Sui - Deep Dive
The full picture on Sui - object model, Move language, Mysticeti consensus, zkLogin, Walrus storage, and what the ex-Meta team is building.
Sui
Sui is what happens when a group of senior systems engineers - most of them ex-Meta working on the abandoned Diem blockchain project - decide to start over from first principles rather than build on existing infrastructure. The result is a blockchain with a genuinely different data model, a new programming language, sub-second finality in most cases, and a set of UX primitives (zkLogin, sponsored transactions) that solve real onboarding problems. It also comes with real trade-offs: a non-EVM smart contract language that fragments the developer pool, an early-stage ecosystem that is only starting to build real liquidity and users, and the institutional baggage that comes from being a well-funded project with high expectations.
Sui mainnet launched in May 2023. It is early but not unproven. The technical foundation is solid. The ecosystem is the open question.
What Sui Actually Is
Meta's Diem project (originally Libra) was a stablecoin and payments infrastructure initiative announced in 2019. It attracted enormous regulatory opposition and was eventually wound down in 2022. Before shutdown, the Diem team had built significant infrastructure - including a programming language called Move and a consensus system called HotStuff. When Diem folded, members of that team started two separate projects: Aptos and Sui. Both use Move. Neither uses the EVM.
The founders of Sui - Evan Cheng, Adeniyi Abiodun, Sam Blackshear, George Danezis, and Kostas Chalkias - made a deliberate choice not to pursue EVM compatibility. The argument: EVM compatibility is a crutch. It brings existing Ethereum developers and tools, but it also imports Ethereum's design limitations. If you're going to build a new blockchain, build it for what you want it to do, not for what already exists.
That choice is defensible on technical grounds. It is expensive on ecosystem grounds. The EVM-compatible chains (Avalanche C-Chain, Polygon, BNB Chain, Base, Arbitrum, Optimism) inherited Ethereum's entire developer and application ecosystem essentially for free. Non-EVM chains have to earn every developer from scratch.
Sui's bet is that the performance and safety advantages of its design are large enough to justify that cost - and that the next generation of consumer crypto applications (gaming, social, payments) will be built by developers who don't have EVM loyalty yet.
The Object Model
Every blockchain needs a model for how state is organized. Ethereum uses an account model: state is a mapping from addresses to account balances and contract storage slots. Everything touches a global state tree. This is conceptually simple and well-understood, but it creates a fundamental bottleneck: transactions that touch the same state (a DEX contract, an NFT collection) cannot be safely executed in parallel, because you can't know the outcome until they're sequenced.
Sui uses an object model. Everything on Sui - tokens, NFTs, smart contract state, even code - is an object with a unique identifier, a type, and an explicit owner. The owner of an object controls it. This ownership structure is the key to Sui's parallel execution.
Owned objects are controlled by a single address (or another object). Transactions that only touch owned objects - sending a token, using an item in a game, updating a personal profile - can be validated directly by the validators holding custody of the relevant objects, without any global consensus round. These transactions achieve very low latency because they bypass the sequencer.
Shared objects can be read and written by multiple addresses. Transactions involving shared objects require full consensus ordering to prevent conflicts. A liquidity pool is a shared object. An auction contract is a shared object. These transactions are more expensive and slower than owned-object transactions, though still fast by blockchain standards.
The practical consequence is that Sui can execute owned-object transactions in parallel at the hardware level - validators process non-conflicting transactions simultaneously rather than one at a time. This is a fundamentally different execution model from Ethereum's sequential processing, and it's the reason Sui's raw throughput potential is significantly higher.
The cost is complexity. Designing Sui applications requires thinking about which objects are owned and which are shared, which affects both security and performance. Developers used to Ethereum's simpler model have to internalize a new mental framework.
Move Language
Move was created at Meta specifically for the Diem project, designed by Sam Blackshear. The core insight driving its design: smart contract bugs on Ethereum are expensive and common precisely because Solidity makes it easy to write code that loses or duplicates digital assets.
The primary innovation is linear types applied to resources. In Move, a resource type (representing a token, an NFT, any asset with value) cannot be copied or silently dropped. It must be explicitly moved, stored, or destroyed. The compiler enforces this. You cannot write a function that accidentally creates a second copy of a token, or one where a token just disappears without an explicit burn. The entire class of reentrancy bugs that drained hundreds of millions from Ethereum contracts is structurally impossible in Move because of how function calls and resource ownership are managed.
Move also has a strong module system that enforces clear boundaries between what a module can do to another module's types. External code cannot arbitrarily manipulate the internals of your resource types.
The honest trade-off table:
| Property | Move (Sui) | Solidity (Ethereum/EVM) |
|---|---|---|
| Asset safety | High - linear types prevent duplication/loss | Lower - requires careful manual management |
| Developer pool | Small but growing | Very large, mature |
| Tooling maturity | Improving rapidly | Mature ecosystem |
| Audit resources | Limited | Extensive |
| Hiring market | Constrained | Deep |
| Bug class prevention | Stronger | Weaker |
The fragmentation problem is real. Every Solidity developer who wants to build on Sui needs to learn a new language. Every Ethereum smart contract audit firm needs to train Move specialists. Every developer tool - IDEs, fuzzers, formal verification tools - needs a Move implementation. These are solvable problems, and they're being solved, but the EVM ecosystem has a decade head start.
Sui's version of Move also diverged from Aptos's version, adding further fragmentation. The two Move ecosystems are related but not directly compatible - a Sui Move contract cannot simply be deployed on Aptos and vice versa. For a language that was already fighting for mindshare, this split was a cost.
Mysticeti Consensus
Sui's original consensus protocol was Bullshark, itself a descendant of the DAG-based consensus work done at Meta. In 2024, Sui deployed Mysticeti, an updated consensus protocol designed specifically to push latency as low as possible.
The key structural feature is a DAG (directed acyclic graph) rather than a traditional linear chain of blocks. Validators broadcast their proposed blocks to each other in parallel, and the DAG structure allows multiple blocks to be proposed and included simultaneously - without waiting for each to be confirmed before the next is proposed. Mysticeti uses this structure to achieve a two-round commit path for shared-object transactions, significantly reducing the round-trip count compared to linear Byzantine fault-tolerant consensus designs.
The result: sub-second finality in normal network conditions for most transactions. Owned-object transactions, which bypass global consensus entirely, complete in milliseconds. Shared-object transactions run through Mysticeti and typically finalize in well under a second.
Why does this matter for applications? Real-time applications - games, social apps, trading interfaces - are fundamentally different to build when state is final in 300ms vs 2-12 seconds. A card game where every move is an on-chain transaction becomes playable when finality is sub-second. It was borderline unplayable on Ethereum L1 and still awkward on most L2s with their withdrawal and state-sync delays.
Finality speed is one of the few places where Sui has a demonstrable, measurable advantage over most of the market - not theoretical, not in test conditions, but in live production.
zkLogin
Getting users onto a blockchain has always required either teaching them to manage seed phrases and private keys, or trusting a custodian to manage those things for them. Neither is good for mass adoption. zkLogin is Sui's solution.
zkLogin allows a user to create a Sui address using an OAuth provider: Google, Apple, Facebook, Twitch, or any compatible OpenID Connect service. Here's how it works: the user authenticates with, say, Google. Google issues a signed JWT (a standard web authentication token). The user's device uses a zero-knowledge proof to demonstrate that it possesses a valid JWT from Google - without revealing which Google account it belongs to, and without Google learning anything about the resulting Sui address. The on-chain address is derived from a combination of the OAuth provider, the user's unique identifier within that provider, and a user-controlled salt that keeps the mapping private.
The privacy properties matter. Google does not know what Sui addresses you use. An observer on-chain cannot link your address to your Google account without knowing your salt. The proof verifies the claim without revealing the underlying identity.
From the user's experience: sign in with Google, have a Sui wallet. No seed phrase. No browser extension. No understanding of cryptography required. If a user loses access to their Google account, they lose access to the wallet - which trades decentralization for usability, but is the right trade-off for users who were never going to manage seed phrases anyway.
zkLogin is a meaningful technical achievement. Zero-knowledge proofs for OAuth tokens were theoretically possible before Sui but had not been deployed in production at this scale. It also represents a philosophically honest approach - rather than pretending users will learn seed phrase management, design for how users actually behave.
Walrus
Walrus is Sui's decentralized storage network, launched in 2024-2025. It is not a blockchain - it is a storage layer designed to complement Sui's on-chain infrastructure.
The core design uses erasure coding, the same mathematical technique used in RAID storage and data center replication. A file is encoded into N chunks such that any K of them are sufficient to reconstruct the original (where K < N). These chunks are distributed across Walrus storage nodes. No single node holds enough data to reconstruct the file, and the file survives even if a significant fraction of nodes go offline or act maliciously.
Compared to Arweave's approach - which relies on miners economically incentivized to store permanent copies of full files - Walrus takes a different philosophy. Arweave prioritizes permanence through redundant full copies and economic guarantees baked into the protocol. Walrus prioritizes throughput and cost efficiency through erasure coding. The two approaches have different trade-off profiles: Arweave provides stronger permanent storage guarantees for archival use cases; Walrus is better suited to high-volume, dynamic storage needs like gaming assets, social media content, and application data.
Walrus integrates tightly with Sui - storage deals are managed through Sui smart contracts, and SUI token is used for payments. This creates a flywheel where Walrus usage drives Sui transaction volume. For applications built on Sui that need to store large assets (game content, images, video), Walrus provides a native solution without requiring integration with an external storage protocol.
Ecosystem - Gaming and Consumer Apps
Sui has made a deliberate choice to target gaming and consumer applications as its primary growth vector, rather than competing directly with Ethereum and Solana for DeFi TVL. The reasoning is sound: gaming requires low latency (Mysticeti), fine-grained asset ownership (object model), simple onboarding (zkLogin), and cheap storage (Walrus). Sui's architecture matches the requirements better than most alternatives.
The gaming ecosystem is early but has genuine momentum. Projects like Panzerdogs, Bullshark Quests (Sui's own gamified engagement system), and several others in development are building on Sui's infrastructure. The Sui Foundation has aggressively funded gaming grants. Notable: the "Capy" NFT brand, built as a showcase of Sui's NFT capabilities.
DeFi on Sui is functional but modest by market-scale comparison. Cetus Protocol and Turbos Finance handle DEX activity. Scallop provides lending. Total TVL reached meaningful levels by late 2024 but remains well below the top DeFi ecosystems. The object model's distinction between owned and shared objects makes some DeFi patterns less natural to implement than on Ethereum.
The honest assessment: Sui has genuine traction in gaming and consumer segments in a way that most new L1s do not. It is not yet competitive with Ethereum or Solana in DeFi depth or developer count. The question is whether gaming becomes crypto's primary growth vector over the next several years - and whether Sui retains its gaming ecosystem leadership as other chains make their own gaming pushes.
Tokenomics
SUI has a total supply of 10 billion tokens. The initial distribution allocated significant portions to early investors, the Mysten Labs team, and the Sui Foundation - a distribution that attracted criticism for being heavily weighted toward insiders. Vesting schedules for team and investor tokens run through 2027, meaning ongoing sell pressure from unlocks is a consideration.
Staking on Sui uses a delegated proof-of-stake model. Token holders delegate to validators and earn staking rewards - current APY has varied but sits in the single digits. The epoch length is 24 hours, with delegation changes taking effect at the next epoch boundary.
Gas on Sui is priced in MIST (the smallest unit of SUI) and adjusts based on network load through a reference gas price mechanism. Validators vote on a reference gas price each epoch, and the protocol targets a price that balances validator compensation with user affordability. Unlike Ethereum's fee burn mechanism (EIP-1559), Sui's fee model routes most fees to validators, with a portion going to a storage fund that subsidizes future data access costs.
The storage fund is a notable design feature: when users write data to chain, a portion of the gas fee goes into a fund that earns returns and is used to pay future validators for storing that data indefinitely. This separates the cost of data creation from the ongoing cost of data storage, solving a problem Ethereum handles poorly.
Value capture is tied directly to network usage: more transactions mean more staking demand, more storage fund growth, and more fee revenue to validators. The tokenomics are straightforward but depend entirely on the ecosystem achieving significant usage.
Investment Considerations
Sui is one of the technically strongest new Layer 1s to launch in the current cycle. The object model is a real innovation, not marketing. Mysticeti finality is measurably fast. zkLogin works in production. The team's depth - systems engineers who built production infrastructure at Meta scale - is apparent in the architecture.
The genuine case: Consumer crypto adoption requires infrastructure that doesn't ask users to manage seed phrases and doesn't make them wait seconds for confirmation. Sui has built that infrastructure. The gaming and consumer focus is strategically coherent rather than opportunistic. If those markets grow, Sui is positioned well.
The genuine concerns: The Move language creates a smaller developer pool, and that pool is split further by the Sui/Aptos divergence. Ecosystem depth - DeFi TVL, stablecoin issuance, real volume - is still modest. High investor/team token allocations mean ongoing vesting pressure through 2027. Being technically superior has never been sufficient to win a platform war; developer ecosystem and liquidity depth tend to win, and Sui is still climbing both curves.
The ex-Diem baggage: The Meta/Diem origin is a double-edged characteristic. It provides engineering credibility and the resources that come with a well-funded corporate predecessor. It also means the team is associated with a regulatory failure and a project that was ultimately killed by political pressure rather than technical flaws. That context cuts both ways depending on your read of regulatory trajectories.
Sui is a project to watch with appropriate position sizing for early-stage, technically-strong ecosystems. The thesis is coherent, the execution so far is solid, and the market it is targeting - consumer apps, gaming, any application where UX matters more than DeFi composability - is both large and underserved by current infrastructure. The risk is that ecosystem building takes longer than expected, and that the Move fragmentation problem limits developer growth more than the upside case assumes.
Learn More
- Token Evaluation Framework → - How to evaluate any token
- Blockchain Ecosystems → - Compare chains side by side
- sui.io - Official Sui site
- docs.sui.io - Technical documentation and Move guides
- suiexplorer.com - Block explorer
Content current as of March 2026. Ecosystem metrics, TVL, and staking rates change frequently - verify current data at DeFiLlama and the official Sui documentation.