the probability of reversal is negligible. This gives fast, probabilistic finality without waiting for a fixed number of blocks, and it means the security argument is statistical rather than «the chain with most work wins». The trade is that the model depends on enough validators actually responding, and Avalanche leans on staking rewards and uptime incentives more than on slashing as a deterrent.

A subnet is a defined set of validators running a chain, with its own virtual machine, its own fee token and its own membership rules. A project can require KYC validators, or a large stake, or a private set. That is genuinely useful for regulated deployments and for games that want their own blockspace. Relaxing the requirement to also validate the Primary Network changed the economics of launching one, and what a subnet inherits: security now comes from whoever validates it.

So «on Avalanche» can mean three quite different things: the C-Chain, a subnet with a large validator set, or a subnet run by a company for its own application.

Fees: the mechanism

On the C-Chain, fees are EVM gas with a dynamic base price: the protocol adjusts the minimum gas price according to how full recent blocks were, and you add a tip. Because the adjustment is gradual, fees do not spike the way Ethereum's do; they drift. During sustained demand they climb and stay up until usage falls.

Two practical notes:

  • Fees are paid in AVAX on the C-Chain. A subnet can charge in its own token, so «cheap Avalanche fees» may not apply to whatever chain you are actually using.
  • Subnet fees are a design choice, not a protocol constant. A private chain can subsidise transactions to zero or price them to ration access.

There is also a fee for creating a subnet and for staking, which is a barrier to entry measured in AVAX rather than in gas, and it moves with the token price.

What is actually built there

DeFi on the C-Chain is real and has periods of strong activity, usually driven by incentive programmes. Lending, DEXs, stablecoin liquidity and structured products exist, and the ecosystem is EVM-compatible enough that teams port quickly.

The distinctive work is elsewhere: gaming subnets, institutional and enterprise chains with permissioned validator sets, and tokenised real-world assets where the operator wants control over who validates. Those are use cases where «your own chain with your own compliance rules» is the requirement, and a general-purpose layer 1 cannot offer it.

Custom virtual machines are the most interesting long-term capability. A subnet can run non-EVM execution, so the chain can be shaped to the application.

Risks and limits

  • Liquidity fragments. Every subnet is a separate pool; value does not automatically compose across them.
  • Subnet security is local. A small validator set is a small security budget, whatever the parent network's reputation.
  • Complexity is a real cost. Running a chain, a bridge and a validator set is an order of magnitude more work than deploying a contract.
  • The deterrent model is unusual. Rewards and uptime incentives do the work that slashing does elsewhere; read the current rules rather than assuming.
  • Launch costs are denominated in AVAX, so the barrier moves with the market.
  • Marketing outruns measurement. «Subnets» and «Avalanche L1s» name an architecture, not a level of adoption.

Official sources