AVAX is the capped supply crypto-asset of the Avalanche network, which is an open source platform based on Proof of Stake which will digitise assets and build a new foundation for finance.
Background and History
The Avalanche network and the AVAX token were created by Ava Labs, and built on the work of Team Rocket’s novel consensus protocol known as Snowflake. Ava Labs was founded by Cornell University professor Emin Gün Sirer, and Cornell University computer science PhD’s Kevin Sekniqi and Maofan ‘Ted’ Yin.
Gün Sirer’s background is interesting, having attended Princeton University on a scholarship, attaining a PhD in Computer Science and Engineering in 2000, and becoming Assistant Professor at Cornell University in 2001. The Ava Labs founder’s first foray into cryptocurrency was around 2002-2003. Along with Vivek Vishnumurthy and Sangeeth Chandrakumar, he contributed to a conceptual peer-to-peer virtual currency called karma six years prior to Satoshi Nakamoto’s release of the Bitcoin whitepaper.
While karma was the first cryptocurrency to use Proof of Work for minting, it did not merge the minting of new supply with the consensus protocol as Bitcoin’s founder did, which became the innovation known as the Nakamoto consensus protocol. Furthermore, the timing of karma was a bit unfortunate, as it emerged in the aftermath of the 9/11 terror attacks, so the prospects of getting funding to pursue the idea further were slim to none and it was likely the state would clamp down on a peer-to-peer digital currency.
Emin Gün Sirer has become a well-known figure in the cryptocurrency scene being part of the cypherpunk movement, the cryptography mailing list where Bitcoin was first presented to the world and was one of the earliest academics that became involved in Bitcoin and Ethereum. His research can be found on Hacking Distributed and is a key member of the IC3 (the Initiative for Cryptocurrencies and Contracts). Gün Sirer studied Nakamoto’s invention in the early years and looked for flaws in the protocol.
This resulted in the Cornell professor discovering one of the biggest flaws in the Bitcoin protocol, known as ‘selfish mining’ and helped to come up with a solution to address this flaw. Gün Sirer is also known for his work on bitcoin covenants and vaults, as well as on scaling solutions such as Bitcoin-NG. On top of his work related to Bitcoin, the Avalanche founder also earned a reputation within the Ethereum community for his research, the most notable perhaps being on the exploits in The DAO before the momentous hack in July 2016. Later on, he founded bloXroute Labs, a team of crypto-researchers and software developers aiming to solve the challenge of blockchain scalability
Before the cryptocurrency market starated to overheat in the 2017 bull run, Gün Sirer collaborated with other academics to assess how decentralised Bitcoin and Ethereum really were, and quantified their evolution and progress. Gün Sirer presented these findings at the Genesis London Conference in February 2018. One of the concluding remarks of this paper was, “further research is needed to decentralise permissionless consensus protocols” which eventually led to the creation of Avalanche several years later.
Avalanche is often described as ‘Blockchain 3.0’, an iteration and improvement on Blockchain 1.0 and 2.0 (i.e., platforms like Bitcoin and Ethereum respectively). The difference is that the first two generations relied on Proof of Work, which comes with several possible drawbacks such as long confirmation times as you wait for miners to produce a new block.
Instead of a single set of miners (or validators), Avalanche is a heterogeneous network of many blockchains and validator sets. By implementing a new consensus protocol, it achieves finality in a few seconds, and provides the same security guarantees as waiting for the recommended number of confirmations with Bitcoin or Ethereum.
An anonymous team called Team Rocket originally proposed the Snow family of consensus protocols in May 2018 and provided a correctness proof to show it could work. It combines the best of both Classical and Nakamoto consensus. While the Snow consensus protocols are permissionless, there is no mining or high energy expenditure as there is with Proof of Work systems.
Prior to the introduction of the Snow family of consensus protocols (which includes Snowflake, Snowball and Avalanche), there were two main consensus mechanisms: Classical consensus and Nakamoto consensus. With classical consensus such as Practical Byzantine Fault Tolerance, which is being used by private, permissioned distributed systems, transactions finalise instantly. However, it does not scale well with the number of nodes taking part.
The Snow family exhibits the high throughput of classical consensus mechanisms, but with a much lower communication overhead. Snow does for distributed systems and consensus what gossip protocols did for data dissemination, by relying on random interactions, fusing a Directed Acyclic Graph (DAG) structure and repeated sub-sampled voting. This ensures that, at the end of a series of interactions, everyone has decided on the same thing.
With this breakthrough in consensus mechanisms, the aim of Avalanche is to replace Wall Street and build a new foundation for finance, providing a platform that can scale and is secure. Digital assets that can be traded anywhere around the world can be created and integrated with the existing financial system. With Avalanche’s subnetworks (or subnets), a network can be configured to a particular jurisdiction, permitting the creation of regulatory compliant assets.
Some major milestones for the Avalanche network and AVAX token are displayed below:
- May 16, 2018: An anonymous group known as Team Rocket released a paper titled “Snowflake to Avalanche: A Novel Metastable Consensus Protocol Family for Cryptocurrencies” on IPFS.
- June 2018: Ava Labs founded by Emin Gün Sirer, Kevin Sekniqi, and Ted Yin.
- February 1, 2019: Seed sale announced. In the seed sale, 18 million AVAX tokens were sold at a price of $0.33 per token with one-year vesting. Seven investors took part, including Polychain, Andreessen Horowitz and Balaji Srinivasan.
- June 21, 2019: Emin Gün Sirer, in collaboration with Team Rocket, published a revised version of the original paper.
- April 2020: Cascade: first public testnet launch.
- May 2020: Private sale: 24.9 million AVAX tokens were sold at a price of $0.50 per token. Raised $12 million from seven investors including Bitmain, Dragonfly Capital Partners, Galaxy Digital, and NGC Ventures.
- May 2020: Denali incentivised test network launch.
- June 25, 2020: Initial Coin Offering for Avalanche’s native token (AVAX) announced.
- July 15, 2020: Initial Coin Offering raised $42 million in under 24 hours.
- August 2020: The Everest test network, a fully featured version of the Avalanche network, was launched.
- September 21, 2020: Avalanche Mainnet Launch.
- September 22, 2020: AVAX price reached an all-time high of $6.98.
- January 2021 (Anticipated): First major upgrade to Avalanche, Apricot.
- March 4, 2022: All AVAX tokens from Public Sale unlocked.
In the months leading up to the Avalanche mainnet launch, the rise of Decentralised Finance (DeFi) saw confirmation times and gas prices on Ethereum reach record highs, limiting participation in open finance to wealthy individuals and large organisations.
DeFi applications based on Ethereum can switch to Avalanche easily, alleviating blockspace demand, congestion and fees. With a throughput of 4,500 transactions per second and security thresholds are above the 51% standards of other networks, Avalanche is well positioned to complement Ethereum as a smart contracts platform.
Just a few months after mainnet launch, the DeFi ecosystem on Avalanche is small but growing, including projects such as non-custodial prediction market Prosper and TrueUSD, the first stablecoin to launch on the network. In 2021, the focus will be on DeFi applications such as stablecoins, lending/borrowing platforms, decentralised exchanges and so on.
Economics of AVAX
The Avalanche (AVAX) token is the native token of the Avalanche platform and is used to secure the network through staking, transact peer-to-peer, pay for fees, and provide a basic unit of account between the multiple subnetworks created on the Avalanche platform.
Demand for AVAX
There are several sources of demand for AVAX:
- Payments: AVAX can transfer value at speed and at scale using the permissionless Avalanche network.
- Staking: users can set up a validator or delegate their tokens to stake and earn AVAX tokens as passive income.
- Utility: Avalanche is a platform for open finance (DeFi, NFTs, stablecoins, etc.), the creation of digital assets and subnets requires AVAX tokens.
Nodes stake AVAX tokens to show that they have some ownership in the system. To stake, a special transaction is sent to the network that locks a minimal amount of AVAX (at least 2,000 AVAX). If the threshold is met, you automatically become a staker. Figure 1 shows that most validators are currently located in the United States and Germany.
The rewards earned by nodes depend on the amount staked, the length of staking, uptime and response latency. The minimum percentage of the time a validator must be correct and online in order to receive a reward is 60%. If you have less than 2,000 AVAX tokens, you can delegate to stake with another node, where the minimum amount is 25 AVAX.
Figure 1: AVAX nodes by country, Jan 2021
Unlike Ethereum and similar platforms, the staked assets (i.e., AVAX) do not serve as collateral and can never be lost by validators. There’s no risk of money being taken from validators, which makes staking more predictable and encourages wider participation. The purpose of staking is ensuring that you cannot impersonate other people, as a protection against a Sybil attack. Staking is completely independent of consensus.
For transactions on the Avalanche network, the fees paid in the native AVAX token are permanently burned, instead of going to miners (or validators in this case) as with cryptocurrencies that depend on Nakamoto consensus. Therefore, as the number of transactions rises and more value is transferred, the available supply is reduced to benefit the ecosystem as a whole, instead of one set of players in the ecosystem.
Another difference to most blockchains is that the fees are not set by the issuer of a transaction, but by a globally verifiable fee function. While simple transactions will have the lowest transaction fees, there will be higher fees for other uses of Avalanche, such as setting up subnets - which commands the highest fees. Individual subnets can issue their own tokens, but AVAX tokens are required to set these up.
Avalanche can be thought of as a safety net while the transition to Ethereum 2.0 takes place ove rthe next few years. DeFi applications and stablecoins are already springing up on Avalanche as compatibility with the Ethereum Virtual Machine (EVM) allows the platform to capture Ethereum’s network effect. Much of the DeFi ecosystem could pivot to Avalanche if congestion becomes a major issue on the Ethereum.
Supply of AVAX
AVAX is a scarce crypto-asset with a total supply cap of 720 million tokens, of which 360 million were brought into circulation at genesis.
The rate at which new coins are created is a function of how much AVAX is staked and ensures the asset reaches a capped supply at a slow and steady rate. The reward rate is not a fixed parameter, since token holders can determine the emission rate. The emission rate depends on several factors: percentage of the total supply staked by the node, the duration of the stake (using a minimum of two weeks and a maximum of one year), node uptime, and node latency.
Figure 2 illustrates the emission rate under two scenarios: if 100% of AVAX are staked, where it follows an emission curve similar to bitcoin’s, and if the staking ratio is only 50%, the supply grows at a slower rate and it will take much longer to reach the 720 million coin limit (the staking ratio is currently just below 70%).
Figure 2: AVAX Emission
Source: AVAX Token Whitepaper
Of the total supply, half of the tokens will be released over decades and allocated to validators as staking rewards. Staking returns are estimated to be between 7-12% in the first year. The default value for staking is 2,000 AVAX, and the minimum time for a node to stake is two weeks. Along with the emission rate, the staking amount and times are also governable.
Staking AVAX currently provides an annual reward of 11.57% and an adjusted reward of 3.03% according to stakingrewards.com if you run a validator for 14 days. The maximum rate is earned by staking for a year, while the lowest rate is for 14 days. According to Avax explorer, there are 251.9 million AVAX tokens being staked, representing 69.98% of the 360 million tokens released at genesis block.
To offset inflation from staking rewards, transaction fees are burned and AVAX tokens are permanently removed from the circulating supply.
Another supply consideration is the tokens sold in seed sale, private sale, public sale and so on that will become liquid in the next four years. The token sale opened on July 15, 2020 and the ICO sold all tokens in less than one day, raising $42 million. The private sale that took place beforehand raised $12 million.
Figure 3 shows the distribution of AVAX tokens:
Figure 3: Distribution of AVAX
- The Seed sale accounted for 2.5% of the total supply. Ten percent of allocation was released on main net launch, while 22.5% will be released every three months.
- The Private sale accounted for 3.5% of the token supply. As above, ten percent of the allocation was released on mainnet launch, then 22.5% is released every quarter for one year.
- For the Public sale, there were three options:
- Option A1: 1% of the token supply distributed to public sale participants. Ten percent was released on mainnet launch, and 22.5% is released every quarter for one year.
- Option A2: 8.3% of the token supply distributed to these public sale participants. Instead of a one-year vesting schedule as with the seed, private, public sale option a1, there is a 1.5 year vesting. 10% of the allocation was released on the main net launch, while 15% is released every three months over a period of 18 months,
- Option B: 0.67% of the token supply with no vesting period.
- Foundation: 9.26% of the tokens allocated to the foundation with a 10-year vesting period. These funds will be used for building the ecosystem, such as bounties, marketing and so on.
- Community and development endowment: 7% grants have a one-year vesting period from the date grant is awarded. Tokens are allocated to individuals and groups that are developing core tooling and infrastructure on Avalanche, as well as grass roots community building and marketing.
- Testnet Incentive Program: 0.27% of the total supply, where tokens locked for one year. Participants of the Denali testnet were able to complete challenges to earn up to 2,000 AVAX.
- Strategic partners: 5% of token supply with 4-year vesting period. These tokens are allocated with the specific mandate of being distributed to groups, organisations, and enterprises that are building businesses using the Avalanche technology and network.
- Airdrop: 2.5% of supply with four-year vesting period. These tokens are to be distributed to various communities in order to onboard more people to the Avalanche community. For example, these may include airdrops to various crypto communities.
- Team: 10% of the token supply given to founding and non-founding members of Ava Labs, four-year vesting period.
Of the total supply of AVAX tokens, 16% were sold in seed, private and crowd sales, and most of these tokens have a strict token unlocking schedule. You can check the dates of unlocking schedule here.
Figure 4 shows the proportion of tokens that are locked/unlocked for the allocations with no vesting periods, or vesting periods of 1.5 years and less. The next unlocking of tokens will be on March 9, 2021. Avascan shows the number of tokens that are time-locked here. There are currently 65.2 million AVAX tokens (out of 360 million generated at genesis) that are not time-locked.
Figure 4: Locked/Unlocked AVAX tokens, Jan 2021
In this section, we summarise the Avalanche platform and the novel Snow family of consensus protocols.
The Avalanche platform comprises three main components: two blockchains (the C-Chain and P-Chain) and a Directed Acyclic Graph (DAG) known as the X-chain. The DAG structure contrasts to the traditional blockchain structure where transactions are organised chronologically. Instead, a DAG links transactions with other transactions so you do not have to wait for them to be bundled together in a block.
All three components (P-Chain, X-Chain and C-Chain) are validated and secured by the Primary Network. All validators are members of the primary network, which is a special case of a sub-network. Snowman is a chain-optimised consensus protocol with high-throughput and well suited for smart contracts. Snowman is powered by the Avalanche consensus protocol and is implemented by both the P-Chain and C-Chain.
The architecture of Avalanche is displayed by Figure 5:
Figure 5: Avalanche Overview
Platform Chain (P-Chain)
The P-Chain is where tokens are locked for staking and coordinates validators to keep track of subnets and make other subnets possible.
The P-chain is also responsible for implementing the Snowman consensus protocol, which achieves chain-optimised consensus, with high throughput and is best for smart contracts. All transaction fees are paid in AVAX which are burned.
Exchange Chain (X-Chain)
The X-Chain is a decentralised platform enabling the creation of new assets, the exchange of assets, and transfers between subnets, as well as being an instance of the Avalanche Virtual Machine (AVM). This component handles the implementation of the Avalanche consensus protocol, which achieves DAG-optimised consensus with high throughput and is simple to prune.
The X-Chain enables anyone to create or mint other smart digital assets, such as stablecoins, utility tokens, NFTs, equities and so on, which commands a fee paid in AVAX. The X-Chain is a UTXO-based system which should be familiar to anyone working on Bitcoin, Litecoin, Bitcoin Cash and so on.
Contract Chain (C-Chain)
The Avalanche Contract Chain (or C-Chain) is an implementation of Ethereum Virtual Machine (EVM) that allows developers to port over Ethereum applications seamlessly, such as the key tooling that has fuelled DeFi’s growth to date, for example, MetaMask and Web3.js.
Using the C-Chain’s API, current Ethereum applications can pivot to Avalanche to take advantage of higher throughput, quicker finalisation, extremely low gas fees and Solidity tooling. All gas fees are paid in AVAX which are also burned.
What are Subnets?
Subnets (short for sub-networks) are a dynamic set of validators to come to consensus on the state of several blockchain networks. These subnets can create their own custom consensus mechanisms, which can be Proof of Work or Proof of Stake, and their own incentive scheme for validators. The primary network is the default subnet which validates the AVAX token’s subnet, i.e., the X-Chain.
To reach Avalanche’s goal of the ‘Internet of Finance’, subnets are a crucial component. Anyone can create tailor-made networks with custom virtual machines (VMs) and validator rule sets, permitting the creation of a spectrum of different networks, from permissionless to permissioned. Developers can use any virtual machine, such as BTC Script, EVM, WASM or any other VM they wish. The ability to move value between subnets will lead to compounding network effects for successful dapps, blockchains and subnets.
Athereum is a subnet utilising the Avalanche consensus engine but ported over the Ethereum state at a yet-to-be determined block height to provide an alternative environment to run for Ethereum developers to run their decentralised applications at a higher throughput and with faster finality. When the Ethereum state is eventually ported to Avalanche, all existing holders of ETH will also have an equal amount of ATH, the native token of Aethereum.
The family of Snow protocols are leaderless Byzantine Fault Tolerant, which eliminates the requirement for staking pools since all validating nodes are rewarded proportionally for their services to the network.
To summarise Avalanche’s consensus mechanisms, let’s look at the life cycle of a transaction.
Validators confirm whether a transaction is spendable with its VM by asking a small, random subset of validators. If the validators part of the sub-sample think the transaction is invalid or they prefer a conflicting transaction, they will reply that they think it should be rejected.
A set of non-conflicting transactions is then created and proposed to the network. Once proposed, a process of repeated random sub-sampling occurs, where a number of validators are selected at random (weighted by their stake) and are queried on their preferred transactions.
If confidence in the transaction is high enough and the repeated random sub-sampling results in a supermajority response in favour of the transaction, then the node will change its own response in favour of that transaction. This is repeated for every node in the network until there is consensus on the conflicting transactions. If there is little confidence in the transaction, then another K random validators are selected and weighted on stake.
Where a transaction has no conflicts, finalisation happens very quickly in a few seconds. When a conflict exists, honest validators quickly cluster around conflicting transactions, entering a positive feedback loop until a supermajority of nodes prefer a particular transaction. This leads to the acceptance of non-conflicting transactions and the rejection of conflicting transactions. If any honest validator accepts or rejects a transaction, there is a high probability that all honest validators will accept or reject that transaction.
One of the major advantages of the Snow family of consensus protocols is that it is not susceptible to 51% attacks like the leading Proof of Work cryptocurrencies. In fact, at least 80% of the nodes in the Avalanche network must collude to break consensus.
Governance is built into Avalanche (expected to be implemented in early 2021).
The polling mechanism used for consensus, which is inspired by gossip protocols, can also be used to agree on critical parameters of the protocol. By asking participants, we can find out what the ecosystem wants. For example, changes to the minting rate, staking amount or staking time, all of which can be increased or reduced depending on the state of adoption and of the market.
While the network can support thousands to millions of block producers and there are already 679 validators since launch, Ava Labs control ten percent of AVAX’s total supply, amounting to 72 million tokens, and a further seven percent of the total supply (50.4 million AVAX) is to be distributed to developers and the community.
The challenge is distributing these tokens to ensure Avalanche is sufficiently decentralised so that any node does not account for more than one percent of the network. The Avalanche-X grant program intends to address this challenge by distributing AVAX to builders and developers within the ecosystem.
The upcoming Apricot upgrade will introduce pruning to reduce the sync time for an Avalanche node, which should improve the decentralisation as it becomes more accessible to earn AVAX.
Privacy subnets are being worked on and will be available soon.
One potential improvement to scaling is the implementation of Frosty consensus, another member of the Snow family of consensus protocols. Both Avalanche and Snowman consensus protocols are leaderless, introducing the possibility of block contention. When there is disagreement amongst nodes in the network, the performance of the Avalanche/Snowman consensus protocols is reduced somewhat. The Frosty consensus engine introduces a leader and improves the performance in scenarios where there is block contention.
The upcoming Apricot upgrade will also help improve scalability by introducing epochs, verifiable pruning and governance. Epochs introduce the notion of time into the DAG X-Chain, where many nodes can extend the chain in parallel. Epochs make it possible to archive large parts of the X-Chain. Once activated, users will be able to run nodes that consume less space, where the space consumed is proportional to user activity.
Another potential scaling improvement relates to databases. The Avalanche consensus protocol relies on LevelDB and is a database-bound process. Scaling of the network can be improved by replacing LevelDB with CedrusDB, as well as through optimisations.