Aptos vs Sui: The Diem Showdown
ETH has reigned supreme for the last few years but will Aptos and Sui overthrow the king?
Welcome to all the new subscribers over the last month! You can read my last article on zkEVMs here.
Stay up to date with Superfluid by subscribing here:
Back in 2017, during the ICO boom, we saw a new Layer 1 protocol being released almost on a weekly basis with lots of hype and fanfare. 5 years later, most of these don’t exist or are holding on by a thread. The winners of this race are pretty clear. Ethereum reigns supreme, with Solana, Avalanche and Fantom also seeing varying rates of adoption.
Despite this, investors continue to back new Layer 1 protocols in the hopes of replicating insane returns seen by early investors in Ethereum and Solana. Most recently, millions of VC dollars have been poured into two incredibly hyped layer 1 projects in Aptos and Sui. But why? What makes these two projects special enough to attract hundreds of millions of dollars, despite strong incumbent layer 1s that already have developer mindshare
Let’s find out!
Aptos is on a mission to create the ‘safest and most scalable Layer 1 blockchain’ to ensure ‘universal and fair access to decentralised assets for billions of people"‘. Born out of the ashes of the ill-fated Diem project (Libra) led by Facebook (now known as Meta), Aptos is led by Mo Shaikh and Avery Ching. Originally, Diem was perpetually plagued by regulatory drama and red tape, and with Aptos, both Shaikh and Ching have set out to build a Layer 1 chain with a focus on reliability, safety and usability and have thought about these issues through the initial stages of Diem.
Aptos builds upon the work conducted at Diem and uses the programming language Move, originally developed for Diem. Move is based on the existing language, Rust, and seeks to be a safer and more secure smart contract language compared to Solidity.
We’ve seen numerous smart contract exploits and whilst some of them might be down to developer laziness and poor security controls. But some of these are also due to nuances of the Solidity language and the EVM. And this is despite spending months and hundreds of thousands of dollars in smart contract audit costs before deployment.
In response to this, Move is a programming language that is centred around scarcity and access control. Digital assets on Move are represented by ‘resources’. These resources (tokens, smart contracts, etc.) are scarce because they cannot be duplicated unless this instruction has been laid out in the bytecode (instruction) layer. To protect these resources, Move has strict access control. Developers can create modules in Move that can make, store or transfer digital assets. However, these assets can only be accessed by the Module that defined them originally.
Aptos specifically was built with security at its core, so has built upon Move’s security features. Aptos uses a powerful consensus mechanism known as “DiemBFT v4”. The Aptos consensus protocol is deemed to be Byzantine Fault Tolerant (BFT) which refers to the Byzantine General Problem.
Sound like gibberish?
Well, imagine you are amongst a group of Generals, planning out your war strategy. You suspect that someone in the group is a traitor and is sabotaging your plans. As such, the dilemma here is deciding how to make a plan given this suspicion. In a blockchain context, this relates to validators who must confirm or deny transactions correctly, without implicitly trusting other validators. This is a common attack vector for blockchain exploits and one that is negated by Aptos’ BFT protocol.
Aptos uses a reputation-based system to update validator rotations in order to account for non-responsive validators. This piece of automation allows Aptos to validate transactions in under a second. In addition, Aptos is built with strong UX at the core, offering users new, secure methods for private key recovery as well as a key rotation system that can prevent fraud. This is a monumental shift in how crypto wallets work. We won’t need to go hunting through the garbage tip to find old hard drives with crazy amounts of crypto on them anymore!
In addition to offering greater security, Aptos is lightning fast. It’s currently able to achieve up to 130k transactions per second, largely due to separating the consensus protocol and execution pipeline and allowing them to run in parallel. Traditionally, blockchains maintain a transaction execution pipeline and use a consensus protocol to agree upon transactions and their order of execution. Instead, Aptos separates the two processes which allow it to increase throughput and lower latency. Aptos uses Block-STM technology as the execution engine for its smart contracts in order to further increase its throughput.
The Aptos’ mainnet hasn’t even been launched, yet over 100 projects and protocols are building on top of it.
This is largely due to a few reasons:
Aptos has raised a ton of money and is offering a generous grants program to developers and creators to build on top of Aptos
The switching costs from Solana to Aptos are low given that Move is largely based on Rust
At the moment, the Aptos ecosystem is focused on building out basic blockchain projects such as DEXs, NFT Marketplaces, Wallets, Oracles, Explorers and Messaging solutions.
Interestingly Sui is also built on Move (with a few tweaks), and is also led by ex-Meta developers, however, approaches Layer 1 architecture from a different angle. Mysten Labs, the lead developers behind Sui, identified two fundamental flaws in the current Layer 1 design.
On-chain storage is expensive, slow and impractical
Smart contracts have too many vulnerabilities and aren’t cross-platform.
Sui tweaked the Move language to create Sui Move in order to solve for better on-chain storage. Sui also uses objects (similar to Aptos) to represent digital assets and smart contracts. To that effect, each object is owned by an address and is mutable by default. They can, however, be shared between multiple addresses and be made immutable. Move receives objects as inputs, and can manipulate them as needed and returned them as objects. In the core Move language, objects are stored globally by default. However, since on-chain storage is expensive and limited (due to indexing costs) Sui Move doesn’t allow any global storage of objects. Instead, Sui replaces this with its own Sui Storage which interacts with it’s Sui Storage Fund (explored later).
On the consensus side, Sui departs from blockchain norms and categorises its transactions as either Simple or Complex. Simple transactions do not have any dependencies and are usually asset transfers from one user to another. As a result, Sui forgoes using a consensus mechanism to validate transactions and instead uses simpler algorithms based on the Byzantine Consistent Broadcast. This is also extended towards transactions that are more involved and make use of Sui Move’s object and ownership model.
Complex transactions typically involve shared objects where multiple users are able to mutate these objects based on smart contract-specific rules. With these transactions, Sui orders all transactions using the novel Narwhal and Tusk consensus protocol (I highly encourage all readers to check out the Sui docs link to get a full explanation).
Sui’s architecture is purpose-built for dynamic use cases and environments. The Sui team are targeting GameFi, DeFi, Commerce, and Social applications. GameFi dApps will be able to take advantage of Sui Move’s dynamic object-centric model which provides flexibility for digital assets to update and dynamically change based on inputs and actions. On the commerce side, fast finality and parallel execution of transactions will ensure a seamless payment process which is crucial to driving large-scale adoption of blockchain technology.
In my article on Lens protocol, we dove deep into the Lens ecosystem and the ability to basically store everything on a social media platform as an NFT. However, this is expensive to do on the first generation of Layer 1 chains. On Sui, its unique data storage economics allow social media apps to efficiently store all their data directly on-chain.
Unlike Aptos, Sui has released some information about its native token and its respective tokenomics. It’s expected that the $SUI token has 10B in total supply, and will be used for Staking/Security, Transaction Fees, Governance and general exchange purposes. Sui also subverts the traditional demand-supply-based transaction cost system that most layer 1 blockchains use.
Instead, Sui runs on 24-hour epochs. At each epoch, the validators change and vote on a reference gas price for the next 24 hours. The protocol then provides incentives to validators to keep transaction costs close to the reference price throughout the epoch. This results in more stable gas prices, and consistent transaction speeds.
As aforementioned, Sui has unique on-chain storage economics. Traditionally, storage costs are paid when a transaction is written onto the blockchain. However, the validators that process and write data into storage at that point in time may differ from future validators that are required to store this data. As a result, it seems unfair to penalise future users who would need to subsidise past users for their storage.
As such, Sui has instituted a Storage Fund. This fund redistributes storage fees from past transactions to future validators. Transaction costs on Sui can be split into two components, computation cost and storage cost. The storage fees are sent to the storage fund to adjust the share of future rewards distributed to validators. If data is deleted from storage, users are rebated appropriately from the fund. This is in order to incentivise users to seriously consider what data is stored on-chain and to continually monitor the necessity of storing data on-chain over time.
On the ecosystem front, Sui seems behind Aptos in terms of the number of projects built on Sui. From preliminary research, I found a handful of dApps on Sui including Ethos Wallet, Sui Name Service and Sui Gallery. This could be due to a lack of a public grants program and a lower amount of (publicly recorded) funding compared to Aptos. It’ll be interesting to observe the level of activity once an incentivised testnet is released in the coming months.
So why build another Layer 1?
Since Ethereum takes the majority of developer mind share at the moment, it feels non-sensical to build another Layer 1 protocol given so many others have wilted away in the past. Why directly compete with a protocol that is so deeply embedded in a sector that is just emerging and is still searching for product-market fit?
One of my biggest gripes with web3 and crypto is that people tend to either over-engineer solutions in an effort to achieve perfection, or simply fork protocols and make a few changes to seem somewhat original. From the outset, Aptos and Sui seem different from other Layer 1s.
Both stem from the same research base, however, have taken different directions and different architectural choices to reach their current state. In my eyes, this is a win already. Different styles and thoughts need to be celebrated in crypto, where it seems like people gravitate to consensus opinions quickly. Instead, let’s subvert the ETH maxi stance and think about whether Aptos and Sui can help crypto find product-market fit faster, and further unlock more use cases for blockchain technology?
Personally, I believe this might be possible and so it’s worth experimenting with new chains and new ideas.
Moreover, Aptos and Sui have played the game well. Choosing Move as their core programming language has allowed them to attract developers from Solana (which uses Rust). This is potentially unintentional but has seemingly worked well for Aptos with its generous grants program.
What this also highlights is that, ultimately, developers will choose which chain wins. At the moment, Ethereum is the dominant chain due to its time in the market, and greater defensibility with regard to using Solidity as its core programming language. Moreover, it has forced other Layer 1 chains to become EVM compatible, which further solidifies it’s stance as number 1. However, as the date for the Merge comes closer, there are murmurs of unrest within the ETH community with some developers and teams protesting against the shift to a PoS mechanism.
This might provide Aptos and Sui with the perfect chance to take advantage of this unrest and attract talented teams away from ETH. In order to do so, they’ll need to provide a strong amount of support for developers and a community for them to feel a part of. Aptos has a headstart with this through their grants program. Will Sui respond accordingly?
It’ll be interesting to see how this battle unfolds over the coming months and years and whether they will truly be able to challenge the incumbent layer 1 chains for both developer and consumer mind share.
What do you think? Do you think Aptos or Sui will win out? Let me know on Twitter!
Make sure to subscribe now to not miss the next article.
How did you like today’s article? Your feedback helps me make this amazing.
Thanks for reading and see you next time!
Well written and great research! I'd be interested to hear your take on Ethereum's method of scaling through sharding vs Sui's goal of horizontal scaling.
Sui's has some interesting ideas for 24 hour epochs but interested to see how that will work with NFT releases. It seems like a lot of NFT collections work as Veblen goods so the stable gas may destroy some demand for the collection, in turn, lowering the price of the NFT. Any thoughts?