The Layer 2 landscape continues to evolve, with the past few months bringing improvements to the underlying technology and new rollups into the sector. Overall, the amount of Ethereum gas consumed by rollups hovers around 9%, showing signs of strength over the past few months as new rollups compete for their share of users and liquidity. L2 gas consumption reached a new all-time high of 12.4% on July 30th due to an uptick of activity on Linea and zkSync. 

The two primary L2s dominating the narrative continue to be Arbitrum and Optimism. These rollups have captured the largest TVL at ~$1.7B and $718M, respectively, with the closest rollups being Base at $193M and zkSync at $120M. Arbitrum and Optimism also generate significant sequencer profits, netting over $10.4M for the Arbitrum DAO and $5.7M for the Optimism Collective, respectively. Despite differences and volatility in revenue, both major rollups have begun to converge upon the same P/S valuation multiple, currently hovering near 200x.

New Layer 2 Launches

Base

Coinbase’s new L2 network, Base, publicly launched on August 9, 2023, at 12 p.m. ET. However, onchain sleuths discovered the L1 bridge contract before the official launch, allowing anyone to bridge their funds to the new L2. This resulted in large speculatory activity on meme coins that early bridgers hoped to cash in on before later entrants came at the official launch. As a result, Base’s DEX volume surpassed that of all chains other than Ethereum and BNB Smart Chain on July 30 and 31. These two days alone generated $665,000 in sequencer profit. Since its official launch on August 9, Base has seen between $70k and $180k in daily profit.

Most of the mid-August activity was due to the rise of Friend.tech, a new social platform that aims to enable creators to connect with their audiences via tokenized attention. This was done by allowing users to buy “keys” of certain X (formerly Twitter) accounts that gave users access to private in-app chats. Well-known crypto-related X accounts began to promote their experience on the app, and as speculation ensued so did the number of users eager to participate. In addition, on Friday August 18, Friend.tech released their first snapshot for its airdrop points system, where users earn points that can be redeemed for a future token airdrop based on their activity within the app and referral codes. Each Thursday at 11:59 UTC will have its own snapshot, with criteria changing week in order to stop anyone from trying to game the system. This application alone has seen over 100,000 new signups in just two weeks and has even seen the adoption of famous athletes, gamers, musicians, and influencers outside of the typical crypto sphere of influence. 

While the application leaves much to be desired from a feature standpoint and introduces the ethical implications of tying speculation to individuals, it has done well in bootstrapping a user base that extends outside of crypto in a small period of time, while showcasing OP Stack rollups’ sheer power to handle user demand. 

Mantle

Mantle is an optimistic rollup that uses Ethereum for consensus and MantleDA as its own data availability (DA) layer. MantleDA is a bespoke solution based on pieces of EigenDA infrastructure that allow Mantle to operate prior to EigenDA’s launch with plans to migrate to EigenDA once it is officially live. While MantleDA has the potential performance power of EigenDA, it does not have any of the security that EigenDA offers: there is no collateral at stake and no slashing conditions. It therefore acts like a data availability committee governed by a permissioned set of nodes chosen by the core Mantle team. Mantle users must trust these nodes not to freeze their funds. There are no publicly facing documents that clarify the number of nodes or who specifically is operating these nodes, a risk that needs to be considered. The L2 does have a permissionless fraud proof mechanism, however, allowing for much better assurances over the validity of the rollup. 

Mantle’s DAO has one of the largest diversified treasuries in the space, with the largest stablecoin balance and the second largest ETH balance behind the Ethereum Foundation. It is therefore well capitalized to facilitate and invest in teams to build on the network. MIP-24 creates a $200M ecosystem fund, with $100M coming from the Mantle treasury and $100M matched by strategic partners. As we saw in 2021, large ecosystem funds had a chance to spur a lot of short term activity and the ability to earn outsized farming rewards. However, it is unclear how much of an impact this will have in current market conditions. It is certainly a much higher ecosystem fund than any L2 we have seen thus far. 

With MIP-25, the Mantle DAO voted to create a Mantle Economics Committee in charge of making decisions over treasury allocation and strategies. Within this proposal was the first strategy of putting treasury ETH to use by depositing 200k ETH into the upcoming Mantle LSD, mntETH, and 40k in to Lido with a revenue share agreement. Mantle will accrue 30% of the rewards that would otherwise accrue to the Lido DAO alongside traditional staking rewards.

Linea 

Linea is a new zkEVM built by Consensys, best known for its work across many popular projects such as MetaMask and Infura RPC. Linea is a Type 2 zkEVM, meaning that it is EVM-equivalent, making it fully compatible with existing applications on Ethereum mainnet and other equivalent L2s while trading off some efficiency in proving times.

It’s important to note that in its current state, there is no forced inclusion or escape hatch for Linea, meaning that all users are relying on the sequencer alone for censorship resistance. If the sequencer did decide to censor, there would be no way to force a transaction or withdraw to the L1, which is standard amongst rollups. 

zk-Rollups Showcase Incremental Progress

Polygon ZkEVM

The Polygon zkEVM has been live on mainnet since March 2023, bringing Ethereum security and improved scalability to Polygon’s arsenal of product offerings. Since launch, the zkEVM has seen lackluster adoption, with just $46M locked in the bridge contract and $25M of TVL. Balancer has been the only major DeFi brand to deploy, but the tides appear to be turning. Last month, Gnosis Pay announced it will use the zkEVM for account management of its Visa-based payments system. Additionally, a recent MakerDAO forum post by a Polygon team member showcases the plan to bring Spark Protocol to the zkEVM, notably offering to seed the initial DAI liquidity through a grant. We also saw Convex Finance testing a new L2 gauge voting system on the Polygon zkEVM. Given recent market conditions, user migration has understandably been slow. However, we are seeing a promising trend of builder adoption.

L2 users rely on sequencers for transaction ordering and soft-confirmations. For all major L2s, the sequencers are currently centralized and run by the development team. There is active discussion around the proper method to decentralize the sequencer without compromising user experience, but no solutions have yet been implemented on mainnet. While the Polygon zkEVM is no exception, the team recently announced a partnership to leverage Espresso Systems’ shared sequencer. The implementation is currently running on Polygon zkEVM testnet.

Polygon also unveiled Polygon 2.0 as the “The Value Layer of the Internet” that delivers an improved protocol architecture. A significant part of this vision is POL, a new token that coordinates activity across the entire Polygon ecosystem. If implemented, MATIC will undergo a 1:1 token swap for POL and become the PoS chain’s gas token as well as the broader protocol’s staking token, reward token, and governance token. POL will inherit the current supply distribution of MATIC, but the supply hard cap of 10B is removed, enabling staking rewards to exist beyond the five year window originally provided by MATIC. While this does create inflation, it is capped at 1% annually and is realistically a requirement for a staking token. POL will likely play a role in decentralizing the sequencer, the prover, or both as the zkEVM matures. According to the mid-July blog post from the team, the token migration could begin “within months.”

Given the nascency of zero-knowledge scaling solutions, the Polygon zkEVM relies on a Security Council that can implement instant upgrades in the event of a security compromise. The council controls a multisig and operates under the mandate to halt the network in case of an emergency state and to push upgrades as quickly as possible if the network is halted. The six-of-eight multisig contract requires at least two signers from the Polygon team to execute transactions. As the zkEVM matures, the Polygon team has stated its intent to deprecate the Security Council, striving to build a more decentralized ecosystem that does not contain singular points of failure.

zkSync Era

zkSync Era also launched its mainnet since March 2023, and has been the beneficiary of aggressive airdrop farming. The protocol boasts ~$15M in sequencer profit since its launch in March. However, sequencer profit does not take into account offchain proofing costs, which eat into a portion of these earnings. 

Users on zkSync Era prove to be quite active, executing 27.6M transactions over the last 30 days which is nearly 16x higher than Polygon zkEVM. Again, much of the activity is likely attributable to airdrop farmers. The team has continually teased that the token will play a role in the decentralization of the sequencer and the prover, but no firm plans have been released.

There is currently $400M locked in the bridge contract and $123M of DeFi TVL. While TVL has climbed since launch, there have been no deployments from major Ethereum DeFi protocols. This is in part attributable to the lack of developer tooling. For example, there is an issue where the timestamp returned on the L2 only updates when new batches of transactions are posted rather than on a block-to-block basis. This creates meaningful gaps in time that make some actions, like distributing liquidity incentives, much less feasible for developers. A patch to this issue is planned to go live later this year. It is reasonable to expect deployments from blue chip protocols as zkSync matures. Balancer, Aave, and Uniswap have already passed governance votes, signaling intent to do so. 

Communication Between zk Ecosystems

Arguably the most significant unlock of zk-Rollups is the native interoperability solution. Chains that share the same prover and bridge contract can agree to trust each other and enable fast asset transfers without additional trust assumptions. The only requirement is that two chains share the same prover and store assets in the same bridge contract on the L1. If these requirements are met, users can access asynchronous composability only limited by the frequency of proofs being posted to the L1.

If anything will lead to the downfall of the Ethereum L2 ecosystem, failing to develop an interoperability standard ranks high on the list. The Cosmos ecosystem smartly achieved consensus around IBC early on, and today IBC does over $735M of value transfer per month in a much smaller ecosystem. Seamless interoperability is a catalyst for growth.

Our recent report on the zkStack details a solution to native interoperability dubbed Hyperbridging which allows users to securely move assets from one zkStack chain to another. Although hyperbridging does not provide atomic composability, where cross-chain transactions cannot be conditionally executed as a single bundle, enabling asynchronous composability improves user experience by decreasing friction between ecosystems. Similarly, Polygon zkEVM’s LXLY enables asset transfers between chains that meet the previously described requirements.

While the technical specifications between hyperbridges and LXLY differ, the end result is the same: two ecosystems with asset transfers that carry no additional trust assumptions and settle in minutes. Ultimately, this serves as a significant edge favoring zk-Rollups over Optimistic Rollups that leverage the same tech stack. For example, Base and Aevo are both built using the OP Stack yet cannot quickly bridge assets with no additional trust assumptions, creating a reliance on external solutions that increase risks for users.

However, zk interoperability solutions are siloed to their tech stack, meaning zk-Rollups using different provers or bridge contracts are segregated from each other. Therefore, the liquidity and activity scaling Ethereum is still fragmented between different “constellations'' of chains.

zkSync and Polygon currently use different proving systems, so today they cannot blindly trust each other. Fast forwarding to the maturity of both tech stacks, the communities could agree that both prover mechanisms are battle-tested and trustworthy, enabling cross-ecosystem message passing. However, the actual transfer of assets would still be impossible since the assets are held in different contracts on the L1. A potential solution, albeit unlikely in the near term, would be to create a shared bridge contract between ecosystems. Since one community would never agree to allow another to govern its assets, the contract would need to be immutable and enshrined on Ethereum. The result would be a network of chains that leverage different tech stacks yet have native interoperability.

This begins to spiral into more problems than solutions rather quickly as there is no clear path to determining which provers are trustworthy enough to be allowed in the shared bridge contact. Adding a new prover system brings in more risk as invalid transactions could be incorrectly proved, but not being part of the approved tech stack would be a significant disadvantage to the scalability of that ecosystem. It is clear this could quickly create a “cartel” of approved zkEVMs. While meaningful progress is being made on zk-Rollups, there is still much ground to cover to efficiently scale Ethereum. 

Interesting Developments

Arbitrum

Offchain Labs recently announced BOLD (Bounded Liquid Delay), Arbitrum’s upcoming fraud proof system that allows for permissionless validation. The current mechanism has only 20 whitelisted validators that can challenge invalid rollup state transitions, largely due to the threat of delay attacks where a validator can repeatedly issue wrongful challenges in order to delay transactions from being confirmed. Under BOLD, the network assumes there is one honest node to dispute any number of malicious claims within a defined challenge period (currently seven days). This is because the L2 state follows a deterministic state transition function, so any honest node can prove correct execution.

This is a very important development for the protocol because it removes the risk of relying on an entirely honest permissioned validator set, solidifies an upper bound on settlement time, and guarantees the safety and liveness of the chain even in times of dispute. When it comes to the utilization and design of fraud proofs, Arbitrum is miles ahead of Optimism, which continues to operate without fraud proofs. While the Optimism team is actively working to implement fraud proofs using Cannon, they currently utilize “fast upgrade keys” where the chain can be upgraded within the typical seven day challenge window. While BOLD has been audited by Trail of Bits, there is no strict timeline with which we can expect it to be implemented. More likely than not, this may occur in the next six months

Optimism 

Following Optimism’s Bedrock Upgrade in June, the network has seen large improvements in transaction fees, L1 data costs, and overall user experience. Since then, Optimism has seen increased adoption through its continued growth in OP Stack chains, such as Worldcoin and Base, with other projects like Farcaster announcing their plans to move to the network.

While Optimism and the OP Stack are gaining momentum, many who look at these networks from a fundamental lens do not see how OP Stack chains actually provide value back to Optimism and OP. While the idea of the “Superchain” had been floated, working mechanics changed frequently, and there wasn’t a clear incentive for OP Stack chains to opt in. In late July, Optimism introduced the Law of Chains, which is another step in defining the Superchain and creating the framework for what opting into the Superchain entails. The Law of Chains is essentially a set of principles that all chains within the Superchain must abide by with governance controlled by OP holders. They do not infringe on any individual chain’s sovereignty but include standards such as being open and permissionless and working towards shared technological improvements. Beyond the Law of Chains, Coinbase also disclosed their revenue sharing agreement with the OP Collective, which will be the largest of 2.5% of revenue or 15% of profit, which could set another standard whether directly or indirectly.

Another development in this realm has been the ideation of Optimism’s shared sequencer, which has been teased as part of the Superchain since its inception in 2022. In July, Optimism partnered with Espresso Systems, a team already building their own shared sequencer, to build the prototype for what a shared sequencer on Optimism would look like. Jesse Pollak from Coinbase has also suggested that in the future, Base plans to plug into this shared sequencer and be one operator in a decentralized set of sequencers that operate for all OP Stack chains that opt-in. As of now, the Law of Chains allows chains within the Superchain to run a single-sequencer network, but governance may change their mind in the future and make it mandatory to plug into this shared sequencer.

The shared sequencer and the Superchain as a whole will be governed by OP, so there could be a source of cash flow that accrues directly to the OP token or to a growing treasury that allocates to public good funding. Given the amount of profits that a single-sequencer can generate for itself, it can be difficult to see the incentive for a chain to plug into this network, as atomic composability has only been proven beneficial for arbitrage bots and may increase the surface of MEV by some small amount. However, the more chains that plug in the more value it does generate, and if Base, Zora, and other large chains choose to opt in, it makes for a potentially interesting catalyst for other networks to participate. 

Starknet 

A recent post in the Starknet governance forums details a new consensus mechanism Starknet is developing. The design pulls heavily from the Tendermint engine used by most chains in the Cosmos ecosystem. However, the team plans to modify the existing Tendermint design by implementing transaction streaming, coupling sequencing and execution, and adding a fork choice rule. The adapted implementation would presumably be used to decentralize the sequencer.

The best mechanism to decentralize the sequencer remains an unanswered question, but will continue to be a narrative as rollups mature as proven by much of the discussion in this report. The team that successfully implements the strongest solution may have an edge over its competitors, given it removes a centralized component and delivers on the promise of decentralization. 

Dencun and EIP-4844

Ethereum’s highly anticipated Dencun Upgrade (Denub + Cancun) has made continued development progress throughout 2023. While a launch was originally slated for Q3,, the timeline has since shifted towards later in the year. As of the August 17 “All Core Developers” meeting, devnet-8 was launched as a new place to experiment with Dencun specs. This will certainly not be the last devnet before things are finalized. Following these tests, we can expect Dencun to go live on various testnets in a similar deployment pattern to the Merge and Shapella before final liftoff on mainnet. The current expectation is November or December of this year, but this is always subject to change.

The upgrade decreases the cost L2s pay to post data to the L1 by creating a separate fee market for data “blobs” that contain rollup data. L2s will no longer compete for blockspace with regular L1 transactions, enhancing L2 profitability.

While the target number of blobs per block has changed a bit over the past year, starting at 8 blobs per block before making its way down to 2 at the end of 2022, the current target is set for 3 blobs per block as both achievable while actually producing meaningful change in cost savings. This means a target and max block size of 0.375 MB and 0.75 MB. As a reminder, blobs will have their own pricing curve that looks exactly as the EIP-1559 pricing curve for transactions but will be competing only amongst other blobs. With the number of L2s that need to post data to Ethereum periodically continuing to rise, demand may outpace blob supply. We could see upscaling of these parameters if demand does rise high enough, but full Danksharding, with a target of 128 blobs per block, is still many years away as it requires ePBS and data availability sampling. 

Final Thoughts

The rollup landscape is continuing to evolve at a very fast pace, with far more rollups now than at this point last year. zkRollups in particular have seen immense strides, with multiple zkEVMs with fully functioning mainnets despite the popular notion that developments were much further away. In addition, research is showing the capabilities of zk rollups to exhibit strong validity guarantees coupled with seamless interoperability. However, teams are incentivized to get their projects out quickly and compete for Ethereum assets before they make their way into some competing rollup’s bridge, causing some rollups to compromise on security. While bridging to these new chains could lead to finding gems early, it also comes with many risks that must be considered.

There are a lot of parallels to 2021, where we saw new alternative L1’s emerge with their own ecosystem funds and incentives that allowed for innovation and cheaper blockspace than Ethereum mainnet, but without the easy and secure bridge to Ethereum that rollups offer. So, while the L2 ecosystem looks potentially brighter than the L1 trade, if similar patterns occur a majority of these ecosystems will become ghost towns by the end of the cycle, while the survivors retain promising communities and development.

The future for rollups using Ethereum for DA and settlement looks very promising. With EIP-4844 going live towards the end of this year, we expect stepwise improvements from a rollup cost perspective, and the main rollup teams building towards more secure and robust architectures. We can expect that this time next year, we will see exponentially more rollups as the tech continues to mature.