# Sui: Web3 at ease of Web2 By [notjuve](https://paragraph.com/@notjuve) · 2024-05-21 --- Key Takeaways: -------------- * Sui uses Narwhal & Bullshark consensus to decouple transacting broadcast and metadata ordering. This reduces consensus overhead and optimizes network resource utilization for higher throughput. * The asset-centric model allows for secure, composable, parallelizable, and data-rich applications. Gaming and Socialfi applications could be their “Only Possible on Sui” moment. * Move is built from the learnings of smart contract applications on Ethereum & Solana. Its type, memory, and resource safety enable secure applications with a superior developer experience. * Sui announced a 157M SUI DeFi incentive program, which has boosted the TVL from $30M to $750M. * Sui allows users to create wallets and make payments natively using a Web2 login. Its focus on gaming and payments demonstrates a clear commitment to onboarding consumers to crypto. * The key challenges for Sui are developer adoption, centralization risk, token unlocks, a lack of crypto-native GTM, and competition. Meta’s Diem and Novi projects aimed to use crypto to provide financial services to the masses. After their shutdown, Sui carries the torch, aiming to simplify the user and developer experience to enable the next generation of consumer crypto applications. Novi’s core team formed [Mysten Labs](https://mystenlabs.com/) in Dec’21 and raised $340 million in 2022 to build Sui. The team spent over 1.5 years addressing Diem’s issues to create a secure and powerful layer-1 platform with high throughput and low latency. Sui is a delegated proof-of-stake system with 81% of all $SUI staked. The protocol uses a DAG and an asset-centric model to enable parallelization. They bring several innovations across the stack; let’s look into them: 1. Networking - Narhwal 2. Consensus - Bullshark 3. Ledger - Object/asset-based 4. Execution - MoveVM & Move Networking & Consensus ====================== [Narwhal](https://arxiv.org/pdf/2105.11827) mempool is a leaderless, round-based DAG that decouples transaction broadcasting and ordering. This reduces the consensus overhead and optimizes network resource utilization for higher throughput. In Narwhal, each validator proposes a block at round R once they receive certificates for n-f blocks from the previous round. Unlike blockchains, which only have one block per slot, a DAG can include all honest blocks that were effectively broadcast. Here is how it works: * **Data dissemination**: Validators broadcast their bundle of transactions and certificate list of n-f certificates from the previous round to all validators. This parallelization improves resource utilization and network throughput. * **Verification**: Receiving validators sign blocks after performing safety and validity checks. * **Certification**: The client/full node creates a certificate of availability with signatures from at least 2/3rds of the stake majority. This is sent back to all validators, and a new round starts. ![](https://storage.googleapis.com/papyrus_images/9bceed7f222098da1e1d8bfd5e873c5df718ed2f26491888603aef1a1c0f753d.png) Narwhal allows asynchronous communication, and the latency depends on the actual network delay rather than the upper bound assumed under synchrony. However, the implementation of transmitting blocks, signatures, and certificates sequentially introduces extra latency. Narwhal-based systems have 2-3s latency vs. ~1s latency with Hotstuff. A major advantage to Narwhal’s reliable storage and broadcast is that it allows for a consensus-less fast lane with <500ms latency for simple actions on user-owned assets. Transactions like payments that don’t require total ordering can be executed after they have been broadcast. In comparison, Solana has no global mempool, and messages are forwarded to a deterministic leader. This reduces networking latency but increases leader resource utilization, decreasing throughput. This type of network architecture also makes Solana susceptible to DOS and liveness failures. **Note**: Narwhal has quadratic communication overhead with respect to the validator set size; this limits the total number of active validators. Currently, Sui has ~105 validators and requires a minimum of 30M SUI (~$30M) to validate the chain. * * * Bullshark --------- Narwhal forms a consensus base, which can be paired with a consensus protocol to offer higher throughput and better liveness under fault/asynchronous conditions. A key challenge is that validators have different local views of the DAG and must be aligned for a global order. Mysten Labs has worked on consensus protocols like DAG-Rider, Tusk, and [Bullhsark](https://arxiv.org/pdf/2201.05677) that allow ordering from the local view with zero communication overhead. Sui currently uses Bullshark, an extremely simple consensus protocol built on Narwhal with just 200 lines of extra code. An Anchor (leader) is selected deterministically for every odd round and voted upon during an odd round. A leader that receives f+1 votes can commit their block and the entire causal history of reference blocks using a deterministic order. An anchor can be skipped during asynchronous conditions, but it can be reincluded later, or a fallback anchor can be included. Bullshark also provides a garbage collection method to cut off dead branches to stop the DAG from growing infinitely. ![](https://storage.googleapis.com/papyrus_images/85b9102990891bdeb837d35786be59ebcee0e60aab5400dc2c96ecc0fce6abd3.png) The key advantage is that Bullshark uses the formation of the DAG for consensus and doesn’t require any extra overhead for a global sync. This significantly improves upon previous DAG systems, which required syncing different local views or extra communication in the form of voting for global ordering. Mysten Labs has launched a new version of their consensus Mysticeti on testnet. It improves upon N/B by: * **Remove Certificates:** This removes the round-trip communication of signing and receiving the certificate, helping reduce latency by a few hundred milliseconds. * **Multiple Anchors and frequent commits:** Multiple anchors reduce the problem of missing or skipped validators and increase the likelihood of a block being committed. Blocks will be committed every round instead of once every two rounds. This allows Sui consensus to scale to 100,000+ tps while maintaining sub-1s latency and <250ms for the fast lane, a major improvement upon N/B with ~2s latency. Sui also offers linear scaling, where you can add more machines per validator to perform more tasks in parallel. With five workers per validator, SUI can multiply its throughput by 5x. **Note**: The TPS figure is theoretical and considers operations as transactions; the mainnet numbers averaged 5K TPS during a test. * * * **Ledger, fee model & economic policy** --------------------------------------- Sui uses a novel asset-centric ledger instead of an account or UTXO-based ledger. The global state stores programmable assets or Objects that can be managed through Move packages. The motivation was to allow for parallelization, ownership rights, composability, and data-rich apps. The object’s metadata includes a globally unique ID, an explicit owner field, and a version/nonce. This provides several features: * **Higher Throughput**: Sui stores the Object and ID pair in a standard database and doesn’t require Merkelization. Merkle trees are used in Ethereum for easy verification and state syncing, but the disk read/write I/O operations have become the biggest scaling bottleneck. Sui allows local transaction-level verification without running the global state and easy sync for nodes through checkpoints. * **Object Ownership:** Sui has four object types based on the object's owner: address, object, shared (no single owner), and immutable.A user-owned object like tokens or NFTs can only be spent using a user signature. These objects can be transferred using the consensus-free fast path with <500 ms latency.Objects can own objects, creating hierarchies, a major unlock for gaming, socialfi, and identity usecases. For example, a character object can collect onchain weapon objects. They could be dynamic and evolving objects, with their history recorded through object versions/nonces.Shared objects have no specific owner, and anyone can read/write to them, making them useful for AMM DEXs or NFT minting contracts. Meanwhile, immutable objects have no owner and can contain immutable code or read-only data. Everything as an object simplifies ledger rules and app logic. **Fee model & Economic policy** Sui has a fixed supply of 10B and annual emissions of 2-3% of the total supply. Validators also earn from user fees, which have four components: compute, storage, storage rebates, and priority. Fees = compute unit x compute base price + storage units (created - destroyed) x storage base price + priority fees After each 24-hour epoch, Sui validators propose a base fee, and the final number should be acceptable to 2/3rds of the stake. This keeps the price stable for 24 hours, and the adjustment stabilizes dollar-denominated prices in the long term. The priority fee allows users to prioritize transactions during congestion or state contention. The storage fee is a deposit that can be rebated upon the object's destruction. A portion of the fees is directed to the storage fund, which helps pay future validators for the large history. This model is created to enable rich apps with large onchain data. * * * Execution: ---------- Sui uses Sui Move, a purpose-built programming language focused on simplifying the development of secure applications. Originally developed for Diem, Move is a Rust-based object-oriented programming language with built-in features and limitations that make programming safer and more efficient. The key features are: * **Object-centric model:** The object-centric model allows parallel execution for independent objects. It also opens a path to intra-validator sharding, where different machines can run different types of objects. Move’s object model provides a secure environment that reduces the errors a developer can make while making it easy to use. Users have lost billions of dollars in hacks, and Move can help reduce them in the following ways:**Approval permit attack:** Objects can only be modified if permitted by the user in a transaction and can’t be hacked due to long-standing permits.**Re-entrancy:** In Ethereum, an actor can enter malicious code while executing a contract and take control of it. Sui prohibits another contract from taking control and allows resources to be used only once, prohibiting re-entrancy.**Account checks:** Solana requires developers to write different checks for access controls and data types. Move Objects have these checks by construction, and every contract on Sui is bytecode-verified not to have such attack vectors. This could have prevented the [$320M Wormhole hack](https://www.certik.com/resources/blog/1kDYgyBcisoD2EqiBpHE5l-wormhole-bridge-exploit-incident-analysis). These are examples of how Move’s type, memory, and resource safety can help improve Web3 security. Move contracts can also be formally proven through the Move Prover. However, a new language and VM can always be vulnerable, and upon widespread adoption, we should better understand its security measures. * **Programmable Transaction Blocks:** A sui transaction can consist of up to 1024 move commands. This allows you to batch different actions and transactions into one atomic block. You can use it to batch payments, batch NFT mints, route trade through multiple DEXs, or chain operations like deposit, borrow, stake, and deposit to another protocol. Batching saves fees, increases actions per second, and improves user experience. According to [Mysten Lab benchmarks](https://blog.sui.io/sui-performance-update/), if all transactions were a PTB of size = 100 vs 1, the chain would do 30x more transactions.Sui allows for one-click apps, like Kriya’s one-click vault strategies, where you can loop or earn points from multiple protocols without any smart contract. Ethereum’s account abstraction roadmap is focused on enabling this for users, but it still has a lot of complexity and concerns. * **Dynamic Fields—Composability:** This allows adding objects to your objects, which creates hierarchies. It is useful for creating tables, arrays, trees, composable NFTs, and gaming/socialfi/identity elements. Sui parallelizes execution by object but doesn’t use optimistic parallelized transaction execution like Aptos. This isn’t currently prioritized but can be implemented in the future. The implicit benefit is that Sui app teams are forced to build parallelizable workflows. * * * Tokenomics ---------- SUI launched in May’23 with a capped supply of 10 billion tokens, with approximately 50% locked until 2030. Its primary utility includes fees, staking, and governance. Here is how it’s distributed: ![](https://storage.googleapis.com/papyrus_images/4fab2cb0e64b16c79e0cb914e8fad37ada08a96665783f83ad9485b6a61e89c8.png) • **Investors**: Mysten Labs raised $300M across two rounds in exchange for 14% of the supply, with a 1-year cliff and a 7-year vesting period. Another 8.9% of the supply, initially marked for FTX and Alameda, was bought back by Mysten Labs. **• Community Reserves & Stake Subsidies:** 10% of the supply is reserved for staking emissions, and the foundation can use another 10% for community programs, educational campaigns, or grants **• Community Access Program:** The team conducted a public IEO event to raise $50M in Apr ’23. Loyal community members were allocated ~1.5% of the supply at $0.03, while another 4.5% was publicly sold at $0.1. Including the 52% unallocated reserve, the Sui Foundation and Mysten Labs control approximately 85% of the tokens. While centralized third parties have custody of the locked tokens, there are concerns over the team’s influence on future allocation. * * * Sui Ecosystem ------------- In Oct'23, Sui announced a 157M SUI [incentive program](https://x.com/SuiNetwork/status/1712906176730746981) to bootstrap the DeFi ecosystem. This initiative and points and liquidity mining programs from new DeFi projects led to a 2,500% increase in TVL, from $30 million in Oct'23 to $750M in May'24. Sui now ranks among the top 15 chains by TVL and has nearly twice the TVL of Aptos. Additionally, Sui has over $400 million in stablecoins bridged, and it's consistently one of the top ecosystems in terms of bridge volume. ![](https://storage.googleapis.com/papyrus_images/13823bd8eac3f2678688420031a938472f64c3a444b7c412e415cf7adc4af254.png) Daily active users have remained fairly consistent at 15-20K, except during incentivized events like [Quest3](https://mystenlabs.com/blog/quest-3) or [Spam](https://x.com/juzybits/status/1785235568089067642). However, the monthly volume increased by 5,400%, from $55M in Sep'23 to $3B in Apr'24. This surge was driven by Cetus' concentrated liquidity DEX and Deepbook's CLOB, which feature active liquidity management and $SUI/$Deep incentives that can inflate volume. The Sui NFT ecosystem is still in its early stages, generating approximately $250K in monthly volume, with Trade Port being the primary marketplace. This volume is expected to increase as gaming projects launch on Sui. ### Lending/Borrowing Lending markets are the leading DeFi category on Sui, with around $325M spread across Navi, Scallop, and Suilend. Due to Sui's incentives and liquidity mining programs, these lending projects offer an attractive yield of approximately 20% on stablecoins, significantly higher than Ethereum's <10% yield and Solana's 10-12% yield. The SUI APY is 15-20%, much higher than the 3-4% staking yield, and could be even higher when considering points programs and airdrops. With a TVL of $135 million, Scallop focuses on advanced features and institutional-grade DeFi. The platform combines Compound v3 and Solend V2 features, including a three-tier interest rate model, soft liquidations, multi-oracle integration, flash loans, and mixed and isolated pools. Notably, Scallop was the first project to receive an official grant from the Sui Foundation and has won multiple grants since. With $130 million in lending TVL, Navi aims to be Sui's one-stop liquidity shop. Its core product is a lending market modeled after Aave v3 and offers liquid staking services. Finally, Suillend, which achieved $55 million in TVL just two months after its launch, is notable for being one of the first major DeFi projects to migrate from Solana's Solend to Sui. The move was motivated by Move and the improved developer experience. ### Spot & Perp DEXs Sui hosts over eight decentralized exchanges (DEXs), with Cetus, Deepbook, Aftermath, Turbo, and Kriya capturing the majority of volume and TVL. Deepbook, a CLOB DEX, was developed as a public good by the Sui Foundation and its partners. The core idea is similar to Serum, aiming to build a liquidity hub for all DEXs. Deepbook is expected to airdrop $DEEP to its users by the end of Q2. Cetus leads the DEX landscape on Sui with $140M in TVL and $1.5B in monthly volume. It's a concentrated liquidity DEX that offers vaults to automate liquidity management. Aftermath is the premier DEX aggregator, utilizing programmable transaction blocks (PTBs) to split transactions into multiple chunks and route them through various pools in a single atomic transaction. Additionally, Aftermath provides a liquid staking service. Finally, Kriya offers a spot and perpetual trading platform focusing on innovation and user-centric applications. It features one-click strategy vaults for leveraged lending and market-making. Kriya's 'degen corner' includes tools such as a token verifier, a no-code token creator, an airdrop tool, and a Pumpdotfun-style fair launch mechanism. Perpetuals and derivatives can be Sui’s breakthrough app. Onchain perpetuals have historically struggled due to high latency and low throughput, which dissuades market makers and results in high spreads/funding rates. We have seen perp projects like Synthetix struggle on Ethereum mainnet and find more adoption on L2s; they went from having [13 daily active](https://x.com/0xtuba/status/1461965502331252738) users to thousands. With Sui’s fast lane, Oracle updates and onchain orders are finalized in <500ms with negligible cost. Mysticeti is expected to enable < 250 ms latency. Perp DEXs can tap into the Deepbook liquidity pool, which reduces the barrier to integrating market makers. Bluefin, the leading perp platform on Sui, generates $150M in daily volume with 30ms optimistic confirmations and 500ms finality. It offers cross-chain assets like ETH, SOL, and SEI, which can be used for cross-margin trades. This offers traders a near-CEX-like experience. * * * Sui Narratives -------------- * **Ease of onboarding:** Sui's zkLogin allows people to create a wallet from email or other Web2 services without requiring a passphrase. Zk proofs ensure that the address isn't revealed to the authenticator, who signs locally and doesn't share any data. Another feature is sponsored transactions, which allow Sui dapps to pay for user transactions. This can allow Sui to onboard people with great ease. * **Payments:** The Mysten team at Novi Financial, part of Meta, focused on making payments more accessible. Sui's fast lane enables quick and inexpensive transactions. Mysten also developed zksend, which uses zkLogin to create payment links, allowing users to transfer and claim crypto payments through a Web2 login. * **Gaming**: Sui has secured major partnerships with RedBull, esports team Team Liquid, and Playtron for Sui gaming consoles and various gaming studios. In March 2024, they hosted a gaming summit and distributed 5 million SUI to gamers on the platform. Sui's object-centric model makes it ideal for creating rich, composable games. Key Challenges -------------- * **Developer Adoption:** New languages and VMs, except for Solana with Rust, have lacked adoption as EVM languages still account for 96% of DeFi TVL. Move is a new language, and Objects is a new ledger model that requires developers to build applications using logic that best utilizes its object model and parallelization capabilities. Sui Move has significant differences from Aptos Move; for instance, Sui requires a transaction to declare the state it touches, but Aptos doesn't. This distinction creates additional challenges for Move adoption. * **Centralization:** Ethereum and Solana have independent researchers and engineering teams actively building on the core client. However, Sui largely relies on Mysten Labs for its functioning and development, which can become a central point of failure under regulatory pressure. Mysten Labs and Sui Foundation have invested in and provided grants for several DeFi and gaming projects on Sui, which can hinder the ecosystem's organic growth. * **High Inflation & Token Unlocks:**  SUI had a very limited public sale, with the vast majority of tokens held by the team and private investors. Sui's supply started at 600 million in May 2023, increased to 2.3 billion today, and is expected to reach 3.3 billion by May 2024. This shift from early adopters to the early majority will result in high volatility. For example, ETH dropped 94% during the 2017 cycle, SOL dropped 97%, and AVAX dropped 95% during the 2020 cycle. * **Lack of Crypto-native GTM:** Ethereum's dominance until 2023 has demonstrated significant user retention. Few chains have attracted users despite substantial incentives and vastly improved user experiences. While SUI has outperformed Aptos, it must develop a crypto-native marketing strategy to drive adoption. For example, Solana's "Only Possibile On Solana" campaign or Monad's community-building efforts have captured people’s attention. * **Competition** - With the new wave of Layer 1s and Layer 2s solutions, block space is abundant, cheap, and fast. To stand out, Sui needs a clear value proposition for users and applications that are "Only Possible on Sui." * * * Conclusion ---------- Blockchains are the only tech that forces users to have a technical understanding and know-how to use the product. Solana has shown us that simplifying the user experience can enable us to use 10-100x crypto. Secure smart contracts, easy onboarding, gaming, and socialfi are some of the clearest paths for Sui to drive the next 10x in crypto usage. Sui’s incentive program has pushed DeFi to new heights. However, Sui onchain daily active users have averaged 15-20K for the past six months. The ecosystem needs to find a way to onboard new users and retain them after incentives. * * * Sorry for the long post; here are some memes ![SUI narrative](https://storage.googleapis.com/papyrus_images/23e171e3af7c86579a1678630aeb0c48def3445c3893b615983f6a66ae833669.png) SUI narrative ![](https://storage.googleapis.com/papyrus_images/b2777e077577275d8ee92145335af851dcb6244f0636ef7057fdc6ee6f06e696.png) ![](https://storage.googleapis.com/papyrus_images/8cda03382742ac9720a1ceb39fd6dda276e5c877b56e57e020e92c3f2174e5ba.png) --- *Originally published on [notjuve](https://paragraph.com/@notjuve/sui-web3-at-ease-of-web2)*