Deep Dive: A Look Into Layer 2s (Part 2/3)

Published: July 17, 2023   |   Last Updated: January 25, 2024
Andrew Cahill
Andrew Cahill
Data Analyst

The State of L2 Adoption

Key Points:

  • L2s come in two flavors: optimistic and zk-rollups. Optimistic rollups have seen the most adoption to date, but new zk-rollups are off to a strong start 🏎️.
  • Arbitrum has emerged as the de-facto leader in the L2 space. But there is plenty of competition from other L2s 🥊.
  • While L2s are seeing secular growth, their underlying tech is still at an early stage. Making them more secure and decentralized will take time 🧱.

This article is the second in a three part series on Ethereum layer-2 (L2) networks. It compares the underlying tech of top L2 networks and their relative levels of adoption. Part 3 of the series will analyze the fundamentals and tokens (where applicable) of these projects.

Under The Hood Of L2s

The technical ins and outs of L2s are complicated. Even for the Milk Man. So let’s not “nerd out” too hard here. When it comes to classifying L2s, two questions are worth asking:

  1. How does the L2 achieve security?
  2. Is the L2 Ethereum compatible?

How Do L2s Achieve Security?

Optimistic rollups execute transactions assuming no “funny business” (i.e., fraudulent transactions) took place. They look at the milk glass half full.

  • They rely on “watchers” to monitor the network. If a watcher notices something is off, it’s on them to dispute the transaction and submit a fraud proof to get things sorted out.
  • Some optimistic rollups (e.g., Optimism) don’t have these fraud proofs active (yet). Instead, their security relies on a multisig wallet controlled by a small group of anonymous individuals.

ZK-rollups rely on mathematical proofs, also known as validity proofs, to secure their networks.

  • If the logic behind the validity proofs is sound, there’s no possibility that fraudulent transactions could be finalized on these networks.
  • All of the zk-Rollups mentioned in this article currently have live proving systems. Phew.

Both approaches “piggyback” on Ethereum’s security by posting data from their chains (i.e., user token balances and, in some cases, transaction data) to Ethereum L1.

Ethereum Compatibility

If an L2 is “Ethereum compatible”, dApps on Ethereum L1 can easily be ported over to the L2. Think of it like a copy-pasta job (with some edits required). This makes life easy for developers – no need to re-write working dApps from scratch. It also makes life easier for L2s looking to rapidly build out ecosystems.

Optimistic rollups have had Ethereum compatibility since they launched in 2021. Coinbase’s upcoming Base L2 will have it when it launches in August 2023. Historically, ZK-rollups have not been EVM compatible. But that changed this year when the first zkEVM-based rollups went live.

In just the past ~6 months, three zkEVM-based rollups (Linea, Polygon zkEVM, and zkSync Era) were launched.

These networks have generated a lot of buzz as they aim to bring “the best of both worlds” – namely, high security (via validity proofs) and Ethereum compatibility (via zkEVM).

Other zk-rollups have opted to go an entirely different route. For example, StarkNet has outlined its reasons for “opting out” of Ethereum compatibility as part of its scaling strategy.

“Solidity contracts on [Ethereum] layer-one were written under a very particular set of computational constraints … moving over with the same set of constraints [to L2] that you’ve lived under on layer 1 makes very little sense.”

Uri Kolodny, CEO at StarkWare (Delphi Podcast, October 2021)

While there could be benefits to this “long game” approach, applications deployed on StarkNet need to be deployed using its Cairo development environment.

What’s Better? Optimistic or ZK?

Debate over which approach is the most secure, scalable, and user-friendly has been heated. Given the complexity and early development stages of these L2s, it will continue raging on. And crowning a “winner” before we see both of them operate at scale and in production for a while seems premature. So, here’s what’s top of mind for the Milk Man:

  • Both approaches have demonstrated they’re capable of delivering scaling gains.
  • Both still have a ways to go before they are “sufficiently” decentralized and secure.
  • With zkEVMs live, both are now capable of being a home for apps that are currently deployed on Ethereum L1.

Let the battle continue🍿.

Current State Of L2 Adoption

Note: All data in this section relates to L2s that are currently live in production environments. Hence, recently launched L2s (e.g., Linea) and soon to be launched L2s (e.g., Base) are not included in the analysis.

Given their head start with Ethereum compatibility, Arbitrum (Arbitrum One) and Optimism (OP Mainnet) have built out the biggest L2 ecosystems to date.

  • They currently have ~$3.1B and ~$1.0B of TVL in applications on their networks, respectively.
  • That translates to ~12% and ~4% of Ethereum’s current ~$25B TVL.

Because they just recently launched with Ethereum compatibility, zkSync ERA and StarkNet have relatively small ecosystems compared to Arbitrum and Optimism.

  • zkSync ERA has ~$200M of TVL which translates to ~1% of Ethereum’s TVL. Though, that number has been increasing quickly!
  • StarkNet has ~$30M of TVL which translates to 0.1% of Ethereum’s TVL.

TVL helps approximate ecosystem size. But there’s also another (simpler way) to analyze L2 usage: transaction counts.

  • Ethereum L1 has processed an average of ~1.1M transactions per day this year.
  • Top L2s have been processing ~500K to ~1M daily transactions per day this year. But on certain days, Arbitrum has processed ~2.5x more transactions than Ethereum layer-1.

What does that mean?

First off, these L2s are actually delivering scaling gains. Another way to see this is by looking at their max throughput, or transactions per second (TPS). Arbitrum and StarkNet have posted higher TPS (on certain days) than Ethereum ever has.

L2s can kick it into “high gear”. Well, relative to Ethereum at least.  

Secondly, the “velocity” of economic activity on L2s is higher than Etheruem L1.

  • Think about it this way: The largest L2 ecosystem (Arbitrum) is currently ~11% the size of Ethereum’s based on TVL. Yet its users are submitting just about as many (and on some days more) transactions than Ethereum users. Wowza.

But let’s not get too ahead of ourselves here. There are still some pretty big challenges for these L2s.

Challenges And Catalysts For L2 Adoption

Let’s start with the challenges.

1) Poor User Experience.

If you’re gonna use an L2, you’ll probably need to use a bridge (or several bridges). Anyone that has used a crypto bridge knows its a pain in the *ss. It takes time. It takes money (i.e., transaction fees). It takes extra clicks.

The process is especially painful when it comes to getting funds off optimistic rollup L2s and back to Ethereum. If you’re using a canonical bridge (the most secure option), it would take you ~7 days to get your money off an optimistic rollup. That’s an eternity in crypto.

The good news is that third party bridges (e.g., Axelar, LayerZero, Wormhole) have emerged to drive these withdrawal times down. But some of them have also suffered some of the biggest hacks seen to date. Safe bridging, frens.

2) Remaining Centralization And Security Risks

Today, most rollups rely on one single entity (i.e., sequencer) to process transactions. In some cases, that sequencer can order transactions as well. Yeah, that’s right. One entity that is critically important to you getting your transactions confirmed. And in some cases, also capable of extracting MEV from you. Yikes.

Put more bluntly, forcing an L2 out of operation would be way easier than forcing Ethereum out of operation. The tech underpinning L2s is still a work in progress (WIP). Crypto data provider L2Beat put together a classification system which outlines the state of maturity of most L2s. It classifies L2 tech into three stages:

  • Stage 0 (“Full Training Wheels”) where the L2 is pretty much run by one sequencer.
  • Stage 1 (“Limited Training Wheels”) where the L2 is partially governed by smart contracts and less reliant on the sequencer.
  • Stage 2 (“No Training Wheels”) where the L2 is fully governed by smart contracts and not reliant on one or few sequencers.

Of the five top L2s in production, only Arbitrum has reached “limited training wheels” status so far.

For those thirsty for more insights on this risk scoring criteria, check out l2beat.

Now onto the catalysts. As we pointed out in Part 1 of this series, there is a “Great Migration To L2” underway. Ethereum L1 can’t scale on its own. Applications and users are slowly but surely making their way to L2s. That’s an organic catalyst. But there are also a couple other things worth mentioning.

1) New L2 Launches Could Speed Up L2 Adoption

Up until now, L2s have been primarily used by crypto native users. They know how to bridge. They know how to scoop up airdrops. They’re not afraid to try new things. But with Consensys (Metamask’s parent company) launching Linea and Coinbase launching Base, we could see more “normies” transacting on L2 in the future. Both of these companies have direct relationships with their customers and will be providing avenues to onboard onto L2.

If only a small percentage of their massive user bases adopt L2s, that could be a big tailwind for all L2s. Whenever Coinbase does something, it’s a big deal.

2) Airdrops And Token Launches

Who doesn’t love tokens? As we pointed out in our “Airdrop Aspirations: zkSync” guide last week, L2s have seen some of the biggest airdrops to date. More appear to be on the way – which will encourage users to continue testing these networks out. Additionally, L2 airdrop criteria have been geared towards incentivizing people to use the networks on a recurring basis and actually become acclimated to it. So, maybe some of these airdrop farmers will actually stick around.

What to expect in next week’s Look Into L2s (Part 3)? An overview of the fundamentals and tokens (where applicable) of L2 networks.