This is an opinion piece and is not to be taken as financial advice. All opinions are solely the author's.

Monolithic -> Modular

The blockchain scalability issue has been a persistent one and has arguably been the driving force behind most developments in blockchain architecture. The quest for a highly scalable Ethereum network led to the rise of proof-of-stake (PoS) blockchains, sidechains (Polygon, Ronin), Layer 2-(L2) solutions such as rollups (Optimism, Arbitrum), and other modular solutions (Celestia, Fuel, Nitro, Eclipse, EigenDA).

Other blockchain ecosystems with their own take on scalability have also materialized, for example Cosmos and its appchains, Polkadot and its parachains, and Avalanche and its subnets, among others. Each of these solutions have their advantages and disadvantages and are not perfect by any means. Of all these scaling solutions, I believe Cosmos’ Appchain Thesis to be the near to mid-term winner.

The Appchain Thesis is essentially this: rather than trying to scale an L1 to support multiple protocols and their growing economic activity, why not build a new L1 customized and optimized for each protocol and connect them all together? The Tendermint + Cosmos SDK stack is unrivalled in the ease that it allows anyone to spin up a fully operational blockchain — from joke chains within hours (Joe Chain) to complex high-performance DeFi-specific chains (e.g. Sei Network). dYdX was the first major DeFi protocol from Ethereum to signal its intention to build an appchain, proving the PMF of appchains. Most recently, Circle announced its appchain to finally bring native USDC to Cosmos.

We’re now seeing Ethereum move towards this same appchain blueprint. The post-Merge roadmap showed that scaling mainnet in the near future will revolve around optimistic rollups (ORUs) and zk rollups (zkRUs). Take that one step further and you have app-specific rollups, or RollApps, the rollup version of appchains.

Gone are the days of monolithic L1s. A host of modular solutions (image below) make building blockchains like Legos, allowing developers to separate the Settlement, Execution, Data Availability, and Consensus layers. Dymension are taking the Cosmos blueprint and applying it to rollups, making it easy for any team to spin up their rollups with ease — just like appchains on Cosmos.

One downside to the Appchain Thesis is that it is done at the cost of synchronous composability. More on that below as we take a deeper look into Cosmos.

The State of Cosmos

As of the time of writing, there are 53 chains in the Cosmos ecosystem connected via IBC — a 65% YoY increase from last year’s 32. The ecosystem has seen a huge drop in IBC volumes since the Terra implosion, which makes sense since Terra accounted for most DeFi activity on Cosmos. However, the launch of perpetuals, money markets, and other lending protocols across the Cosmos recently and in the upcoming quarter could be a catalyst for user activity. We’re also are seeing efforts to connect:

1. IBC chains with non IBC chains (Polymer)

2. Cosmos SDK chains with non-Cosmos SDK chains (Penumbra)

3. EVM and Cosmos chains (Electron Labs)

Reference: IBC Protocol Review 2022

Despite the apparent PMF of appchains, I am personally looking at DeFi-specific chains like Kujira, Sei, Injective, and even Terra, who have been continuing to build innovative infrastructure and tooling after the crash. Delphi’s Astroport and Mars protocols remain on my watchlist. The reason why I am bullish on “DeFi-specific” chains and prefer the idea of such (as opposed to “application-specific chains”) is that meaningful economic activity can only occur when there is an ecosystem of products that users can use. “But Ian, isn’t that the whole reason for IBC, that appchains can talk to each other and are collectively the Cosmos ecosystem?”

Well, the current state of IBC relayers is not conducive to achieving synchronous composability between appchains. It may enable asynchronous composability, but it is not enough. Synchronous composability is when interactions between applications can occur within the same block. This is vital for trading and yield strategies to reduce slippage and prevent frontrunning. IBC enables asynchronous composability, which is composability between protocols where interactions occur within an undefined and variable amount of time. IBC relayers are slow, unprofitable to run, and unreliable — thus are only useful for transferring assets between chains and is not ideal for things like cross-chain trading or advanced yield strategies incorporating protocols on multiple chains. While ICS 29 may help improve reliability, reliability alone won’t be enough if transactions between chains take longer to settle than on-chain transactions.

As such, DeFi activity on Cosmos in the near future will be confined to DeFi-specific chains. Just to prove that I’m not crazy to think so, Mars Protocol in their whitepaper, states their plans to launch native outposts on various chains to achieve synchronous composability, as it is simply not possible with cross-chain calls. However, we can look forward to improved asynchronous composability with developments like Axelar’s GMP, which Squid is using, or Circle’s CCTP. Keep an eye also on cross-chain messaging protocols like Router Protocol, Synthr, etc.

A Modular Future

So which blockchain scaling method is better? This sounds like a cop out but I do think it depends on what each protocol needs. The following opinions are based on the state of Ethereum L2s and Cosmos appchains at this point in time.

Scenario 1: If a protocol requires a highly-performant chain that can be customized and optimized for their needs, then a Cosmos appchain would be my pick. This is because I believe that nothing yet rivals the customizability of the Cosmos SDK + Tendermint stack. The team at Sei Network have shown just how this is the case with their work on intelligent block propagation and optimistic block processing. With Celestia coming online soon and Nitro building the first Cosmos L2, this will kickstart the modular era of Cosmos appchains and provide more customization for developers.

Ethereum L2s should not be underestimated however, as application-specific rollups are coming very soon. Modular blockchain solutions like Celestia, Fuel, Eclipse, EigenDA, etc are all coming online to facilitate this. However, “RollApps” will suffer the same fate as Cosmos appchains in terms of having asynchronous composability at best, at least for now.

Scenario 2: If a protocol requires proximity to Ethereum for its network effects and superior liquidity, then an L2 rollup makes more sense. This is not to say there is a lack of customizability — in fact, the host of modular solutions as discussed previously points to exactly the opposite. I just think that the Cosmos stack has had much more development behind it, has had ample time to deal with critical security issues, plus upgrades from various chains can be up-streamed to the rest of Cosmos since each chain is using the same stack.

Thus, I am fairly certain that the space is large enough for both ecosystems (Ethereum and its rollups and Cosmos and its appchains) to do well. IMO, the near future of blockchain scalability will converge towards one that is modular and filled with DeFi-specific "appchains" and "rollapps". In the far future, both ecosystems (and others) will ultimately converge and be completely composable.