TLDR

• Avalanche is horizontally scaling through subnets. Subnets are a subset of validators from the Avalanche Primary Network and can contain one or multiple customizable blockchains with different VMs. Subnets benefit from Avalanche’s high performance.
• Subnets must incentivise a validator set from the Avalanche Primary Network to secure the subnet. This validator set can be permissioned, semi-permissioned, or permissionless. This means subnets can be designed so that validators pass KYC/AML checks and comply with regulatory requirements from a certain jurisdiction, for example.
• The primary use case for subnets at the moment is gaming, with DeFi Kingdoms and Crabada establishing subnets.
• Ultimately, subnets will need to bootstrap and ensure their security through their validator set. DeFi Kingdoms and Crabada are carefully vetting their validator set so that malicious actors do not drain funds. It will be up to the subnet if and how they want to scale their validator set and thus their level of decentralization.
• Subnets are isolated from the rest of the Avalanche ecosystem. A general mechanism for cross-subnet communication has not been developed yet.

Introduction

Avalanche burst onto the scene as a high-performance alternative layer-1 blockchain. Users delighted in its lucrative farming opportunities, fast transactions, and ultra-cheap fees. However, during periods of hyperactivity, the network started to show some strain, becoming slower, with failed transactions and higher fees. This is the problem blockchains encounter as they reach capacity.

Architecture Overview

Avalanche has subnets to address this scalability issue and ensure high performance. Subnets are a way of horizontally scaling the blockchain. Instead of popular dApps clogging up the mainchain (C-Chain) and rendering it near unusable, these applications can launch their own subnet (which may contain one or multiple blockchains) that can be optimized for their specific use case. This enables these applications to ensure high performance while still remaining a part of the Avalanche ecosystem. Crabada, for example, accounted for 63% of all activity on Avalanche C-Chain over the last 7 days (20-27 April), making transactions more costly as a result. This high activity makes operating on a subnet more optimal - and Crabada will be moving to its own subnet.

The Avalanche Primary Network is a special subnet comprising 3 blockchains:

• X-Chain: (Exchange chain) for creating and exchanging digital smart assets e.g. a representation of a real-world resource like equity and bonds, with a set of rules that govern its behaviour.
• P-Chain: (Platform chain) enables the creation of subnets, coordinates validators, and manages metadata of the entire Avalanche ecosystem.
• C-Chain: (Smart contract chain) EVM-compatible chain where the majority of Avalanche applications are built on e.g. Platypus, Trader Joe.

To learn more about Avalanche generally and its adoption, check out Nansen’s Q1 2022 report on Avalanche here.

What are subnets?

A subnet is Avalanche’s answer to scaling similar to Polkadot’s parachains, Cosmos’ Zones, and Ethereum rollups. It is a set of validators that are also validators on Avalanche’s Primary Network.

• Each subnet can have multiple blockchains
• A blockchain can only be validated by one subnet
• Validators can validate multiple subnets.

Subnets are Avalanche’s way of horizontal scaling, spinning up more chains = more scalability.

At present, it is necessary to stake at least 2000 AVAX (~$140,000 at the time of writing) on the Primary Network to be a validator. Subnets can tap into this validator set.

• Subnets are required to incentivize their own validator set. They can do this by whatever means they see fit (most will use their native token).
• Subnets are free to choose which virtual machine to build their blockchain with. This can be the EVM, Avalanche Virtual Machine (AVM), or any other custom VM the blockchain desires.
  • There can be multiple VMs within a subnet
• In addition, subnets can choose which token(s) are required for transaction fees, including their own token (which could drive additional value to it).
• Subnets may be public or private blockchains. In addition, subnets can have requirements for validators such as KYC/AML and even hardware.
• Subnets are by default isolated from the Avalanche Primary Network and other subnets but can be interoperable. However, the tooling around this is in its infancy, and trusted bridges will be required for interoperability across Subnets.

• That said, interoperability between chains within a subnet does not require a bridge.¹ It is worth noting that subnets at present are whitelisting their validators. This will likely be the most common approach for subnets in general.

• If becoming a validator on a subnet is permissionless, it is more exposed to malicious validators entering and attacking it.
• Subnets will need to be careful who their validators are. For example, an issue that could arise where a subnet with 5 validators has 3 validators maliciously colluding.
• Ultimately, it is up to the subnet to determine the validator set and the level of decentralization. Right now, strict control over who is permitted to be a validator is probably important and the approach that current subnets are taking.

Incentives

Avalanche has launched an up to $290m Incentive Program to incentivize the adoption and growth of subnets, across all ecosystems including gaming, DeFi, institutional, NFTs and more.

Benefits

Subnets are customizable

• This is important for optimizing the chain for the specific use case as generic VMs like the EVM incorporate various design decisions that are not optimized for specific use cases. This differentiates Avalanche from other EVM alternatives that have displayed little innovation (Vitalik Buterin once commended Avalanche’s tech).

Barriers to entry are (relatively) low

• A subnet can be created by burning 1 AVAX.
• A subnet is permanent. This can be seen as advantageous to Polkadot, which requires parachains to compete in candle auctions every 2 years for a slot.
• However, subnets need to bootstrap a validator set from the Primary Network. As previously stated, subnet validators are required to stake at least 2000 AVAX.

Interoperability

• For blockchains within a subnet, a general cross-chain mechanism exists, as demonstrated in the Primary Network with the interaction between the C-Chain and X-Chain. However, custom VMs would need to implement this. Further, smart contracts can not access it (given it is outside of the EVM).
• EVM subnets should be able to communicate using trusted bridges.
  • There is no stated method for subnet-to-subnet interoperability yet. It has been speculated that the Avalanche Primary Network may act as a bridge between subnets, although this is not confirmed.
  • Subnet-to-subnet interoperability is problematic due to no shared security and various trust assumptions required.
• IBC adoption speculation. There is no official confirmation about this, however, Kevin Sekniqi (COO and Chief Protocol Architect) arguably hinted that this is something Avalanche are considering/planning for.

Avalanche Consensus

Tapping into Avalanche’s fast consensus is a key benefit of being a subnet. This enables very high performance through high TPS and fast finality.

• Each validator randomly selects a number of other validators to query for a preferred decision. The querying node will then update its answer with the majority of responses.
• Therefore, one validator does not query every validator (like in classical consensus). This enables Avalanche to scale their validator set without significantly increasing the burden on each node as they are doing the same amount of work.
• The process is leaderless - making attacks very difficult, even if a significant proportion of nodes are malicious.

Use cases

Gaming

• Gaming is quickly becoming the most prevalent use case for Avalanche subnets. In addition to enabling far higher performance, subnets (and app-specific chains in general) give projects far greater control over the game economy as the native token is used for transactions etc. In addition, the subnet can be optimized for the game - rather than generalistic design tradeoffs. A key example is DeFi Kingdoms.
• DeFi Kingdoms (DFK)
  • DFK was originally a Harmony One blockchain game but has since expanded to Avalanche with the launch of its own subnet.
  • JEWEL is required to pay for transactions on the subnet, giving the token additional utility and demand. Establishing a subnet can enable DFK to cultivate a larger gaming ecosystem - something that would not be possible on a mainchain.
  • DFK is the recipient of Avalanche’s Multiverse incentive program.
  • Check out their subnet here.
• Crabada
  • Crabada has accounted for ~65% of all traffic on Avalanche over the past week.
  • It is migrating to its subnet Swimmer Network which Crabada will say will reduce costs for users by ~85%.
    • If successful, this will be a powerful example of the benefits of subnets.

Institutional Applications

• Institutional/enterprise adoption of subnets could be a very interesting growth area. Subnets can have full control over their validators and enable KYC/AML requirements and private blockchains that will allow enterprises to work within the remit of regulations while taking advantage of the benefits of blockchain technology and being part of the Avalanche ecosystem.

Security

• Subnets do not receive shared security like Polkadot, which enables them to be limitless in the number of subnets launched. Each subnet can decide and execute their own level of security.
  • Subnets will have varying security assumptions (similar to Cosmos Zones). All security is not created equal.
• Subnets need to incentivize Avalanche validators to validate their subnet. Subnets have complete control over how they would like to incentivize validators. This will likely be the native token in many cases.
  • Validators (who validate multiple subnets), will therefore be able to receive revenue streams from multiple tokens, enhancing their APR. This is important, as the more validators each subnet receives, the greater the security.
• That said, subnets are not optimized for security (although they can still attain it). Subnets introduce some centralization risk. Subnet creators can add malicious nodes and remove benevolent validator nodes to gain control of the subnet. This risk will be greatest when the creator is anonymous. Doxxed and well-known administrators will have a much lower risk.
• It will be interesting to see how pertinent these risks are in practice, and what the Avalanche ecosystem does to address/mitigate them. It appears that although it is relatively easy to set up and deploy a subnet, those that do will require the necessary resources to ensure its security.
• It is likely that there will be a marketplace(s) for matching validators with subnets.

How Important Is Decentralization?

Subnets potentially make Avalanche highly scalable, while they themselves sacrifice decentralization to achieve this. However, the importance of maximal decentralization is arguably overstated by much of the crypto community. While certain applications are better decentralized, many do not require this high level of decentralization. If certain projects have the resources to ensure that their subnet has sufficient security, then they present an excellent means of scaling the Avalanche network and the benefits of blockchain technology.

Quick Comparison

AVAX Tokenomics and Value Accrual

One criticism of layer-2s is that they are parasitic to the main chain. Subnets themselves can customize which token is used for gas fees (which will likely mostly be their native token). However, subnets have been designed to benefit AVAX, as validators are required to stake AVAX. If subnets become popular this will drive additional demand for AVAX.

All transaction fees on Avalanche are burned, and there is a capped supply of 720m AVAX. This can be a compelling narrative, as ‘competing’ ecosystems such as Polkadot and Cosmos have an inflationary supply.

Key Takeaways

• Subnet’s high performance combined with Avalanche’s $290m subnet incentive program will likely see some adoption of subnets over the coming year, especially in gaming.
• Subnet’s potential for enterprise and institutional applications is particularly interesting. This is because the validator set can be permissioned, enabling projects to select their validators and implement KYC/AML requirements etc while benefiting from Avalanche’s high performance.
• Ultimately, subnets do not offer a one size fits all solution, and like parachains and Cosmos Zones certain tradeoffs have been made. The key tradeoff made by subnets is higher performance but lower security due to lesser decentralization guarantees (that’s not to say that subnets cannot be satisfactorily secure). It is ultimately up to the project to determine if the subnet model is the correct fit for them and if they can satisfactorily secure their subnet.
• Only projects with the necessary resources should consider a subnet. Necessary resources include the ability to bootstrap a non-malicious validator subset, and the ability to incentivize this validator set indefinitely and potentially scale it safely (if they desire). Most projects will fall short of this, however, they may still want to establish a blockchain within a subnet or launch their project within a blockchain within a subnet.
• The ease of interoperability within subnets and between subnets will be important to their long-term success. It will be interesting to see how this develops.

Notes

A general cross-chain mechanism exists, as demonstrated in the Primary Network with the interaction between the C-Chain and X-Chain. However, custom VMs would need to implement this.