An overview of blockchain scaling and interoperability solutions, including: rollups, data availability, arbitrary messaging bridges, and more.
Rollups saw their TVLs decline throughout 2022, a likely impact of the bearish market outlook, with a handful retaining users as a result of their airdrop or potential airdrop.
Other scaling solutions made significant development, from data availability solutions like Celestia to zkEVM efforts like Polygon Hermez.
Liquidity bridges have also lost signficant TVL, a result of declining asset prices as well as multiple bridge hacks in 2022.
State of Blockchain Scaling Solutions in 2022
As blockchain adoption grows, we begin to see accelerated developments in scaling technology. There are multiple approaches to improving blockchains' throughput, from rollups to data availability optimization to modularization. The current scaling efforts are predominantly Ethereum-based or data availability-based. While there are scaling efforts on other blockchains such as Cosmos, the implementations are typically marginally different.
Overall, scaling solutions attracted significant attention with the launch of Optimism's governance token, OP. This resulted in speculation on the airdrop criteria for existing rollups that are yet to launch a token, such as Arbitrum and StarkNet. While the crypto market has generally slumped, a token airdrop usually stimulates short-term hype and activity that can bootstrap signficant adoption. With zkSync 2.0's and Polygon's zkEVM launch, along with Arbitrum's and StarkNet's highly anticipated token drop, it is likely that we see signficant growth in on-chain activity for the L2 ecosystem in 2023.
Ethereum-based Scaling Solutions
The overall landscape for Ethereum-based scaling solutions has largely been dominated by Arbitrum, Optimism, and dYdX, currently accounting for close to 90% of the TVL across all Ethereum-based rollups.
Figure 115: TVL of all rollups in 2022 Source: L2Beat
The reason for their dominance would be a mix of ﬁrst-mover advantage, strong utility and deep liquidity. We will look at some of the existing rollups in-depth in the next section.
Optimistic rollups (ORs) rely on fraud proofs to ensure the validity of their states on Ethereum. This requires vigil watchers to call out any invalid states proposed, which can take some amount of time. As a result, all ORs necessitate some arbitrary period of time (i.e., a challenge period) when facilitating withdrawals from the OR back to Ethereum. This time period is typically 7 days (i.e., Arbitrum and Optimism) but can be arbitrarily shortened to 7 minutes (i.e., Metis). Generally, the shorter the challenge period, the lower the level of security.
While a robust watcher network could improve security to some degree, by having watchers on alert for all new states proposed, it involves some trade-off between centralization and performance. For instance, a series of newly proposed state roots would need to be checked by a handful of watchers (centralized) or apportioned to be checked by multiple watchers (trust every watcher) or all checked by multiple watchers (slow). With the right incentive design, the ideal approach would be trusting every watcher. However, most ORs today rely on centralized watchers to submit fraud proofs.
Arbitrum currently has a permissioned fraud-proof system where only whitelisted validators are allowed to submit fraud proofs.
Figure 116: Dapp dominance on Arbitrum in 2022 Source: DeFiLlama
In terms of Arbitrum dapps growth, GMX would be the most significant one for 2022, sitting at ~43% of Arbitrum's TVL. GMX is a decentralized perpetuals exchange, where users can trade perpetuals contracts or provide liquidity for traders.
Figure 117: GMX usage metrics in 2022 Source: GMX
GMX attracted over 57,000 users and accrued $21 million in fees over the past 2 months and is still steadily growing throughout the bearish climate of the past year. GMX also expanded to Avalanche, though adoption is relatively modest there for now. There is contention on the viability of GMX's model now, after LPs were economically exploited on Avalanche in September 2022 as a result of GMX facilitating zero-slippage trades. Essentially, GMX serves as a platform where active traders are effectively trading against passive LPs. Only time can tell if such a dynamic is sustainable for all stakeholders.
Arbitrum Nova is an AnyTrust variant that lowers costs by posting transaction data to a Data Availability Committee (DAC) instead of on-chain. A DAC has N members, of which AnyTrust assumes at least two are honest, which is attested with a Data Availability Certificate that ensures the data will be made available if required.
Figure 118: Dapp dominance on Arbitrum Nova in 2022 Source: DeFiLlama
While Arbitrum Nova's adoption is relatively modest, at around $4.8 million TVL, it primarily serves as a cheap blockchain with sub-cents fees instead of a platform for meaningful financial activity as of its current implementation. This is because relying on an off-chain DAC is still less secure than on-chain data availability, which is an important criterion for a rollup to inherit Ethereum's security. The concept of data availability and its implications will be discussed in a later section.
Optimism's fraud-proof system is still under development, as such, users are currently trusting one block proposer, OP Labs PBC, to submit valid L1 state roots.
The most notable event for Optimism in 2022 would be the launch of its governance token, OP, in June 2022. This brought significant attention and hype for Optimism and also boosted Optimism's on-chain activity. In fact, the OP airdrop garnered so many users that it caused a noticeable slowdown for Optimism'sRPC. Additionally, it was announced that 5.4% of OP's supply would be distributed to projects on Optimism over the next six months via governance, which helped bootstrap significant liquidity on Optimism.
Figure 119: Dapp dominance on Optimism in 2022 Source: DeFiLlama
Optimism was largely dominated by Synthetix in 2021 but had seen the proliferation of other dapps in 2022, such as Aave and Velodrome, both of which were recipients of OP token grants. Most of Synthetix's TVL on Optimism came from the staking of SNX tokens, which does not stimulate much on-chain activity. Velodrome, an Optimism-based DEX, saw decent growth in TVL, bootstrapping significant liquidity for major altcoins on Optimism, such as ETH and OP, as well as stablecoins, such as USDC and sUSD.
Figure 120: Value locked in Velodorome in 2022 Source: DeFiLlama
The spike in TVL in July was the result of a 3 million OP token grant that was awarded to Velodrome by OP Labs PBC to deepen the liquidity on Velodrome. This had a great impact on Velodrome's ability to attract liquidity, especially for stablecoins, overtaking Curve in terms of TVL on Optimism. Another notable design of Velodrome would be their vote-escrowed (ve) token model, where veVELO holders are incentivized to vote on the distribution of rewards to different liquidity pools. This resulted in a generally increasing trend for the number of veVELO holders as there are sustainable incentives for LPs to hold veVELO. However, this trend could change quickly should liquidity on Velodrome start leaving.
Metis' fraud-proof system is a fork of Optimism. Like Optimism, Metis' fraud-proof system is under development, so users are expected to trust Metis' block proposer to submit valid L1 state roots. In April 2022, Metis' transaction data started being posted to MEMO-distributed data storage instead of on-chain, meaning that Metis no longer relies on Ethereum for data availability. MetisDAO had announced that there would be an additional challenge process to force Metis' sequencers to post missing data to Ethereum, but it has yet to be implemented.
Figure 121: Dapp dominance on Metis in 2022 Source: DeFiLlama
While NetSwap dominated Metis' TVL for most of 2021, it saw various competitors in 2022, such as Hummus and Hermes, which had their own liquidity mining incentives that caused mercenary capital to ﬂow from NetSwap to these other DEXs. Aside from DEXs, the only dapp on Metis with notable liquidity would be Synapse bridge.
Boba is another fork of Optimism whose fraud-proof system is also currently under development. Similar to Optimism and Metis, Boba users are required to trust a single block proposer to submit valid L1 state roots. In Boba's case, enya.ai, the core developer of Boba, is responsible for this role.
Figure 122: Dapp dominance on Boba in 2022 Source: DeFiLlama
OolongSwap, the first DEX on Boba, dominated the TVL for most of 2021, but SushiSwap quickly ate into OolongSwap's market share when it was brought onto Boba's ecosystem in October 2022. Similar to Metis, Boba's TVL is mainly concentrated in DEXs, in particular, OolongSwap, SushiSwap, and Gin Finance. In comparison, there is lackluster adoption for other dapps on Boba, such as Outcome Finance, a liquidity bootstrapping service, as well as Bodh Finance, a decentralized lending platform.
Zero-knowledge Rollups & Validiums
Zero-knowledge rollups (ZKRs) differ from ORs in that they utilize advanced cryptographic techniques to generate a validity proof that ascertains the validity of a particular state. There are two general types of validity proofs, Succinct Non-Interactive Arguments of Knowledge (SNARK) and Scalable Transparent Arguments of Knowledge (STARK). The former is generally more widely adopted though the latter is arguably more optimized for proving more information.
Another unique feature of scaling solutions utilizing validity proofs is that they can leverage off-chain data availability, though there is a different classification for such solutions: Validiums, a term coined by StarkWare. This is because submitting a valid validity proof on-chain is considered highly improbable without a valid execution trace, meaning that a malicious attacker would find it nearly impossible to submit a valid proof for fraudulent states. As such, the data used in the generation of the validity proof can be hosted off-chain with a marginal trade-off in security.
Various Validiums were launched with the help of StarkWare, each with its own unique DAC, a handful of entities that guarantees all transaction data will be posted on-chain. In the event that the Validium operator refuses to service withdrawal requests, the transaction data posted on-chain will allow individual users to submit withdrawal requests on their own by submitting a Merkle proof for the latest state. That said, the trade-offs for data availability modes are nuanced and will be further discussed in a later section.
The leading ZKR by TVL is dYdX and it leads the other ZKRs by a significant amount. As of this writing, dYdX holds $385 million in TVL, which is more than 3 times larger than Loopring's TVL of $114 million, the second largest ZKR. The reason for dYdX attracting so much liquidity is that it offers a perpetuals trading experience much better than its competitors for now. This is apparent from the high trading volumes on dYdX, and although volumes fell significantly in 2022, dYdX still managed to retain a significant amount of trading activity, averaging $1 billion in daily trading volume in the month of November.
Figure 123: dYdX trading volume in 2022 Source: dYdX Metabase
Another indicator that stands out for dYdX would be the number of unique depositors, which has been climbing steadily since its inception. While the growth in unique depositors tapered off in 2022, the fact remains that dYdX is gaining traction in adoption even in a bear market. The collapse of FTX may well have been the catalyst to push the crypto community to start adopting DEXs with full transparency instead of relying on CEXs.
Figure 124: dYdX unique depositors since 2021 Source: dYdX Metabase
While dYdX has been steadily onboarding new users, it is not unaffected by the ongoing bear market. Daily open interest fell significantly from the highs of $1 trillion in early 2022 to an average of $225 billion for the month of November.
Figure 125: dYdX open interest in 2022 Source: dYdX Metabase
It is worth pointing out that dYdX is a self-custodial solution, which means that traders remain full control of their funds. Even when funds are bridged to the ZKR, users are able to withdraw their full account balance at any time without relying on any centralized entity. This is because dYdX posts all data, including its state root, on-chain, so users are able to submit a withdrawal request on Ethereum itself, to pull their own funds out from dYdX.
That said, dYdX announced that they will soon be migrating to the Cosmos ecosystem. The migration comes with the launch of dYdX v4, which is supposed to bring improvements in terms of throughput and decentralization. dYdX also indicated that the existing perpetuals exchange will be deprecated once the migration to a standalone Cosmos chain is complete, which raises the question of whether all of dYdX's users will migrate to the Cosmos ecosystem, given that there is likely a significant number of users who chose dYdX because it was built on top of Ethereum. The migration is estimated to start at the end of 2022.
Loopring was launched in March 2020, making it the oldest ZKR in the space. It attracted a considerable TVL since and remained one of the top few ZKRs by TVL. That said, Loopring's TVL is largely comprised of the LRC token, their native governance token.
Figure 126: Value locked in Loopring in 2022 Source: DefiLlama
The other metric that provides some insight into Loopring's usage would be the daily trading volume. The trading volume on Loopring's automated market maker (AMM) has been relatively modest, averaging $1.6 million in daily volume, with large spikes on November 4 and 10, 2022. The reason for the spike in volumes might be due to the collapse of FTX, which saw high trading volumes throughout that period in general and also pushed traders to find alternative trading avenues instead.
Additionally, Loopring has an order book exchange built on L2, though its trading volume is far lower than that of the AMM. This trend indicates that most of Loopring's users prefer using the AMM interface instead of the order book. There could be various reasons for this, though the most likely one is that the AMM interface is simply easier to use.
Aside from the AMM and order book, Loopring also supports the minting of NFTs and has an NFT marketplace. The traction for NFTs on Loopring is relatively slow, averaging $145,000 in daily volume over the past three months.
ImmutableX is an NFT-specific Validium developed by StarkWare. Such solutions are cheaper than their on-chain data availability counterparts, such as dYdX, but there are additional trust assumptions made for Validiums. The trade-offs between on-chain and off-chain data availability will be addressed in a later section.
In terms of trading volume, ImmutableX facilitates an average daily trading volume of $850,000 over the past three months. Additionally, in recent months, Gods Unchained began dominating the trading volume on ImmutableX, accounting for 90% of ImmutableX's trading volume since September 2022. While Gods Unchained has grown significantly in terms of trading activity, the same cannot be said for the other collections. Given that ImmutableX's revenue is dependent on the trading volume it facilitates, it would certainly want to incentivize the trading activity of all NFTs listed on its platform, lest it cause ImmutableX to be over-reliant on a single NFT project for its revenue.
Figure 128: ImmutableX trading volume in 2022 Source: Immutascan
zkSync experienced significant growth in TVL in Q1 2022 which declined afterward, likely caused by falling ETH prices rather than users' withdrawals, as the amounts of stablecoins on zkSync remained relatively unchanged. zkSync's TVL is largely dominated by ETH, a likely consequence of the lack of dapps on zkSync, which provides zero incentives for bridging altcoin liquidity to zkSync.
Figure 129: Value locked in zkSync in 2022 Source: L2Beat
The reason for the lack of dapps on zkSync 1.0 is that zkSync 1.0 does not have a generalizable framework for applications to be built on top of zkSync 1.0, making it hard for it to scale in terms of dapp development. However, with zkSync 2.0's zkEVM, it will likely resolve this issue as dapp developers can use Solidity as the framework for building dapps.
zkSync announced the launch of their baby Alpha on October 29, which comes with on-chain proof generation and verification. However, their mainnet block explorer indicates that L2 blocks are being generated intermittently, with only 14 blocks beingvalidated since its launch. According to Matter Labs, the full public launch will only come in early 2023. Another advancement that zkSync 2.0 intends to introduce is Layer-3 (L3) Fractal Hyperchains. The main gist of L3s is that they offer a customizable execution environment and leverages recursive proofs, where the validity proofs of L3 are combined into a single L2 validity proof, which is posted on Ethereum mainnet itself, which allows the L3 to inherit some of Ethereum's security. However, L3s are merely an extension of L2 and this concept can be abstracted infinitely, though the computational overhead increases with every additional layer.
ZKSpace saw a relatively similar trend in TVL to zkSync, with a spike in Q1 2022, followed by a decline throughout the rest of the year. It should be noted that most of ZKSpace's TVL is made up of WBTC, ETH and ZKS, their own governance token. More notably, ZKS makes up an average of 28% of ZKSpace's TVL throughout the year, meaning that ZKSpace's TVL has a sizable composition of an endogenous asset as well, similar to Loopring.
Figure 130: Value locked in ZKSpace in 2022 Source: L2Beat
It should be noted that ZKSpace currently comprises three products: ZKSwap, an AMM; ZKSea, an NFT marketplace; and ZNS, a domain name service. However, most of ZKSpace's on-chain activity stems primarily from ZKSwap.
In terms of trading volume, ZKSwap is a far cry from Loopring's AMM, averaging only $200,000 in daily volume over the past month. Interestingly, it saw the same spike in trading volume over the same period as Loopring, which corroborates the theory that crypto users were using alternative trading avenues post-FTX collapse amidst a volatile trading environment.
Figure 131: ZKSwap trading volume in 2022 Source: ZKSpace
Previously DeversiFi, Rhino.fi is another Validium that was developed by StarkWare that provides DeFi capacity. Rhino.fi's TVL generally declined throughout 2022, despite a short spike in April 2022. In terms of TVL composition, DVF, Rhino.fi's governance token, reduced in weightage throughout 2022, falling from an average of 52% in January to 20% in November. This highlights that while Rhino.fi's initial TVL was substantially bootstrapped by DVF, it managed to attract liquidity for other assets such that its TVL is not primarily backed by endogenous assets.
Figure 132: Value locked in Rhino.fi in 2022 Source: L2Beat
Despite Rhino.fi's decent TVL growth, it had lower trading activity, averaging $235,000 in daily volume over the past 30 days. Similar to Loopring's and ZKSpace's AMM, Rhino.fi saw a spike in volumes from November 8 to November 10. More notably, there was another spike on November 25 and 28, though it seems to be spontaneous rather than event-specific.
Figure 133: Rhino.fi volume in 2022 Source: L2Beat
Sorare is another Validium solution developed by StarkWare that primarily serves as an NFT marketplace where users can trade NFTs related to Sorare's fantasy sports games. Sorare started out with only a single collection of soccer-related NFTs, but quickly expanded into the baseball and basketball domains as well. That said, Sorare has yet to see a significant surge in its TVL. This is likely attributed to the ongoing bear market, which would likely negatively impact the adoption of NFT-based games as well.
Figure 134: Value locked in Sorare in 2022 Source: L2Beat
APEX is another Validium solution by StarkWare. Similar to dYdX, APEX offers a perpetuals trading experience, albeit for fewer pairs and lower order-book depth. However, the use of off-chain data availability will likely allow APEX to charge significantly lower fees than dYdX. The growth in APEX's TVL in November is likely attributed to the earlier conjecture that crypto users are looking for decentralized alternatives to trade after FTX's collapse.
Figure 135: Value locked in APEX in 2022 Source: L2Beat
However, it remains to be seen if APEX can truly become a perpetuals exchange that rivals dYdX's liquidity and order-book depth. As mentioned earlier in dYdX's subsection, some users choose dYdX because it inherits Ethereum's security. An off-chain DAC may not ever be a substitute for that.
Aztec is a privacy-focused ZKR that currently provides private DeFi yield aggregation on Ethereum. In 2021, Aztec launched zk.money, an application that allows users to shield transactions on L2. However, that has been deprecated with the launch of Aztec Connect, as the new zk.money now allows users to interact with Ethereum's DeFi ecosystem from L2.
This is a remarkable feat, considering that users can now access Ethereum native DeFi apps while maintaining their privacy. As a result, Aztec's TVL remained relatively consistent throughout 2022, even with the declining asset prices. There was a migration process in June 2022, where we saw a fall in Aztec's TVL, only to return back to Aztec Connect two months later.
Figure 136: Value locked in Aztec in 2022 Source: L2Beat
Aside from the initially supported assets of ETH, DAI, and renBTC, Aztec Connect also started supporting wstETH as it is commonly used as collateral in Ethereum DeFi applications. Overall, Aztec's TVL still largely comprises ETH and DAI.
Figure 137: Value locked in Aztec in 2022 by asset Source: L2Beat
While Aztec made significant headway in providing privacy for DeFi on Ethereum, it remains to be seen if such privacy layers will eventually become the target for regulators. Just like how OFAC sanctions suddenlyhit Tornado.cash, Aztec's zk.money may one day become a similar target for regulators.
StarkNet is StarkWare's generalizable ZKR. While StarkNet Alpha launched on the Ethereum mainnet in November 2021, it only started accepting deposits in May 2022. Since then, its TVL has been increasing steadily, causing StarkWare to continuously raise the cap it had initially placed on StarkNet's TVL.
It is clear that ETH dominates most of StarkNet's TVL, though it has a marginal amount of USDC and DAI. It is peculiar that StarkNet attracts this level of liquidity when there are no applications that are production-ready on StarkNet just yet. This is likely due to speculations that bridging liquidity to StarkNet is one of the potential criteria for the STARK token airdrop.
While there is a substantial level of speculation on the STARK token airdrop, it remains to be seen if the STARKtoken will be well distributed. Though there are a handful of precedents for StarkWare to learn from, optimal tokenomics has always been relatively challenging to attain, given the unpredictability of the stakeholders' actions. Nonetheless, StarkWare has mentioned that the current tokenomics are not finalized and will likely consider the feedback from the community and relevant stakeholders before making a final decision.
In terms of technological developments, the most notable one would be the use of Cairo, a programming language that optimizes the generation of validity proofs.
Figure 138: Value locked in StarkNet in 2022 Source: L2Beat
Polygon forayed into the ZKR domain in 2022, acquiring multiple ZKR teams, such as Hermez and Mir (now Polygon Zero). Polygon also experimented with zk-STARKs with its own initiative, Polygon Miden. To date, none of Polygon's ZKR has been launched on the Ethereum mainnet, though Polygon Hermez is allegedly the closest.
Hermez saw its TVL declining steadily throughout the year, as it announced that it would no longer be developing its ZKR but instead, contribute to Polygon's ZKR efforts. Naturally, users would likely withdraw their assets from Hermez to repurpose their capital elsewhere. However, it would appear that some liquidity has remained even up till today. Given that there are hardly any transactions on Hermez, it would stand to reason that the zkEVM version of Polygon Hermez will likely be on a different platform from the existing Hermez ZKR.
Figure 139: Value locked in Polygon Hermez in 2022 Source: L2Beat
One of Polygon's other ZKR efforts, Polygon Zero, is focused on creating SNARK proofs which has been claimed to have the fastest proof generation andverification times. Polygon Miden, on the other hand, is a Polygon initiative that focuses on creating a STARK-based zkEVM rollup that would provide significant scaling advantages for applications.
Regardless, all of Polygon's ZKR efforts seem to coalesce towards a zkEVM solution that will utilize both STARKs and SNARKs. The main idea is to combine multiple fast-generating SNARKs into a single STARK, which is conceptually identical to proof recursion. Only time will tell if Polygon's foray into the ZKR space is successful.
Outlook on Rollups
Overall, the outlook for rollups remains ever-positive, owing to the fact that rollups generally have an optimal trade-off between security and costs. After all, gas costs on rollups are amortized over all transactions. However, at current ETH prices, it would appear that the difference in costs across rollups makes marginal differences. This will not be the case in a bull market should ETH trade at higher prices, where the minute difference of cents today can become a stark difference of dollars. Essentially, this means that over the long run, only the most gas-efficient rollups will outcompete the others.
Figure 140: Gas cost to send ETH or a token swap Source: L2Beat
It also remains a question if rollups can attract liquidity on a comparable scale to Ethereum, with one of the major concerns being that many rollups today are still centrally operated. Moreover, there is no incentive for rollup operators to decentralize their current role anytime soon. As such, centralization concerns for rollups will likely remain for the foreseeable future, at least, until one major rollup starts decentralizing its operator nodes.
In terms of increasing adoption, one of the most significant ways to bootstrap liquidity and incentive on-chain activity is through a token. With some of the major rollups, such as Arbitrum, zkSync, and StarkNet, that have yet to launch their token, it will be interesting to see how the respective teams will utilize token incentives to drive adoption.
Overall, rollups are becoming quintessential for Ethereum users, as it provides an Ethereum-like experience with near-Ethereum-level security, at a fraction of the costs of Ethereum. As Ethereum adopts danksharding in the foreseeable future, rollup technology will only stand to gain from building on top of Ethereum – a synergistic harmony between ecosystems.
Data Availability Solutions
Data availability refers to the ability to ensure that all required data can be made available to a party at any point in time, should it be requested.
In the context of rollups, the cost savings are generally either from posting transaction data as calldata so as to save on computation costs, or posting transaction data off-chain and only post it on-chain when necessary. The former is the approach by ORs and ZKRs while the latter is the approach by Validiums.
We will first look into Validiums and their cost savings, followed by diving into the trade-offs made by utilizing an off-chain DAC. Thereafter, we will look at Celestia, a DA-focused scaling solution.
The existing Validiums today offer significantly lower fees than ZKRs like dYdX and Loopring, and even ORs like Arbitrum and Optimism. This is because the gas costs of a rollup are essentially passed on from the gas costs incurred on Ethereum itself. Thus, the gas that a rollup charges is a function of the volume of data it posts to Ethereum, as well as the level of computation it requires. Both ZKRs and ORs ofﬂoad computation from Ethereum to their respective layers, and Validiums go one step further by ofﬂoading data from Ethereum to an offchain data committee as well. As such, Validiums typically consume far less gas on Ethereum than ORs and ZKRs.
Figure 141: Gas used by rollups in 2022 Source: Dune Analytics (@funnyking)
At first glance, it would seem that ZKRs consume much less gas than ORs. However, existing ZKRs are usually far more specific, whereas ORs are usually generalizable. As a result, there are far more complex interactions that ORs can support as opposed to ZKRs. Not only that, but the gas consumption of a rollup is also dependent on the volume of transactions on the rollup itself. As such, a more relevant metric for rollups' gas consumption would be the average gas consumed per transaction.
Figure 142: Monthly average gas consumed per transaction by rollups in 2022
Source: Dune Analytics (@funnyking)
When comparing the average gas consumption per transaction, it becomes clear that ZKRs are not significantly outperforming ORs. In fact, in 2022, ORs gas consumed per transaction fell with improved technology, namely in calldata compression methods. For example, Optimism launched its calldata compression in March 2022 and Arbitrum launched Arbitrum Nitro, with calldata compression, in August 2022, evidenced by the relative decrease in gas consumed per transaction during those periods. Calldata compression sacrifices computation overhead for using lesser block space but the relative trade-off varies with the compression method used.
Validiums save on the gas required for posting transaction data on-chain by having an off-chain data committee post a data availability attestation, which is essentially a new Merkle root agreed upon by the majority of the DAC. In Figure 142, it would appear that rhino.fi is becoming increasingly inefficient but this stems from the main ﬂaw of ZKRs, especially for those using STARKs, where posting and verifying a validity proof on-chain costs a fixed, significant amount of gas. As such, ZKRs are susceptible to becoming inefficient if the transaction count begins to drop, as the fixed amount of gas consumed by the validity proof is no longer amortized over as many transactions as before. In rhino.fi's case, as DeFi activity tapered off in the bear market, the transaction volume on rhino.fi fell off, which consequently caused gas consumed per transaction to spike.
Figure 143: Monthly averaged gas consumed per transaction for Loopring, dYdX, and ImmutableX in 2022 Source: Dune Analytics @funnyking)
However, if one were to simply compare rollups with relatively stable growth trends, it becomes apparent that a ZKR with a much higher transaction count can amortize gas costs more efficiently, whereas a Validium is almost always certainly more performant in gas costs, albeit at the cost of data availability. This is pertinent because the DAC is the users' last line of defense against a malicious Validium operator. Without the transaction data to create a valid Merkle proof, users' funds can be held hostage indefinitely by the operator. That said, it should be noted that at no point can the operator direct users' funds to any arbitrary address, which eliminates the incentive for the operator to misbehave.
The identities of DAC members are usually kept private, such as that for ImmutableX, which uses a 5-of-7 multisig, and Sorare, which uses an unverified smart contract for data availability attestation. Such obscurity might protect these entities from being targets of cyber-attacks but also forces users to blindly trust the DAC without knowing their identities. Thus, some Validium protocols, such as rhino.fi, chose to publicly announce the identities of the DAC members instead.
Figure 144: Entities in Rhino.fi's DAC in 2022 Source: Rhino.fi
The conclusion on Validiums is that they will generally out-scale ZKRs and ORs because they reap the benefits of more efficient scaling by leveraging validity proofs and also the advantage of reduced block space usage by posting data off-chain.
In the domain of data availability, other solutions such as Celestia sprung up to serve as a specialized data availability layer. Such a layer holds the attestations of block data except, instead of utilizing a DAC, it is a permissionless layer where anyone can stake tokens (TIA for Celestia testnet) to participate in the network as one of the attesters. Celestia is able to provide scalable data availability because it uses Reed-Solomon (RS) erasure coding, a technique that has been used in various consumer technologies. Implementing the RS algorithm in the blockchain context requires some parameterization, such as the minimum number of nodes as well as the maximum number of bad actors.
Figure 145: Visualization of Celestia's Reed-Solomon algorithm Source: The Block Research
The more nodes there are in the network, the less data each node has to store to reconstruct the entire database, as illustrated in Figure 145. Additionally, a higher degree of RS algorithm can be used to lower the percentage of dishonest nodes required to compromise the network, on the condition that there are sufficient nodes in the network.
All of Celestia's nodes essentially play the role of a DAC, that is, posting attestations for the availability of data on-chain. Celestia's documentation specified that it implements the 2D RS encoding scheme, an optimized adaptation of the RS algorithm for Celestia's block data structure.
Celestia is also able to act as a consensus layer as its nodes will be putting up an economic stake when Celestia moves to mainnet. This will allow any sovereign rollup to utilize Celestia as a consensus and data availability layer, though it remains a question if rollups will choose to use Celestia over Ethereum as a consensus layer.
Figure 146: Visualization of Celestia's modularization Source: The Block Research
Celestia secured partnerships with L1s like Cosmos, to launch joint efforts such as Cevmos, a settlement layer built on top of Celestia that connects multiple EVM rollups. This architecture is similar to L3s built on top of L2 ZKRs.
Figure 147: Visualization of Cevmos's and StarkNet's modularization Source: The Block Research
Celestia testnet's adoption metrics are currently the best proxy for the mainnet's adoption, though it will likely vary, depending on the state of the markets. Moreover, a testnet has limited functionality whereas a mainnet will very likely see EVM-compatible applications migrating over to capitalize on potential incentives that may come with Celestia's mainnet launch.
Figure 148: Distribution of voting power on Celestia testnet Source: Celestia testnet explorer
It is apparent that most of the testnet activity stems from 10 entities, though there are a total of 150 activevalidators. The top 10 entities hold over 99.9% of the voting power on Celestia, which would be a concerning centralization problem on mainnet. That said, most protocols are usually highly centralized when they are first launched, as they usually require a vigil watch over potential bugs. It would only be a significant concern if Celestia starts amassing a sizable TVL while maintaining a relatively similar distribution of voting power depicted above.
With the rollup space becoming increasingly competitive, Celestia may offer rollups a cheaper alternative for data availability instead of Ethereum. Though this narrative will allow Celestia to boost its adoption, it should be noted that Celestia's main revenue stream will come from rollup operators who are willing to pay for the data availability attestation that Celestia provides. This means that Celestia's long-term sustainability is dependent on rollups using it, not to mention competing with potential competitors in the future, such as Polygon Avail.
Outlook on Blockchain Scaling
While Ethereum-based scaling technology saw many developments, there are also many noteworthy efforts beyond the Ethereum ecosystem that made significant headway. From the Cosmos ecosystem to sharding technology to zkEVM, many of these developments are also pertinent to blockchain scaling.
As discussed in Layer-1 section, one of Cosmos's main scaling efforts is the development of application-specific chains. An example of an app-specific chain is Osmosis with a native DEX that offers advanced features such as bonding curves, dynamic fee swaps, and multi-token liquidity pools. Another example would be Sei Network, a DeFi-specific chain that seeks to share liquidity across all applications built on it. This should grant users access to the deepest liquidity regardless of which application they use, thereby solving the problem of fragmented liquidity. Naturally, ofﬂoading each unique use case to a specific chain, it would allow each chain to optimize its performance for the use case it handles, not to mention reduce the bloat that would come from handling multiple use cases in a single environment.
Another notable effort is Dymension, a framework for modularizing Cosmos-based rollups. Any arbitrary application can build its own customized rollup (RollApp) on top of Dymension while Dymension serves as the settlement and consensus layer. As such, every RollApp would have the choice of its own DA layer (such as a Cosmos-based chain or Celestia), or even its own external DAC if it prefers to. This would effectively allow any application built on Cosmos to have full customizability of its settlement, consensus, and data availability layer, thereby granting developers great versatility with the extent of modularity and choice of layers they wish to use.
Figure 149: Visualization of Dymension's modularization Source: The Block Research
Enshrined rollups are blockchains with a modular architecture that is enshrined natively in the protocol, presumably to maximize long-term stability.
We first look at Polkadot's architecture, where it uses the relay chain as the settlement and data availability layer, while parachains are expected to handle state execution as well as consensus.
Figure 150: Visualization of Polkadot's modularization Source: The Block Research
Various other architectures exist, such as the four data shards that Near uses, the transactional OR mechanism that Tezos uses, and the validity-proof-settled Mina.
Figure 151: Visualization of hybrid modularization architecture Source: The Block Research
As the blockchain space continues to grow, it is likely that we will see more unique architectures with different optimizations. However, every additional siloed ecosystem fragments both liquidity and adoption further, so it may not necessarily be in the best interest of the crypto space to be headed in that direction.
L3s are an abstraction from ZKRs, where an additional layer can be built on top of the existing L2s, only for the L3's validity proof to be combined into the L2's validity proof that is submitted to L1 (typically Ethereum). However, this concept can be abstracted arbitrarily many times, making layer numbers relatively meaningless. It would be less confusing to refer to such layers as customizable execution layers. This classification would include Cosmos' RollApps together with zkSync's Fractal Hyperchains.
Figure 152: Visualization of execution layers in modularization Source: The Block
Regardless of nomenclature, L3s do bring great ﬂexibility in optimizing an existing settlement layer for a specific application. For example, zkSync's L3 effort,Opportunity, is meant for public experimentation.
Figure 153: Visualization of zkSync's Fractal Hyperchains Source: Matter Labs Medium
Another notable effort in the L3 domain is Slush, which aims to build an SDK that allows L3s to leverage the Tendermint consensus mechanism. Theoretically, Slush could work for any ecosystem, but it is currently focusing on StarkNet mainly due to StarkWare's shared prover (SHARP) technology, as parallel Cairo program executions can be verified with a single validity proof. That said, Slush needs to address the challenges of being able to run a Tendermint light client in Cairo as well as ensuring Tendermint's compatibility with StarkNet.
Of the four main shard chains that exist today, namely Elrond, Near, Harmony, and Zilliqa, only Elrond and Aurora, an EVM emulator on one of Near's shards, managed to retain a notable fraction of their liquidity and users in 2022. Zilliqa's liquidity appears to fell along with the bear market, while most of Harmony's liquidity was drained as a result of the Harmony Horizon Bridge hack in June 2022.
Figure 154: TVL of shard chains in 2022 Source: DeFiLlama
On November 4, 2022, Elrond rebranded itself to MultiversX at its XDay Conference. This rebranding also indicates MultiversX's shift in focus from developing DeFi to Metaverse applications. With a shift in focus, it remains a question for MultiversX's existing community if DeFi development would be neglected. As it has only been a month since the rebranding, there has not been any significant change to the liquidity on MultiversX and the only DEX in its ecosystem, Maiar. On the other hand, transaction count declined in November, despite the rebranding.
Figure 155: Elrond usage metrics in 2022 Source: Elrond explorer
For Aurora, it remains relatively vibrant, with new dapps surfacing in 2022, such as Bastion, a lending protocol, and Aurora Plus, a staking service. While the main DEX on Aurora, Trisolaris, retained a significant market share, the aforementioned applications saw considerable growth in TVL as well.
Figure 156: Dapp dominance on Aurora in 2022 Source: DeFiLlama
Harmony's main activity has always been dominated by DeFi Kingdoms, a popular blockchain game. This is in spite of DeFi Kingdoms' cross-chain expansion to Avalanche with its own subnet. However, as liquidity gets pulled from Harmony's ecosystem from June 2022 onwards, DeFi Kingdoms' relative dominance shrank and SushiSwap rose instead. This is because a P2E game such as DeFi Kingdom loses profitability once the liquidity of the overall ecosystem has been impacted. Not only that, the fact that no substantial liquidity has returned to Harmony post-hack implies lowered expectations for Harmony's recovery. Consequently, other Harmony-based dapps dwindled as well.
Figure 157: Dapp dominance on Harmony in 2022 Source: DeFillama
While the development of shard chains has not been promising, it does show that having cross-shard state execution capabilities is extremely difficult to achieve. More notably, none of the above shard chains came close to implementing cross-shard atomic transactions. This suggests that it may be extremely difficult for Ethereum shard chains to ever implement state execution capabilities. Thus, it has been proposed to use danksharding instead.
Danksharding allows more space for data to be posted on Ethereum, without needing to be interpreted. These "data blobs" are only checked if they are available, that is, if they can be downloaded from the Ethereum network. This additional space will allow Ethereum to serve as a data availability layer for rollups more cost-effectively. Essentially, these data shards are just meant to hold "data blobs" cheaply, which rollups can reference, while the state execution remains on a single chain.
The zkEVM narrative grew with Polygon's announcement of the Polygon zkEVM launch. Though Polygon zkEVM was more accurately a rebranding of Polygon Hermez, it sparked much discussion on what constitutes zkEVM. More importantly, which of the current efforts were focused on building a workaround and which were actually focused on building an optimized zkVM that could support EVM toolings.
Figure 158: Visualization of zkEVM Source: Foresight Venture
Protocols such as Scroll and Polygon Hermez were much more suited to be EVM-compatible, as they aimed to support EVM at the opcode-level, which would allow their native VM to carry out instructions almost identical to the EVM. However, the extent of EVM-compatibility is not inversely correlated to the VM's performance. There is no data to corroborate that but it is clear from zkSync's 2.0, Polygon Hermez, and Scroll that zkEVM definitions and designs can differ significantly at the zk-circuit implementation aspect and yet, have a similar approach towards conferring EVM-compatibility.
More notably, ZKRs like StarkNet are choosing to adopt a different programming language instead, Cairo. However, there are efforts to either transpile Solidity to Cairo using Warp developed by Nethermind, or run a zkEVM within StarkNet's Cairo VM itself such as Kakarot.
Blockchain Interoperability Solutions
Cross-chain interactions are becoming an integral part of the blockchain space, stemming from the various siloed ecosystems proliferating. This calls for interoperability frameworks that would allow individual blockchains to interact with one another. However, interoperability necessitates trade-offs, such as those introduced by the Interoperability Trilemma, coined by Arjun Bhuptani, the founder of Connext. Generally, the more chains an interoperability protocol supports, the more trust is required.
The most common interoperability protocol today would be a liquidity bridge, as liquidity has been fragmented across multiple siloed chains. Fragmentation of liquidity causes higher slippages for DEX trades and higher costs of borrowing from lending protocols. As such, transferring tokens across chains became one of the commonly used solutions for traders. However, most of these bridges are Arbitrary Messaging Bridges (AMBs) that are capable of transferring arbitrary data across chains.
Figure 159: Value locked in cross-chain bridges in 2022 Source: L2Beat
The TVL of cross-chain bridges fell significantly in 2022, peaking at over $58 billion in January and declining ~90% to $6 billion as of this writing. While this is largely due to the majority of assets on bridges falling in price, there were also numerous bridge hacks in 2022, such as the $600 million Ronin hack and the $323 million Portal hack, as discussed in the DeFi Exploits subsection.
Multi-chain bridges, as the name suggests, are capable of supporting multiple blockchains and their assets. Competition is generally stiff as users are usually not privy to the trust and security trade-offs made, and opt for the cheapest alternative instead.
Figure 160: Value locked in multi-chain bridges in 2022 Source: L2Beat
Multichain is the leading multi-chain bridge by TVL, which mostly comprises USDC, WBTC, and ETH. Thus, it stands to reason that Multichain would facilitate the largest bridging volumes as compared to other bridges from this category. Multichain facilitated over $60 million in daily average volume over the past 30 days to at least 70 different chains.
Figure 161: Value locked in Multichain in 2022 Source: L2Beat
Currently, there are 21 nodes on the Multichain network, which receive and validate cross-chain messages. This poses some extent of centralization risks as the existing Multichain nodes have the ability to censor cross-chain interactions.
Orbit is another multi-chain bridge that supports 12 blockchains, with Ethereum, BNB Chain, and Polygon as the most notable ones. Orbit facilitated slightly over $4 million in daily volume on average over the past 30 days.
Figure 162: Value locked in Orbit in 2022 Source: L2Beat
Portal by Wormhole
Portal by Wormhole is a multi-chain bridge that was primarily meant to bridge liquidity across Ethereum, Solana, and Terra blockchains. However, after Terra crumbled in May 2022, Wormhole was facilitating mostly Solana-Ethereum transfers.
Figure 163: Value locked in Portal in 2022 Source: L2Beat
The sharp dip in TVL in February 2022 was the result of the $323 million Portal hack that depleted all of its liquidity until Jump Crypto intervened by replenishing funds on Portal.
Satellite by Axelar
Satellite is another multi-chain bridge that supports 22 different blockchains, with a focus on blockchains in the Cosmos ecosystem, such as Osmosis and Secret Network. In terms of TVL growth, Satellite is one the few bridges that managed to retain most of its TVL throughout 2022, which is a positive indicator for the bridge's growth.
Figure 164: Value locked in Satellite in 2022 Source: L2Beat
Stargate is a multi-chain bridge for stablecoins, and most of its TVL comprises USDC and USDT. It currently supports 7 different blockchains, averaging roughly $25 million in daily bridging volume over the past 30 days. For a bridge that was launched in March 2022, it certainly saw a relatively high growth in its TVL and usage.
Figure 165: Value locked in Stargate in 2022 Source: L2Beat
Synapse supports 19 different blockchains and has facilitated $4.8 million in daily average volume over the past 30 days. In terms of TVL, Synapse experienced the same generic decline as most other bridges. Synapse holds a large majority of its TVL in stablecoins and ETH.
Figure 166: Value locked in Synapse in 2022 Source: L2Beat
Hop Protocol is an Ethereum-focused bridge, as it only supports chains such as Polygon, Gnosis, Arbitrum and Optimism. Hop airdropped its governance token, HOP, to users and LPs in June 2022, which might explain the small spike in TVL shortly after. However, its TVL has since plateaued.
Figure 167: Value locked in Hop in 2022 Source: L2Beat
cBridge currently supports 36 different blockchains and averages a daily volume of roughly $15 million over the past 30 days. Similar to most other bridges, most of cBridge's TVL stems from stablecoins and ETH, though there is a significant amount of WBTC as well.
Figure 168: Value locked in cBridge in 2022 Source: L2Beat
Aptos Bridge by LayerZero
Aptos Bridge was recently launched and it aims to bridge liquidity between existing blockchains and the new Aptos chain. As Aptos uses a different architecture relative to other chains, LayerZero has built the Aptos bridge to handle these differences. Currently, the bridge liquidity is limited to $1 million per 24 hours after its launch for security reasons but it will be lifted over time.
Figure 169: Value locked in Aptos Bridge in 2022 Source: L2Beat
Across is an Ethereum-focused bridge, supporting only ORs and sidechains on Ethereum, namely Arbitrum, Optimism, Boba and Polygon. Its TVL has been plateauing since June 2022 and even with the recent ACX token airdrop, it did not see a significant uptick in its TVL. This is a concern as a token launch is typically one of the best tools for incentivizing usage and liquidity. Only time will tell if the ACX token can bring sustained growth for the Across bridge.
Figure 170: Value locked in Across in 2022 Source: L2Beat
Two-chain bridges typically facilitate transfer between two chains only. Examples include canonical rollup bridges such as Arbitrum and dYdX, and L1 native bridges, such as the Polygon and Avalanche bridges.
Figure 171: Value locked in two-chain bridges in 2022 Source: L2Beat
Polygon PoS & Plasma
Polygon is an Ethereum sidechain that offers cheaper and faster transactions. The Polygon team has been actively developing a ZKR for Polygon, though none of their ZKR efforts are production-ready just yet. That said, its PoS chain and Plasma chain have accrued significant TVL. The Polygon PoS bridge accounts for more than 75% of Polygon's TVL throughout 2022, which corroborates the fact that most of Polygon's dapps and transactions occur on the PoS chain.
Figure 172: Value locked in Polygon PoS and Plasma in 2022 Source: L2Beat
Polygon PoS chain holds various assets, though it comprises mainly stablecoins as well as ETH and WBTC. This is expected, since these are the most liquid assets and would be necessary to support Polygon's DeFi ecosystem.
Figure 173: Value locked in Polygon PoS by token in 2022 Source: L2Beat
Most of Polygon's MATIC tokens are held on the Plasma bridge, which holds nearly 99% of its TVL in MATIC.
Figure 174: Value locked in Ronin bridge in 2022 Source: L2Beat
Rainbow bridge is the bridge that facilitates transfers between Aurora, Near's EVM emulator, and Ethereum. Most of Rainbow bridge's TVL is in AURORA, the governance token of Aurora, which is a concerning trend, since it means that most of Aurora's TVL is that of an endogenous asset. As the relative composition of other assets has not grown over time, it might mean that the Aurora bridge was not able to bootstrap liquidity from Ethereum. That said, Rainbow's TVL of $55 million is $20 million short of Aurora's TVL of $75 million. This indicates that a substantial portion of the liquidity on Aurora is bridged in from Near.
Figure 176: Value locked in Rainbow in 2022 Source: L2Beat
The Avalanche bridge only facilitates interactions between Avalanche and Ethereum or Bitcoin. As such, it is expected that the bulk of Avalanche bridge's TVL is made up of stablecoins, ETH and WBTC. However, Avalanche bridge's TVL generally declined throughout 2022, likely due to the ongoing bear market, which has taken a toll on Avalanche's DeFi ecosystem as well.
Figure 175: Value locked in Avalanche bridge in 2022 Source: L2Beat
The Gravity bridge is technically a "multi-chain" bridge, as it bridges Ethereum to multiple Cosmos app chains. However, the Gravity bridge is classified here as a two-chain bridge as it facilitates cross-chain interactions only between the Cosmos ecosystem and Ethereum. Gravity's TVL grew slightly in the past three months, which might indicate a growing interest in the Cosmos ecosystem.
Figure 177: Value locked in Gravity in 2022 Source: L2Beat