In-depth analysis of the differences between Meta's public chains Aptos and Sui

Author: State.io

Compiled by: The Way of the Metaverse

About two weeks ago, media reports indicated that Mysten Labs was preparing for a Series B funding round, which sparked considerable interest. This led to a flurry of Twitter discussions surrounding Mysten Labs' Sui native L1 blockchain. On the surface, Sui and Aptos share striking similarities—most notably, both projects are initiated by teams of former Meta executives (formerly Facebook). Therefore, I find it easy to compare the two.

It is important to note that a direct comparison is unlikely until both L1 blockchains launch their mainnets. The information disclosed by the two is also different; for example, Aptos's economics will not be made public before its mainnet launch, while Sui has already released its economic white paper.

Let's begin…

Overview of Mysten Labs & Sui

Mysten Labs was founded by the Novi research team at Meta, which also worked on Meta's Diem blockchain. Like Aptos, Sui has no affiliation with Meta, although it is also composed of former Novi-Meta personnel.

Mysten Labs designed its own high-performance Layer 1 ("L1") proof-of-stake ("PoS") blockchain, Sui. While Sui aims to address the blockchain trilemma, its focus is on scaling composable and dynamic NFTs for a wide range of metaverse applications, including gaming, social, and commerce. In contrast, Diem was initially designed to handle lightweight payment flows between a small number of custodial wallets (10 to over 100). Diem's original architecture could not support mass adoption.

Additionally, the team has established partnerships with several protocols to apply its research on scalability and consensus improvements. These partnerships require open-sourcing and deploying its Narwhal & Tusk consensus mechanism (discussed below) as well as the programming efficiency of Move. So far, there are two formal partnerships—Celo (a mobile-first, EVM-compatible network) and Sommelier (a transaction co-processor for Ethereum).

Team: Before founding Mysten Labs/Sui, the project's co-founders all worked at Novi:

• Evan Cheng, CEO of Mysten Labs: Former head of R&D at Novi and engineering director at Meta (programming, languages, and runtime); worked at Apple for 10 years
• Sam Blackshear, CTO of Mysten Labs: Former chief engineer at Novi, focused on the Move programming language
• Adeniyi Abiodun, CPO of Mysten Labs: Former product lead at Meta Novi; previously worked at VMware, Oracle, PeerNova, HSBC, JPMorgan
• George Danezis, Chief Scientist at Mysten Labs: Former research scientist at Meta Novi; previously worked at Chainspace, Microsoft
• Kostas (Konstantinos) Chalkias: Former chief cryptographer at Novi; previously worked at R3, Erybo, Safemarket, NewCrypt

Interestingly, the Mysten Labs team seems to have managed to recruit more crypto human capital from Meta. Team continuity and maturity have been attractive factors for startups, and Mysten Labs benefits from this.

Funding Situation

Series A

• Date: December 2021
• Funding Size: $36 million, Series A
• Asset Type: Equity financing
• Notable Investors: Led by a16z and NFX, Scribble Ventures, Redpoint, Lightspeed, Electric Capital, Samsung NEXT, Slow Ventures, Standard Crypto, Coinbase Ventures, among others

Series B

• Date: Currently TBD; in July 2022, news of this funding was leaked to the media
• Funding Size: $200 million, Series B
• Asset Type: Equity financing
• Notable Investors (rumored/known): FTX Ventures
• Expected total fundraising of $236 million.

Economics

White Paper Summary:

• Code: $SUI
• Supply: Fixed total of 10 billion; not yet released. A portion of the total supply will begin to circulate at the mainnet launch, with the remainder unlocked or distributed as future staking rewards over the coming years.
• No deflationary/burning mechanism
• Utility: 1) Staking for network security, 2) Transaction fees, 3) Storage fees, 4) Governance voting, 5) As a medium of exchange for Sui native assets. More details on 2) and 3) to come.
• More details pending ("TBD")

Similarities with Aptos

Sui and Aptos, aside from both teams being spun off from Meta, have some surface-level and fundamental similarities, summarized as follows:

1. Both are committed to solving the blockchain trilemma.
2. Both use Move as their native programming language, although Sui's object model differs slightly from Aptos.
3. Parallel transaction execution and consensus based on software transactional memory ("STM") (achieving low latency and higher throughput)—as opposed to ordered transactions and sequential execution blocks in other L1 designs.
4. State synchronization to optimize data availability.
5. SDKs and APIs for improving DX/UX.
6. Tier 1 venture capital investor caps.
7. Valuation (more or less).

Advantages (Different from Aptos)

Based on publicly disclosed information today, key differences warranting further investigation lie behind the scenes. They appear as follows:

Gas Mechanism: Sui is able to maintain low and predictable gas prices while incentivizing validators to optimize transaction processing and prevent denial-of-service ("DoS") attacks.

At the start of each epoch, validators vote on a gas price across the network, which serves as a reference for users when submitting transactions for processing. Validators are incentivized to keep gas prices near the reference price for each epoch and respond accordingly. Those who submit high bids will be penalized with discounted rewards. Similarly, unresponsive validators receive relatively fewer rewards, and the risk of delegating stakes is reduced as users want their transactions processed effectively. Additionally, predictable gas facilitates predictable transaction processing. Even during high activity periods, gas prices can remain low because throughput can linearly scale with more machines from each validator; validators can add more machines accordingly to meet increased network demand. Traditional blockchains are typically designed to run on a single machine for each validator (even on a single CPU).

A unique aspect of Sui's design is that storage fees are paid separately from transaction execution. In principle, storage costs are denominated in off-chain dollars for storing arbitrary on-chain data. It is determined through governance polls as an exogenous pricing mechanism. From the user's perspective, they do not distinguish between the fees paid. However, when users submit on-chain data, they pay both gas and storage fees simultaneously. The storage portion of the fee is deposited into a "storage fund" to cover the costs of validators storing user data. This fund is used to subsidize future storage costs that increase as the network develops and matures. Conversely, users can delete on-chain data to offset storage fee rebates, doing so for economic optimization.

Here is an overview of Sui's economics:

Transaction Execution: Consensus paths are divided into the following two types of transactions:

1. Simple Transactions: Defined as independent transactions with no causal relationships, using Byzantine consensus broadcasting within the transaction. It is a leaderless protocol that eliminates the overhead of global consensus without sacrificing safety and liveness guarantees. Faulty validators do not significantly impact performance, while most leader-based blockchains see throughput drop and latency increase when one or more validators crash. Transactions are optimistically validated and executed in parallel individually, rather than sequentially or in batches like most traditional blockchains. This is nearly instantaneous. Most transactions fall into this category, such as when users simply want to transfer assets to a recipient, where the only required data is the sender's account, with no interdependencies with any other part of the blockchain state.
2. Complex Transactions: Defined as interdependent/entangled with shared objects and using a variant of Byzantine Fault Tolerance ("BFT") protocol. Sui uses Narwhal to order and execute them sequentially (more details below). Transactions are validated within 2-3 seconds.
3. Transaction dependencies need to be explicitly stated. The consensus subsystem can also scale as it can order more transactions by adding more machines for each validator. This allows Sui to scale.

BFT Consensus: Known as Narwhal & Tusk. Similar to Aptos, Sui's consensus protocol is also a derivative of HotStuff. Both Sui and Aptos's consensus mechanisms minimize the communication required between validators to process transactions for lower latency. AptosBFT is partially asynchronous. Sui's design for implementing consensus separates the protocol's memory pool from the consensus layer, while most PoS L1s have monolithic consensus protocols.

• Narwhal (memory pool protocol): The memory pool stores unconfirmed blockchain transactions until they are validated by the consensus protocol. Narwhal's task is to broadcast transactions for consensus while maintaining data availability. Narwhal is independent of the consensus protocol and can tolerate forms of asynchronous or intermittent liveness loss.
• Tusk (asynchronous consensus protocol): Responsible for ordering transactions. Tusk uses shared randomness to eliminate extra communication between nodes and allows each node to determine the total order of transactions. When combined with Narwhal, it ensures high performance in the event of failures.

Narwhal & Tusk can be composed with other Byzantine Fault Tolerance ("BFT") consensus protocols. The team has open-sourced this code to allow the broader crypto community to benefit from their research (starting with their partnerships with Celo and Sommelier).

Theoretically, the throughput scalability of this design has no known upper limit. As there are no testnet results, it will be interesting to see how it performs in practice.

Sui's security properties can withstand adverse network conditions, network partitions, or DoS attacks on validators, as it does not impose any synchronization assumptions on the network. In contrast, sustained network attacks on synchronized blockchains (i.e., most proof-of-work blockchains) can lead to resource double-spending and deadlocks.

Hardware Requirements: The requirements for running a full node can provide a sense of decentralization, although it is still too early to determine at this stage. Below is a summary of the requirements for Sui, Aptos, and other L1 peers:

As of June 30, 2022, over 5,000 Sui nodes are operating in 65 countries. When both Sui and Aptos launch their mainnets, other factors driving decentralization will be discussed at a later stage.

Sui SDK: While Aptos also has an SDK to improve DX/UX, Sui's SDK makes interesting attempts to connect to other ecosystems and non-crypto use cases.

• Game API: Directly connects game developers to a general L1, rather than a sidechain or game-focused L2. Game developers will then be able to seamlessly interact with Sui's ecosystem (users, other dApps, assets) and the Move language.
• "SuiEcho": Allows dApp developers to guide the community by facilitating the portability of digital assets from other ecosystems to Sui. For example, Ethereum-based Bored Ape Yacht Club NFT holders can use their ETH-native PFP as proof of ownership to mint Sui-based equivalents through SuiEcho. While NFTs minted on Sui are independent assets, their avatars can be used within Sui's network. Similarly, it allows Sui dApps to enter communities in other networks. While high-value NFTs can bridge elsewhere, value ultimately accumulates on the original chain. However, this approach allows Sui and newer L1s to benefit from existing communities from another L1. The Aptos SDK also has NFT functionality, but it is currently limited to Aptos-native capabilities.
• "Handshake": A front-end tool that serves as a channel for users to distribute, claim/redeem Sui digital assets (e.g., payments, merchant coupons) to both crypto and non-crypto users. For non-crypto users, there is no need to pre-create accounts or interact with the underlying blockchain. The roadmap for Handshake includes onboarding non-crypto users.

Performance: As of June 29, Sui has just begun launching its incentivized testnet. Once more testnet data is available, I will update this benchmark again. Otherwise, the following reference points for available throughput include:

• An unoptimized 8-core MacBook Pro achieving 120k TPS
• Achieving 130k TPS when Narwhal is combined with Hotstuff, cited as verified evidence in Mysten Labs' consensus research
• 160k TPS when Narwhal is combined with Tusk

However, Aptos's Avery Ching stated that these throughput numbers do not represent the end-to-end throughput of the blockchain, as they only consider consensus, network, and some storage aspects. They do not take into account other important factors, such as transaction execution time in the production blockchain, account access patterns, or authenticated data structures (e.g., Merkle trees). In terms of transaction finality, block time is often mistakenly equated with finality, whereas, in reality, block time is just one input factor for finality.

Ecosystem: As of July 25, see below for a comparison of key alternative L1s based on GitHub stars, Discord members, and Twitter followers. Note that Ethereum is not included due to its well-known dominance; also because these new L1s are informally viewed as "SOL killers."

Sui DApp:

• Sui Wallet (native wallet)
• Ethos (wallet)

Aptos DApp:

Other Aptos-funded program participants: Martian DAO, Solrise Finance, Protagonist

Final Thoughts

For both L1s, it is still early days, and the smart contract market is undoubtedly competitive. Both companies were founded around the same time. Similarly, they are only a few months apart. While there are many overlaps between Aptos and Sui, there are also some interesting technical differences.

The technology is compelling, but it ultimately needs to solve meaningful problems that give people reasons to pay for it; technological innovation must be a sufficiently significant improvement to justify the transition costs from existing technologies. It is also not the only single determining factor for success—Jason Choi has stated that a technological edge alone almost never wins—success also depends on business development and market entry. Technology needs to translate into performance, and then into results.

When considering multiple aspects of each ecosystem, Aptos stands out compared to Sui and other competitive L1s. In a short time after launch, despite having only a few months of first-mover advantage over Sui, its ecosystem is larger:

• ~2 times the community/users (based on Discord members and Twitter followers);
• About 3 times the developer activity (based on GitHub stargazers);
• Similar throughput size, but largely still unclear;
• More metrics can be analyzed post-mainnet launch, such as TVL, active users (wallet addresses), active nodes, minimum staking requirements, etc.

Overall, when conducting a layered valuation at the time of writing, Aptos appears to demonstrate relative value given the progress of its ecosystem, based on the latest funding rounds. Note that Aptos has just raised a total of $150 million in Series A funding at a valuation of at least $2 billion (total funds raised of $350 million). For a $2 billion valuation, Aptos offers more in terms of a strong community, developer activity, and dApp ecosystem, while being more advanced in its testnet. Sui needs to close the gap with Aptos on every metric to justify its valuation. This will come down to the primary principle of attracting quality developers.

That said, their team is credible, and I have no doubt they will be able to execute their vision. They are well-equipped in terms of pedigree, skills, and resources. While this post is more of a comparison between Sui and Aptos, Sui's enhancements undoubtedly fill a gap in the current L1 smart contract market. If they successfully execute their plans, they are likely to be competitive post-mainnet.

Some industry participants are frustrated with new L1 entrants like Aptos and Sui. However, the reality is that there is currently no chain capable of mass adoption. While existing chains (L1s and L2s) will certainly benefit from first-mover advantages, new entrants (Aptos, Sui) are closing the gap—assuming they can address substantive scalability pain points and provide strong DX/UX. Nevertheless, the market may be large enough to support a multi-chain future. The key question is what it will look like and how many there will be.

Sui's testnet results should provide a better sense of performance. Similarly, it will be interesting to revisit this comparison after the mainnet launch to analyze L1 performance/attributes across more metrics, including their token economics, unlocking schedules, and stakeholder distributions. Time will tell how the underlying technology will drive overall DX/UX and who will attract more users/liquidity.