Layer 2 Scalability Explained: How ZK-Rollups Measure Up to Subnets


Recently, popular Layer 1s (L1s) like Solana and BNB Chain have succumbed to network outages as a result of the concurrent running of DeFi protocols, NFT marketplaces, and web3 games on their respective networks. In a similar vein, Ethereum has also been constantly criticized for the network congestion and expensive gas fees that have been plaguing its network. From this, the space began to realize the importance of Layer 2 (L2) scalability solutions.

According to the Ethereum Foundation, rollups are currently the preferred L2 solution for their scalability issues. On top of rollups, other L2 solutions like sidechains and subnets also exist to tackle the scalability trilemma.

In particular, Subnets, Avalanche’s L2 scalability solution, have gained quite a bit of popularity in recent times with the deployment of the DeFi Kingdoms and Crabada subnets earlier this year. For some context, subnets are essentially sidechains, with the exception that subnets can choose to employ validators from the mainnet. Currently, subnets and sidechains seem to be favored by web3 games as their go-to L2 scaling solutions.

With all these different L2 solutions floating around in the space, how do they actually stack up against one another? In this article, we’ll dive into the rollups landscape to discuss the latest developments and pain points within it. We’ll also analyze rollups in relation to custom chains (subnets or sidechains), and look at how these various L2 solutions actually stack up against each other, particularly in the GameFi space.

ZK-Rollups Landscape, Recent Developments, and Pain Points

Rollups serve to reduce gas fees and scale up the capacity of transactions for any given L1 by bundling compressed off-chain transactions into a single order on said L1 network. This way, the L2 can perform computation while the L1 merely has to validate them. Flowing from this, the gas fees will be shared among a bundle of transactions, and thousands of transactions can be finalized en masse.

There are two different variations of rollups: Optimistic Rollups and zero-knowledge rollups (“ZK-Rollups”). The two differ in terms of their verification methods.

Optimistic Rollups assume all the transactions are valid, and post them to the base layer without verification. Witnesses then amend disputed transactions during a seven-day challenge period.

On the other hand, ZK-Rollups create a cryptographic proof and post it to the mainnet as a form of validity proof. Currently, transactions with ZK-Rollups can be finalized in approximately 10 minutes.


From the chart above, and when compared to Optimistic Rollups, ZK-Rollups appear set to empower a faster and more scalable Ethereum ecosystem.

However, it’s worth noting that ZK-Rollups are still a nascent form of technology, and EVM compatibility is lacking. Although zkSync did launch an EMV-compatible zkEVM on their public testnet earlier this year, there’s still a long road ahead with regard to EVM compatibility for ZK-Rollups. Without that, ZK-Rollups will be limited to token transfers and exchanges.

Source: (data as of June 7, 2022)

From the graph above, it can be observed that Optimistic Rollups currently hold a significant 72% of the overall market share in terms of TVL. The aforementioned lack of EVM compatibility appears to have left ZK-Rollups on the sideline, with large DeFi players opting out of ZK-Rollups when expanding into rollups for that very reason.

With the maturation of Optimistic Rollups, most large DeFi projects have already adopted this solution as needed. However, the one-week waiting period often underwhelms new users in the space. As an alternative, when considering GameFi Scalability, NFT marketplaces and web3 applications, ZK-Rollups could be a possible solution to networks that demand higher throughput and cheaper transaction fees.

ZK-Rollups Landscape

Source: and protocols’ websites (data as of June 7, 2022)

As of the time of writing, there are over 12 ZK-Rollup solutions in the space, with underlying technologies developed by StarkWare, Loopring, Matter Labs and Polygon Hermez.

There are two main cryptographic proof methods in ZK-Rollups. The technology adopted by Loopring and zkSync creates a zero-knowledge, succinct, non-interactive argument of knowledge (ZK-SNARK) proof to validate bundles of transactions in the mainnet. ZK-SNARK came to the spotlight in 2012 and was widely adopted by ZK-Rollups pioneers.

On the other hand, a zero-knowledge, scalable, transparent argument of knowledge (ZK-STARK) technology is relatively new and was released by StarkWare in 2018. ZK-STARK was perceived to be more transparent and scalable than ZK-SNARK. The promising technology recently spurred the valuation of StarkWare to $8B, with the funding round being led by Tiger Global.


StarkWare offers two main ZK-STARK solutions: StarkEx and StarkNet.

StarkEx is a Validium solution designed for application-specific ZK-Rollups, adopted by the derivative trading platform dYdX, web3 games, and the NFT Marketplace Immutable X.

Validium is a scaling solution that adopts the same validity proof as general purpose ZK-Rollups, but doesn’t store the data on the mainnet. However, the risk of adopting Validium is that data availability managers could control the users’ assets, raising security concerns. Yet, off-chain data storage bypasses the processing of transactions on the mainnet, and in turn, it remarkably enhances scalability compared to general purpose ZK-Rollups.

Source: (as of 8 June 2022)

StarkNet is a general purpose rollup launched by StarkWare, and completed its launch in February this year. It’s worth mentioning that Immutable X embraces a Volition model which allows users to choose between the Validium model or a general purpose ZK-Rollup. StarkWare has focused on speeding up the TPS of StarkNet to its optimized level as the TVL gradually picks up.

Cairo is the programming language for StarkNet. StarkNet doesn’t natively support EVM, and the Warp team at Nethermind is developing a Solidity-to-Cairo transpiler at the time of writing.

StarkEx’s Web3 Game Footprint

With ZK-Rollups, web3 games can notably slash transaction costs for users, avoid congestion, and reap the benefits from Ethereum’s battle-test security. As such, ecosystems such as Sorare, Immutation X, and more, have adopted StarkWare’s scalability solutions in efforts to support the web3 games on their platform.

Sorare, a famous football card game, gained popularity in the Ethereum mainnet and then moved to L2 by adopting StarkEx. Sorare opted for ZK-Rollups to scale up while making sure not to sacrifice security and decentralization.

Immutable X is a L2 platform for NFT Marketplaces based on Ethereum. It offers zero gas fees and instant trades by moving computation off-chain. Based on DappRadar, four active web3 games currently perform NFT transactions with the help of Immutable X. Gods Unchained is the most prominent one, with around five thousand users and a $1.5 million daily transaction volume at its peak.

Illuvium is the latest web3 game to join the Immutable X platform, hitting the headlines by selling over $72 million worth of digital land plots before its official game release.

In sum, building up games in ZK-Rollups can be done in two ways. The first, through solutions like Immutable X, and the second is by building up a separate ZK-Rollup, similar to what Sorare has opted for. Comparing the two methods, Immutable X’s solution has proven to be more cost-effective. However, it’s important to note that Immutable X is still lacking in adoption, and that ZK-Rollups have yet to gain traction in the web3 gaming space thus far.


Loopring was an in-house developed ZK-Rollup solution. The protocol is the largest general purpose ZK-Rollup chain by TVL, as of the time of writing. Loopring strategically focuses on trading and payments, whereas its efforts to support web3 games manifests only in the building of the Gamestop NFT Marketplace in cooperation with GameStop. There is currently no web3 game running on the platform.

Matter Labs (the Team Behind zkSync)

Matter Labs is the team behind zkSync scaling solutions. It uses ZK-SNARK, and offers off-chain Validium solution zkPorter and on-chain general rollup zkSync.

ZKSpace was forked from zkSync, and adopts the same ZK-SNARK technology to address scalability issues. As pointed out by zkSync, ZKSpace has remained close-sourced since the fork, resulting in the possibility of a security risk. As of the time of writing, ZKSpace (previously known as ZKSwap) owns the protocols ZKSpace, ZKSwap and ZKSea (yet to go live).

Matter Labs launched zkEVM, the first EVM-compatible ZK-Rollup, on the public testnet in February 2022. zkEVM is still under development and is designed to run 99% of Solidity contracts. This signifies a great milestone for ZK-Rollups to run parallel with the Ethereum mainnet and compete with Optimistic Rollups.

Matter Labs is also currently designing a Validium solution, zkPorter. Should Matter Labs intend to lure web3 games onto its platform, zkPorter may be just what it needs. However, Matter Labs still seems to be focusing on building general purpose ZK-Rollups.


Polygon has been quite aggressive in forming its own L2 scalability solutions. Its ZK-Rollups products encompass Polygon Hermez (ZK-SNARK), Polygon Zero (ZK-SNARK) and Polygon Miden (ZK-STARK). Polygon acquired ZK-Rollups scaling project Hermez and startup Mir in 2021, and garnered know-how to build on its ZK-Rollups.

As of the time of writing, only Polygon Hermez is live. It serves as a general purpose ZK-Rollup to DeFi projects with a TVL of merely $366k. Polygon Hermez hasn’t yet attracted many projects, possibly due to a lack of EVM compatibility. However, Polygon may have strategically targeted web3 game developers with its PoS sidechain or supernets, and not yet shifted its focus to ZK-Rollups.

Pain Points and Brief Summary

A recurring issue for general purpose ZK-Rollups lies in their EVM compatibility. We’ve noticed that most DeFi players in Ethereum select Optimistic Rollups or Polygon PoS as their L2 scaling solution. For example, Curve opted for Optimism; Aave V3 for Polygon, Optimism and Arbitrum; and Uniswap has expanded to both Optimism and Polygon. zkEVM is being tested by Matter Labs, and Nethermind is developing a transpiler for StarkNet. It is possible that any breakthrough in EVM compatibility will eventually help to lure DeFi projects to ZK-Rollups in the future.

Another concern is the lack of communication within ZK-Rollups. This causes fractured liquidity and a centralized sequencing process that could induce security concerns. However, Immutable X’s newly unveiled cross-rollup liquidity solutions might bode well for a more liquid rollup ecosystem.

ZK-Rollups have yet to gain traction in the NFT and web3 gaming space, and with good reason. Gods Unchained has continually lost its users, and its trading volume has plunged since its adoption of Immutable X’s solution. In a similar vein, Sorare wasn’t able to repeat its success in the Ethereum mainnet after moving into rollups. Despite the lack of success for ZK-Rollups in the aforementioned examples, in contrast, subnets from Avalanche have found a more solid footing in both NFT and web3 games.

Avalanche Subnets and Polygon

Introduction to Avalanche Subnets


Avalanche’s Primary Network consists of three blockchains — the X, P and C chains. Each chain serves a different purpose in maintaining the integrity of Avalanche, and in creating new assets or blockchains.

· The Exchange chain (X-chain) is the default asset blockchain on Avalanche, and enables the creation of new assets, exchanging and cross-subnet transfers.

· The Platform chain (P-chain) is the metadata blockchain which coordinates validators, keeps track of active subnets and enables new subnet creations.

· The Contract chain (C-chain) is the smart contract blockchain. It enables the creation of EVM-compatible smart contracts.

Using the scaling approach, Avalanche creates subnets to scale horizontally. Subnets differ from a single blockchain, and their relationship is as below:

· A subnet is a set of validators on a blockchain.

· Each blockchain has only one subnet (set of validators), while each subnet can maintain multiple blockchains concurrently.

Each subnet can start off with five validators, but having 10 is recommended for network health and security. The creation of subnets on Avalanche only requires one AVAX. However, each validator of a subnet is also required to validate the Primary Network as well — requiring 2000 AVAX tokens, worth approximately $50k as of June 9, 2022.

Subnet creation is designed to be permissionless and flexible in its requirements to allow for different use cases. For example, private blockchains may require transactions to only be viewed by the subnet validator; or validators have to be KYC/AML verified before being allowed on the subnet.

Web3 Games on Subnets

Subnets have been the holy grail for L2 scalability, particularly for web3 games. Avalanche has been supporting and growing its GameFi ecosystem and finding a home for them in subnets. Crabada and DeFi Kingdoms were their two forerunners, and Castle Crush is currently working on its testnet.


From the table above, Crabada’s share of gas fees on Avalanche’s C-chain tops out at 38.93%. However, similar to many web3 games, Crabada faced the problem of network congestion in mainnets, thus spurring them to move to their own chains, such as subnets. Recently, Crabada moved from the Avalanche C-chain to their very own subnet called Swimmer Network. With its migration to Swimmer Network, Crabada players have seen up to 85% reduction in gas fees, and much shorter transaction times.

Beyond the obvious benefit of lower gas fees and higher TPS, there are also other benefits luring web3 games to adopt subnets. By having their own subnets, web3 games essentially have greater control over the predetermined parameters of the blockchain. It’s possible to have their gas fees be paid in their token of choice. Crabada now uses $TUS as the gas token instead of $AVAX, burning up to 80% of the gas used as the network scales. Token burning curbs token inflation and enhances token utility.

It’s important to note that gaming studios have been building a subnet for a whole ecosystem of games, instead of the typical one-game-per-subnet model adopted by Crabada or DeFi Kingdoms.

Avalanche has also shown its support for web3 games building on their subnets. After its initially planned $190 million Avalanche Rush kick-started the DeFi ecosystem at genesis, the following $290 million Multiverse program aims to accelerate the ecosystem’s growth.

Polygon Supernets and Sidechain

Source: Polygon Blog/Supernets

A competitor to Avalanche subnets is Polygon supernets, a scaling solution from Polygon Edge and Polygon PoS, the prominent sidechain solution. As mentioned previously, Polygon has been rather aggressive with their scaling solutions, with ZK-Rollups that coexist with both supernets and sidechains.

Polygon supernets are still new to the web3 gaming space, and haven’t gone through many practical applications. Yet, the supernet solution is positioned as an alternative to Avalanche subnets with more functions. To highlight, Polygon supernets offer to adopt either proof of authority (PoA) or proof of stake (PoS) chains; choose from a group of professional validators for enhanced decentralization; and the possibility of shared security with the Polygon mainnet.

Polygon PoS has been a hub for web3 games, with successful games such as Aavegotchi and Crypto Raiders seeing great success in 2021. With popular games come both high volume and transaction issues. Similar to Crabada on Avalanche, Sunflower Land caused major network congestion issues, guzzling up to 42% of gas used on Polygon at its peak.

To be brief, it’s still early for Polygon’s various scaling solutions (rollups and supernets). Polygon PoS, the sidechain hosting a myriad of web3 games, is the most formidable choice. However, Polygon PoS has more similarities to a Layer 1 general purpose sidechain, as it owns its own set of validators and still suffers from network congestion. We regard its supernets as being more akin to subnets by Avalanche, and we perform the comparison with ZK-Rollups below.

Comparing Custom Chains to ZK-Rollups

At their core, both L2 ZK-Rollups and custom chains (Avalanche subnets and Polygon supernets) are solutions to the scalability issues from which their respective L1 blockchains suffer. Both have vastly different technical specifications and benefits to using them, and we’ll try to break them down to better understand these differences.

If we were to solely focus on the technical aspects for web3 games to decide between building on L2s or Avalanche Subnets, one might argue that L2s offer a far better solution. However, L2 solutions are still in development, and not yet battle-tested. Subnets seem to gain more popularity for web3 games, due to the demand for their own chains.

In general, web3 games don’t need to interact much with a larger ecosystem as compared to DeFi projects. They typically have closed-loop ecosystems within which their tokens and NFTs support one another. The only interactions outside of their ecosystem would be through on-and-off ramping of their network tokens, as with AVAX or MATIC.


On the other hand, top web3 game studios Gala Games or the soon-to-be-released Myria prefer to build on their own blockchains. These two studios have multiple game titles that interact with one another at some level; thus, it makes more sense for them to build games that exist together on the same blockchain.

Despite this, given that most games are currently made by individual project teams, and not part of a larger collective, there’s minimal collaboration between games. It might therefore be more rational for smaller game studios to adopt subnets, instead of building their own chains.

Beyond the technical specifications, the core difference between Avalanche Subnets and ZK-Rollups is the isolation of web3 games on individual subnets or sidechains. At this stage, Immutable X is still an ecosystem where web3 games and NFTs can interact and build off one another, while Avalanche subnets are individually siloed off into modular blockchains.

How Do Projects Choose Between ZK-Rollups and Subnets?

Due to a lack of composability with the mainnet, subnets and supernets weren’t designed for DeFi projects. EVM-compatible ZK-Rollups fit projects that deploy smart contracts and demand a high level of security. However, as the EVM compatibility of ZK-Rollups is still in the bud, Optimistic Rollups seem to be the only available option for DeFi projects.

It’s apparent that subnets have manifested as better options than ZK-Rollups for web3 games. As demonstrated by the popularity of Avalanche subnets, the separate custom chain for the one-game/one-chain model has gained traction.

ZK-Rollups and Validium may have better technical specifications than custom chains, but they’re still in development. The ZK-Rollups scalability solution doesn’t offer a separate custom chain, and it might succumb to abrupt hype from individual projects. In addition, creating a separate rollup such as Sorare entails high cost — and transition pain.

Source: Immutable X, StarkNet, StarkEx

On the other hand, Immutable X has recently released an innovative Layer 3 solution for web3 games. With this approach, game studios would be able to enjoy hyper-scale solutions in Layer 3 without sacrificing security and composability. The Layer 2 and Layer 3 integration consolidates liquidity across rollup solutions, and bolsters the interaction between DeFi and web3 games. While it still seems to be in its early stages of development, we’re excited to see what else Immutable X has to offer.

In summary, StarkWare and Immutable X’s flying valuations from VC funding have led institutional investors to wager a great deal of money on them. This in turn boosts confidence in ZK-Rollups’ ability to expand scalability within Ethereum’s ecosystem. Moreover, Polygon still takes an aggressive approach to tapping into ZK-Rollups. Hence, it’s apparent that ZK-Rollups’ potential is not yet fully unleashed — and we’ll be looking closely to see how ZK-Rollups flourish in the future.