For everyone in crypto, layer 2 scaling solutions like Arbitrum are not just an investment question. They are the basis for ongoing and future projects in DeFi and the NFT space. L2s will sustain DAOs and virtual worlds, but also support the entire creator economy and the Metaverse.
In the history of crypto, there has been three instances where gas fees on Ethereum spiked to astronomical heights: the bull run of 2017, after DeFi summer 2020 and NFT summer 2021.
Every time gas fees skyrocketed again, a more pressing discourse around the scalability of the Ethereum blockchain returned to center stage while more people rushed to find the next competing blockchain to invest in. Effectively, the meteoric gas fees that we have to deal with are only the tip of the iceberg. The reason why gas fees are so high today is because Ethereum has not been able to scale, meaning that there still is more demand for transactions than what the Ethereum blockchain can actually process at a time.
You may have the impression that the scaling debate is led by crypto masterminds and developers who make it hard to understand why exactly Ethereum is facing scaling issues, what scaling even means and what the stakes of this scaling race are for yourself and the crypto space. Here’s what’s going on.
Blockchains were created as a model to process transactions and record them in blocks to avoid double spending without the need of a trusted authority. Early blockchains like Bitcoin and Ethereum based their consensus model to process and record these transactions on proof of work.
In the proof of work model, once a transaction is initiated, it is placed in a pool of pending transactions and propagated to all nodes of the network. Although it is propagated to all nodes, only one miner who solves the transaction’s computational problem validates it by adding it to a block alongside other transactions, and is rewarded with transaction fees for mining.
Firstly, the fact that transactions have to be propagated to the network before and after their validation makes the processing time of transactions long. Then, the computational power required to solve mathematical problems consume huge amounts of electricity. Finally, the limited space for how many transactions can be included in one block creates competition between users and raises gas fees -the minimum price miners are willing to accept to process a transaction- to prices sometimes exceeding the value being transacted. This is increasingly becoming problematic as more dApps that leverage blockchain technology to process and store transactions are built on the mainnet (currently around 3000).
It follows that to resolve these issues, Ethereum must scale, by increasing the number of transactions that can be processed by the network (measured in TPS, transactions per second) as well as the transaction speed. To this end, a number of scaling solutions are already being developed, the main ones being Ethereum 2.0, competing Layer 1s, sidechains, and Layer 2s. While Layer 1s such as Ethereum 2.0, Polkadot and Solana, structurally change the underlying consensus and block rules of Ethereum, L2s are built on top of the Ethereum mainnet and its protocol.
The stakes are high. The few blockchains that will come out of this race as king will power the whole DeFi space, NFTs, sustain DAOs and virtual worlds, but also support the entire creator economy and the Metaverse.
We elaborate below on why we believe L2s like Arbitrum (in which Marc Cuban, Polychain Capital and Pantera Capital invested) will be leading Ethereum scaling over -at least- the five coming years.
As discussed above, scaling refers to increasing the transaction throughput (number of transactions that can be processed at a time) and transaction speed. Below a brief overview of the current existing options that aim to achieve this:
There are two ways to scale a blockchain ecosystem.
Layer 1 solutions are solutions that deal with the underlying protocol, which is the code of the mainnet blockchain itself, to increase the transaction capacity of the blockchain. They can be further classified into:
Protocol improvements are changes made to the underlying protocol to enlarge the transaction throughput, either by increasing the number of transactions that can be placed in a block (sustainable in the short term only), reducing the time difference between block-creation or through a structural shift from proof of work consensus model to proof of stake. Unlike in a proof of work model, proof of stake selects validators according to the amount of the blockchain’s native token they have staked. Since validators are chosen on this basis, no substantial computing power is needed and there is no competition between miners that drives gas fees to astronomical heights. Transactions take much less time to verify because individual nodes do not have to dedicate so much processing power as is the case for instance with PoS based blockchains Solana (50,000 TPS) and Polkadot (1000 TPS). Ethereum can only process 16 TPS as a matter of comparison.
Sharding divides the computational tasks and the data space of one blockchain into multiple chains (63 more with Eth 2.0). Sharding blockchain protocols intuitively means dividing nodes of the initial network into smaller groups, where each group is in charge of approving a distinct subset of pending transactions and of storing a subset of the global state. The way nodes are assigned to a shard is through a cryptographic sortition mechanism implemented by verifiable random functions, to avoid shard takeovers attacks if a shard comprises a majority of malicious nodes. Because transactions specific to a node are validated by it and not by the entire blockchain, there is no competition that leads to high fees and transactions are faster, which in turn increases the number of transactions that can be processed in a second. Sharding, alongside the PoS consensus model should enable 100000 TPS on Ethereum 2.0, according to Vitalik Buterin.
While layer 1 solutions alter the protocol of the blockchain, L2s are extensions to L1s enabled by a smart contract built on chain. They create additional space for transactions to be processed by outsourcing the execution of transactions before reporting back to L1. There are multiple ways to do this.
First, you can transfer assets to side chains on which transaction fees and speed are much more optimal (e.g. xDai uses a delegated proof of stake consensus mechanism, which allows for 5 seconds TPS at $0.000021 USD). Cross chain transfer of assets is enabled by a two-way peg (2WP) protocol that locks the assets on the first chain, then creates a transaction on the second blockchain whose inputs contain cryptographic proof that the lock was done correctly. A perfect example of this is Polygon.
Then, Plasma (link to the white paper by Joseph Poon and Vitalik Buterin) is a construction of blockchains in blockchains. Assets are sent to the smart contract managing the Plasma chain, which executes the transaction. Only blockheader hashes (used to identify a particular block on an entire blockchain) are submitted on the root chain unless there is proof of fraud, in which case the block is rolled back and the block creator is penalized. This ‘only if’ scheme brings tremendous scalability because minimizing the state updates in the root blockchain makes transactions faster.
Channels are open-source protocols and smart contracts that allow participants to transact x number of times off chain while only submitting two on chain transactions to the Ethereum network. Users have to create and pay for an Ethereum transaction when they first open up the channel. When they are ready to close the channel, they have to pay fees once again to process a transaction on the Ethereum blockchain. This decreases the number of transactions that must be processed and stored and reduces gas fees to only opening and closing a channel. The main projects leveraging state channels on Ethereum are State Channels, Celer, Perun, and Raiden.
Rollups scale the mainnet by rolling up transactions in a batch, using compression tools (e.g. scientific notation to reduce length of value in bytes) and verifying them off chain, before storing state data on L1. Compression and batching allows much higher throughput, speed and minimizes per transaction cost.
Then, there are two different types of rollups that depend on the verification method. Verifying transactions means checking whether their post-state root (meaning, the account balances, contract code, etc, that are "inside" the rollup) after they have been processed in the batches are correct. While in zk rollups a validity proof called a zk-SNARK is generated for every bundle, in optimistic rollups like Arbitrum, proof of computation is only performed if a node suspects that a transaction is fraudulent, thus further increasing transaction speed and throughput.
As a user, you would most likely interact with Arbitrum to use any dApp that you would also normally use on Ethereum and that requires transactions, (e.g. Uniswap or Aave). More specifically, you interact with Arbitrum by first adding the Arbitrum One network extension to your MetaMask wallet, connecting to it, and then bridging over assets.
Scaling is not enough. While increasing their transactional throughput, blockchains must preserve two fundamental properties of blockchain technology: decentralization and security. This is known as the blockchain trilemma. As of today, the only Ethereum scaling solution that satisfies all three elements is rollups like Arbitrum. The total number of transactions that have been processed on Arbitrum amounts to 3.56 million, with a maximum of 268k daily transactions reached on September 12, 2021.
In terms of processing capacity, Arbitrum should allow for 40,000 TPS and a 5x average cost advantage over using Ethereum at the base layer (currently $2 USD vs $10.38 USD on Ethereum, see live L2 fees). In addition to this, Arbitrum is actually working on further reducing fees by 90%-95%, meaning that you will be able to mint a NFT or transfer eth for only tens of cents. The total daily amount of gas fees spent on Arbitrum has also been consistently lower than on Ethereum. The only noticeable spike for September 12 is due to the launch of an L2 yield farm called ArbiNYAN.
Arbitrum achieves this high throughput while deriving its security from L1 consensus. As a matter of comparison, early L1s like Ethereum and Bitcoin prioritized decentralization and security but sacrificed scalability, as evidenced by the high gas fees today. Likewise, other competing L1s such as Solana and EOS sacrificed decentralization since only 150 and 21 nodes respectively can control their network. In turn, centralization also affects security because it increases the chances of a 51% attack. Similarly, sidechains could potentially bring attack vectors in the mainnet since they rely on consensus and block validation models of their own. Another example with limited security is protocol improvements because blockchains with bigger blocks are inherently more difficult to verify and likely to become more centralized, and in turn less secure. Note however that not all rollups are decentralized in their early days, although most if not all are committed to gradual decentralization.
Arbitrum is the one of the most EVM compatible layer 2 solutions as of today. This makes it almost trivial for developers to migrate existing Ethereum applications to rollups because they don’t have to rewrite most of the code. Arbitrum is the biggest Ethereum layer 2 network with over $2.3 billion total value locked in DeFi protocols. It already supports established dApps such as Uniswap, 1inch, Aave, Balancer, Curve, MakerDAO, Gnosis Safe etc. As a matter of comparison, ZK-rollups and payment channels only support simple payments, exchange and other application-specific use cases. Moreover, it currently has a total of 2361 verified contracts and has been adding on average 12 more contracts every day since September.
Not only is Arbitrum very attractive to developers but also consequently to users, as more dApps get integrated into Arbitrum rollup. This is evidenced by the increasing number of unique addresses using Arbitrum.
Among these addresses, smart money addresses seem to have been moving to Arbitrum. As shown below, 50% of ETH Millionaires that use Ethereum use Arbitrum too. For those who are not familiar with Nansen labels yet, ETH Millionaires are addresses with an ETH balance of at least US$1,000,000.
Another reason why we believe that rollups like Arbitrum will dominate Ethereum scaling in the coming years is that they are built on Ethereum, king in the crypto space. This gives them the first mover advantage. Firstly, Ethereum remains the most popular blockchain protocol in the world in terms of usage with more than 3000 dApps. It powers the whole DeFi ecosystem, NFT projects, DAOs and virtual worlds. Bitcoin is the only comparable blockchain in that regard but lacks the ability to host rollups.
Then, unlike what most people believe about Ethereum 2.0 making rollups obsolete, rollups like Arbitrum should be the first de facto scaling solution considering that eth 2.0 will only be fully deployed in years from now. Quite on the contrary, data sharding, the second step in eth 2.0 is focused on accelerating rollups. With rollups splitting the state stored, shards splitting the history of transactions and rollups registering themselves to specific shards, the scalability of rollups’ throughput is exponential and could reach up to 14 million TPS by 2030.
For these reasons, it is clear that rollups like Arbitrum are uniquely positioned to lead this wave of Ethereum scaling solutions in the near future.
Unless you are a developer or early stage investor, you can still get exposure by purchasing a L2 project’s native token. Although most rollups like Arbitrum do not have native tokens yet, we can expect most rollups to eventually release one.
The first critic is that the withdrawal period is long for optimistic rollups. It can take 7 days as withdrawals need to be delayed to give time for someone to publish a fraud proof and cancel the withdrawal in case a transaction is suspected to be fraudulent.
Then, the condition for optimistic rollups to be secure is that at least one node is honest and identifies fraudulent transactions.
In addition to this, rollups are still in their early stages and are not interoperable yet, although we can expect that moving assets and data across rollups will become increasingly easy. There are nonetheless side interoperability solutions like Hop, Connext, cBridge and Biconomy.
Finally, many argue that the liquidity of Ethereum is fragmented between different rollups, although since users are incentivized to extract the best price possible we could see over time liquidity accrue to rollups like Arbitrum.
Realistically, the potential of rollups in the long run is not limited to Ethereum. The future of scaling solutions is in fact a much more complex system of interdependent scaling projects on multiple L1 chains. In this complex future, we can nonetheless expect three main trends.
As mentioned above, in the short run, rollups on Ethereum such as Arbitrum will dominate the scaling solutions landscape and will then be further strengthened by the deployment of eth 2.0 and its sharding scheme. L2s -including but not limited to- rollups will show progress and become increasingly performant, thus competing with L1s when it comes to execution.
Next, when other competing L1s reach full capacity, they will start building rollups on top of their mainnet. In fact, what most people still fail to understand about the potential of rollups is that every L1 will need rollups. Ethereum was simply the first to have been preparing for it and for quite some time now (since 2015). For instance, Tezos is embracing a rollup-centric roadmap. Likewise for NEAR, Celestia and Polygon who just recently announced advanced zk-STARKs solution Polygon Maiden.
In reality, considering that the demand for Ethereum and blockchain technology is only going one way, we can also expect that all scaling solutions will be valid and necessary to meet increased demand from current and future market participants.
In conclusion, high gas fees were indeed just the tip of the iceberg. Ethereum has been facing scaling issues since its inception which a multitude of scaling solutions are aiming at tackling today. Among them, Arbitrum rollup, the current biggest Ethereum layer 2 network, but also the one with the greatest potential to dominate scaling solutions for years to come. Arbitrum rollup has proven capable of scaling Ethereum without compromising the decentralization and security of the mainnet. As users seeking to extract the best price possible will move to Arbitrum more developers will build on it because it is one of the most EVM compatible L2s. Beyond Ethereum, rollups is a paradigm that will be exported to other competing L1 and fuse in an increasingly complex landscape of scaling solutions.
For everyone in crypto, L2s like Arbitrum are not just an investment question, they are the basis for ongoing and future projects in DeFi, the NFT space; will sustain DAOs and virtual worlds, but also support the entire creator economy and the Metaverse.
Does this sort of data interest you? Arbitrum ecosystem dashboards are now live on nansen.ai!