區塊鏈經常被比喻為數碼孤島——安全而封閉，但彼此孤立無援。過去十年，不同項目試圖連結這些孤島，使資產與數據得以跨網絡流通。不過，早期的「連接區塊鏈」方法，大多依賴臨時橋樑或聯盟式系統，反而帶來新風險。如今，有一種嶄新模式正崛起——模組化區塊鏈架構。這種模式下，不同區塊鏈功能（如執行、共識、數據儲存及安全）可由互相配合的獨立網絡提供。三個前瞻性項目正領導這場變革：Celestia、Avail 及 EigenLayer。它們各自處理不同拼圖，令區塊鏈變得更易連結、可擴展及多元。本文將深入探討這些項目的運作原理、解決的問題，及其如何共同重塑整個區塊鏈生態。
從單體鏈到模組化未來
傳統「單體式」區塊鏈（如比特幣及早期以太坊）中，每個節點需要負責全部功能：交易執行、共識、數據可獲得性及最終確認。這種「一刀切」的設計雖簡單且安全，但存在天然擴展限制。所有操作都發生於一層，當用量提升時，鏈容易成為瓶頸。相比之下，模組化區塊鏈將職能拆分為不同層或模組。例如，一層專責交易執行（處理智能合約），另一層只負責交易排序及驗證區塊數據是否公開予全網查閱。這種職責拆細的方法，允許更高靈活度及處理量，同時不犧牲安全。
單體式與模組化區塊鏈架構對比。在單體設計（左），一條鏈同時負責執行、結算、共識及數據可用性；在模組化架構（右），這些功能細分於專門的層——例如 rollup 負責執行，其他網絡則提供共識及數據可用性。這種分工可讓多條鏈共享安全或數據層，從而提升擴展性與互通性。
模組化興起，源於多年可擴展性挑戰的經驗教訓。早期僅靠增加單體鏈容量，發現有中心化風險（比特幣及以太坊區塊大小爭議即一例）。Layer-2 解決方案如 rollup 橫空出世，將執行移離主鏈，同時繼續用主鏈確保安全及儲存數據。像 Arbitrum 及 Optimism 這類 rollup 已顯著提升吞吐量，卻暴露了另一問題：數據可用性困難。在 rollup 上，交易數據必須可靠地發佈於某處（以便任何人能重建狀態或挑戰欺詐行為）。如果要在以太坊這類繁忙的 Layer-1 上發佈，成本高昂且容量有限，進一步限制 rollup 的發揮。這正是 Celestia 及 Avail 以專屬數據可用性網絡針對的痛點。
同時，許多新區塊鏈或服務面臨自建安全網絡的啟動難題：獨立新鏈的安全取決於其驗證人或礦工隊伍的實力。Polkadot 及 Cosmos 提供所謂「共用安全」或互通框架，卻有不同權衡——Polkadot parachains 共享中心中繼鏈的安全，而 Cosmos 鏈則可透過 IBC 協議互連，但各自維護驗證人隊伍。EigenLayer 則提出全新以太坊共用安全方案：讓現有以太坊持幣人「再質押」資產，支援其它鏈或模組。換句話說，就是循環利用以太坊龐大驗證人基礎，助力新加密基建起步。
這些創新（專屬數據可用性層及再質押共用安全）共同構建未來多鏈、模組化加密生態骨幹。區塊鏈可插入共享數據層、共用安全資源，更易互通。細看每個項目前，先認識模組化架構內「層」的意義：
- 執行層：交易在此執行並更新狀態（如 rollup 處理智能合約）。
- 共識層：由驗證人排序及最終確認區塊（鏈的「心跳」）。
- 數據可用性層：確保每區塊所有交易數據都有公開發佈，網絡能驗證內容。
- 結算層（選用）：處理糾紛或驗證證明之層，rollup 可在此提交欺詐或有效性證明作結果裁定。
在單體鏈中，上述功能全融合於單一平台。在模組架構下，各網絡分擔不同角色。例如，Celestia 提供共識及數據可用性，多條獨立執行層（rollup 或專用鏈）則建立於其上。Avail 同樣主攻多條鏈的數據可用性。EigenLayer 則聚焦安全層——讓多項服務借用以太坊的共識作信任基石。這些可視為互補合作——Celestia 及 Avail 針對擴展型數據及共識，EigenLayer 則專注共用安全與互通。以下逐一細看其背景、運作方式及現時狀況。
Celestia：模組化數據可用性的先行者
Celestia 經常被譽為首個完全自零設計的模組化區塊鏈網絡。於 2023 年底主網 beta 上線，Celestia 的核心理念簡單而有力：只提供共識及數據可用性，故意不涉及其他功能。不會執行用戶交易或智能合約，甚至不驗證用戶鏈的狀態轉換。其職責僅為為各鏈（如 rollup）排序區塊並確保數據在網絡中能被所有人隨時下載與驗證。專注於這兩項任務，Celestia 致力成為讓不同區塊鏈可安心用作「離鏈執行」的底層基座。
技術上，Celestia 最突出之處為數據可用性取樣（DAS）。這是一種密碼學技術，使輕節點無需下載整個區塊也能驗證其數據已發佈。在 Celestia，每個區塊的數據會被糾刪編碼並分割為小片。輕節點隨機取樣部分片段測試可否檢索。若隨機樣本足夠多都能取回，該節點即高度相信整個區塊數據完整可用。這大大減輕個別節點負擔——即使智能手機或瀏覽器裝置亦可協助檢查數據可用性，同時能夠偵測惡意驗證人隱瞞數據嘗試。形象點說，就像數千名獨立審計員各自隨機抽查一本巨著的不同頁數——即使只漏掉一頁，大概率都會被發現。只要部分誠實輕節點積極取樣，隱瞞數據幾乎無法不被揭發。
Celestia 採用糾刪編碼（erasure coding）及 KZG 多項式承諾技術，令此新型取樣機制成為現實。糾刪編碼為數據添加冗餘備份，就算部分缺失也能「補全復原」。KZG 承諾（以太坊後來 proto-danksharding 升級亦應用）能生成簡潔證明，令輕客戶端快速驗證樣本對應正確的原始多項式。這些技術，意味 Celestia 可以安全支援遠大於傳統區塊鏈的區塊尺寸。2023 年 10 月主網「Lemon Mint beta」上線時，區塊已達 2–8 MB，遠超以太坊大約 0.1 MB 的區塊。而根據 Celestia 路線圖，未來甚至可擴展至 1000 MB（1 GB）級別，有望支持每秒數萬筆總交易量。早期測試顯示效能驚人：2025 年 4 月，Celestia 測試網使用 128 MB 區塊達到約 21 MB/s 吞吐率。這設計另有一種反直觀現象——輕節點（非驗證人）越多，區塊容量反而可增長且保持安全，等於「越去中心化網絡容量越大」。
為何要如此重視數據可用性？核心好處是 Celestia 令新區塊鏈或 rollup 開發人員毋須自建驗證人網絡或擔心數據可用性上限。只需設計任何執行邏輯（無論 EVM、遊戲鏈還是注重私隱鏈），便能把交易數據發佈到 Celestia；Celestia 驗證人會為這些交易排序並確保數據可取得，由 rollup 自身節點或用戶負責執行及驗證狀態轉移。這賦予開發者極大自主——可自由選擇執行環境（EVM、WASM、Cosmos SDK 等），甚至設立自有代幣和治理，卻同時享有 Celestia 強大數據可用性和共識。創辦人 Mustafa Al-Bassam 形容這種模式為從單體區塊鏈的「撥號年代」進入模組化的「寬頻時代」—— block space is abundant and cheap. Instead of competing for limited throughput on a single chain, many specialized chains can run in parallel, all anchored by Celestia. Developers no longer have to shoehorn their app into someone else’s chain or clone a Layer-1 and recruit validators from scratch – they can start a sovereign chain that plugs into Celestia’s consensus and data layer.
區塊空間而家又多又平。開發者唔再需要爭一條主鏈有限嘅吞吐量，而係可以開發專門用途嘅鏈，全部並行運行，而且都以 Celestia 作為錨點。開發者唔需要再逼自己嘅應用落入其他人嘅鏈，或者自己重新複製一條 Layer-1 再逐個搵驗證者 – 佢哋可以直接起一條自主鏈，接駁 Celestia 嘅共識同數據層。
Notably, Celestia does not enforce any execution rules on the chains using it. This means it doesn’t check fraud proofs or validity proofs for rollups – those are handled either by the rollup’s users or by a separate settlement layer if one exists. This approach is called “sovereign rollups”, where the rollup is sovereign over its own state (no higher authority automatically slashes bad behavior). If a sovereign rollup using Celestia produces an invalid state transition, Celestia’s validators will still include and publish the data (as long as it’s properly formatted), because Celestia’s job is not to know what’s valid or invalid in that rollup – that’s up to the rollup community or an optional settlement layer. This design maximizes Celestia’s neutrality and simplicity, but it means rollup projects have a choice: they can be fully sovereign (with social consensus to handle any faults), or they can introduce their own fraud proofs or proofs and maybe post those to some security layer (even potentially Ethereum or another chain). In practice, some teams plan to use Celestia for data while using Ethereum as a settlement layer – achieving a hybrid model where Ethereum verifies proofs for security, and Celestia provides cheaper data availability. The Cosmos-based rollup Eclipse is one example that considered using Celestia for data and Solana’s VM for execution, while settling to Ethereum – illustrating the creative combinations modular architecture enables. Celestia itself is built with Cosmos SDK and uses the Tendermint (now CometBFT) consensus algorithm (PoS based). It currently has 100+ validators and implements features like namespaced Merkle trees to allow efficient retrieval of specific rollup data from blocks. The network achieved near-instant finality (~6-second block times with fast finality) with a 2023 upgrade, making it responsive enough for practical use.
值得留意，Celestia 本身唔會對使用佢嘅鏈執行任何規則。代表佢唔會查核 rollup 嘅詐騙證明或者有效性證明 – 呢啲嘢要靠 rollup 本身嘅用戶，或者如果有獨立結算層，由嗰個層處理。呢種做法叫「sovereign rollups」（自主 rollups），即係每條 rollup 都對自己嘅狀態有自主權（冇更高權限自動懲罰惡意行為）。如果用 Celestia 嘅 sovereign rollup 出現無效狀態轉移，只要數據格式正確，Celestia 嘅驗證者都會照樣包入同發佈個數據，因為 Celestia 嘅職責唔係判斷某條 rollup 嘅數據有效定無效——要靠 rollup 社群自己或者某個結算層決定。咁樣設計令 Celestia 保持中立同簡單，不過對 rollup 項目嚟講有個選擇：可以全自主（透過社交共識去應對問題），或者自己寫詐騙證明、有效性證明，甚至放上第二層安全層（例如可能用 Ethereum 或其他鏈）。實際上，有啲團隊已經計劃用 Celestia 做數據層，同時用 Ethereum 做結算層——打造混合模式，Ethereum 負責驗證安全證明，Celestia 提供更平嘅數據可用性。來自 Cosmos 嘅 Eclipse rollup 就係一個例子，佢諗過用 Celestia 裝數據、Solana VM 執行、落到 Ethereum 做結算——完全展示咗模組化設計有幾咁多變。Celestia 自己係用 Cosmos SDK 起、用 Tendermint（依家叫 CometBFT）共識協議（基於 PoS）。而家有超過 100 個驗證者，仲有 namespaced Merkle trees 呢啲功能，方便喺區塊中有效抽取指定 rollup 數據。2023 年升級之後，Celestia 達到近乎即時最終性（大約六秒出一個快最終性區塊），實際應用都夠快。
The momentum behind Celestia has grown rapidly. The project (originally named LazyLedger in its research phase) secured significant funding to realize its vision. In March 2021, Celestia’s team raised a $1.5 million seed round to develop a “modular consensus and data availability layer”. A year later in October 2022, they raised $55 million in Series A funding led by Bain Capital Crypto and Polychain Capital. And by September 2024, the Celestia Foundation raised another $100 million (led again by Bain) to fuel development, bringing total funding to $155M. Such backing underscores the high hopes for modular blockchains. Celestia launched its first public testnet (Mamaki) in 2022, followed by developer testnets like Arabica and Mocha, before achieving its Mainnet Beta launch in October 2023. With the mainnet live (albeit labeled “beta”), attention turned to building out the ecosystem of rollups on Celestia. Over half a million users participated in Celestia’s early testnets and were rewarded in an airdrop of its native token $TIA in late 2023. This broad community test hinted at the demand for Celestia’s blockspace: indeed, Celestia’s token airdrop was called one of the “hottest” of 2023.
Celestia 嘅增長勢頭好快。項目本身研究時期叫 LazyLedger，搵到唔少融資實踐佢嘅願景。2021 年三月，Celestia 團隊種子輪集資 150 萬美金，目標係建一個「模組化共識及數據可用層」。2022 年十月完成 A 輪融資 5,500 萬美金，由 Bain Capital Crypto 同 Polychain Capital 領頭。2024 年九月 Celestia Foundation 再融資一億美金（又係 Bain 帶頭），總融資額升到 1.55 億美金。呢啲支持反映市場對模組化區塊鏈好有期望。Celestia 2022 年推出第一個公測測試網（Mamaki），之後有 Arabica、Mocha 開發者網，2023 年十月正式主網 Beta 上線。主網一出焦點就放喺開發 Celestia 嘅 rollup 生態圈。超過 50 萬用戶參加 Celestia 前期測試網，2023 年底仲收咗 $TIA 原生代幣空投。呢個大規模社群試驗已率先顯示大家有幾想用 Celestia 嘅區塊空間：Celestia 空投更被譽為 2023 年最「火熱」之一。
Several projects are already integrating with or building on Celestia. For example, Nexus – a network of bridges using Cosmos’s Inter-Blockchain Communication (IBC) protocol and Hyperlane – launched alongside Celestia’s mainnet to help bootstrap liquidity and connectivity between Celestia rollups and other ecosystems. Celestia’s upcoming upgrades aim to further enhance interoperability; the Lotus upgrade slated for mid-2025 will integrate Hyperlane interoperability, allowing Celestia rollups to easily communicate with Ethereum and other chains. In short, Celestia is positioning itself not just as a data layer, but as the hub of a new modular multi-chain world – one where new blockchains can sprout with minimal friction, inheriting security (via shared data availability and consensus) and easily interacting with each other through standard protocols.
有唔少項目已經同 Celestia 集成或者喺佢之上建設。例如 Nexus ——一個用 Cosmos IBC 協議嘅橋接網絡，同 Hyperlane——喺 Celestia 主網推出時同步上線，主要目標係幫 Celestia 上嘅 rollups 同唔同生態鏈打通流動性同互聯互通。Celestia 未來升級會再推高跨鏈能力；2025 年中 Lotus 升級將會集成 Hyperlane 互通性，到時 Celestia 上嘅 rollups 可以簡單同 Ethereum 或其他鏈互相溝通。簡單講，Celestia 唔只係個數據層，佢想做新一代模組化多鏈世界嘅樞紐——新鏈可以無縫誕生、無痛繼承安全性（共享數據可用性同共識）、再用標準協議互相對接。
Celestia’s approach is not without trade-offs. Because it doesn’t validate what’s in the data it carries, there is a risk that a chain using Celestia could go rogue (fail to publish fraud proofs, etc.) and Celestia would continue to publish its data regardless. The safety of end-users still depends on the chains built on top of Celestia doing their job correctly (much as Ethereum rollup users depend on the rollup operators and fraud proofs). However, by removing execution from the base layer, Celestia greatly simplifies the consensus engine and maximizes throughput. The promise of cheaper data availability is a big draw – Celestia is pitched as a cheaper alternative to storing data on layer-1 blockchains like Ethereum. This could alleviate the high fees and congestion that L2 rollups face when posting data to Ethereum today. It’s notable that Ethereum itself is evolving in a similar direction: with Proto-Danksharding (EIP-4844) introduced in 2023, Ethereum began adding dedicated “blob” data space to blocks, specifically to make rollup data cheaper. In the coming years Ethereum plans to implement full Danksharding with data availability sampling, effectively adopting many of the techniques that Celestia uses (though integrated into Ethereum’s beacon chain). This raises a big question: will Ethereum’s own upgrades negate the need for external data layers like Celestia, or will the multi-chain world still favor independent, specialized layers? Proponents of Celestia argue that a sovereign modular layer can innovate and scale faster, and serve many ecosystems (not just Ethereum’s rollups). Moreover, Ethereum’s focus is on scaling its rollups, whereas Celestia is neutral ground for any chain or rollup, whether Ethereum-based or not.
Celestia 嘅方法唔係冇取捨。由於佢本身唔驗證數據入面內容，理論上有機會有條鏈「做壞」——例如冇發 fraud proof——Celestia 都會照發佢個數據。最終安全始終靠建立於 Celestia 上面嘅鏈自己做好本份（就如 Ethereum rollup 用戶要信得過 rollup 運營同 fraud proof）。但係將執行拆走，Celestia 大幅簡化咗共識引擎，吞吐量又極大提升。數據層平咗係佢個主賣點——Celestia 俾你個廉價替代品，比 layer-1（如 Ethereum）儲存數據更加平。咁樣有助減輕今日 L2 rollup 上鏈 Ethereum 嗰啲高昂費用同擁塞。值得留意，Ethereum 本身都向住呢個方向進化：2023 年出咗 Proto-Danksharding（EIP-4844），開始出「blob」專用數據空間，就係要畀 rollup 數據平啲。Ethereum 接下來會推真正 Danksharding 同可用性採樣，實際上會落實好多 Celestia 現時用嘅技術（不過變做集成落 Ethereum beacon chain）。咁會唔會令 Celestia 類外部數據層變得冇必要？又或者多鏈世界依舊更加需要獨立、專門層？Celestia 支持者認為自主模組化層創新空間更大、擴展得更快，又可以服務唔同生態（唔淨係得 Ethereum rollup）。再者，Ethereum 只係專注自己 rollup，而 Celestia 係任何鏈、rollup 嘅中立地帶，唔理佢有冇用 Ethereum。
As of mid-2025, Celestia stands as a trailblazer for modular blockchains. It has proven the feasibility of data availability sampling at scale on a live network and gathered a community of users and developers eager to launch new chains. The race to connect blockchains has one strong contender in Celestia: a base layer that many chains can share. But it is not the only player. In parallel, another project emerged from the Ethereum community to tackle the data availability challenge – from a different angle.
到 2025 年中，Celestia 已經成為模組化區塊鏈嘅先驅。佢證明到大規模可用性採樣喺 live network 都走得通，而且有一班開發者同用戶社群等緊 launch 新鏈。連接區塊鏈之爭，Celestia 無疑係一個勁有力參賽者：一個好多條鏈可以共用嘅底層。但亦唔止佢一個。在同一時間，Ethereum 社區都出現另一個專攻 data availability 嘅新項目——不過路徑略有不同。
Avail: Data Layer for an Interconnected Multi-Chain Ecosystem
Avail: 為互聯多鏈生態提供嘅數據層
Around the same time Celestia was taking shape, Polygon (the team known for its Ethereum scaling solutions) was quietly working on a similar concept called Avail. First announced in mid-2021, Polygon Avail is a general-purpose, scalable data availability layer (DAL) for blockchains. The premise of Avail is much like Celestia’s: provide a reliable, decentralized ledger where other chains can dump their transaction data, thereby decoupling the data availability and ordering function from execution. As Polygon co-founder Anurag Arjun put it, “Avail decouples the data availability layer, making it easier for chain developers to focus on execution and settlement”. By late 2022 and 2023, Avail began to gain its own identity separate from Polygon. In fact, in March 2023 Polygon decided to spin off Avail into an independent project, with Arjun leaving Polygon Labs to lead Avail full-time. This spin-out underscored how important Avail’s mission had become: it wasn’t just an internal feature for Polygon, but a standalone attempt to unify and connect many blockchains via a shared data layer.
大約喺 Celestia 逐步成形嘅同時，Polygon（出名做 Ethereum 擴容方案嘅團隊）其實默默咁做緊同類型概念：叫 Avail。Polygon Avail 係 2021 年中正正式式宣佈，主打面向多條鏈嘅通用、高擴展性數據可用層（Data Availability Layer，DAL）。Avail 嘅基本理念同 Celestia 好似：提供一個可靠、去中心化台帳，畀其他鏈 dump 交易數據，即係將數據排序同可用度職能拆出嚟、唔再同執行層綁死。Polygon 聯合創辦人 Anurag Arjun 話，「Avail 將數據可用層解耦，令鏈開發者可以專注做執行同結算」。去到 2022、2023 年，Avail 已經有自己個品牌同定位。事實上 2023 年三月，Polygon 決定將 Avail spin off 成獨立項目，Arjun 離開 Polygon Labs 全心帶領 Avail，咁都證明 Avail 嘅重要性——佢嘗試唔只係 Polygon 嘅內部配件，而係一個用共用數據層統一、連接多條區塊鏈嘅獨立工程。
Technically, Avail’s design philosophy converges with Celestia’s on many points. Avail uses erasure coding and polynomial commitments (KZG) to implement data availability sampling and proofs. In a 2021 introductory post, Polygon’s team described Avail’s architecture in detail: Avail arranges each block’s data in a two-dimensional matrix, applies erasure coding to each column (doubling the data with redundant pieces), and then uses Kate (KZG) commitments to commit to each row. Light clients sample random cells of this matrix and use the KZG proofs to verify consistency. If any part of the data were missing, a light client would catch an inconsistency in its random sample with extremely high probability. This ensures that a block is only considered valid in Avail’s consensus if all its data is available – exactly the same principle Celestia follows. As the Polygon team noted, “Avail reduces the problem of block verification to data availability verification, which can be done efficiently with constant cost using data availability checks.” In other words, Avail validators don’t execute transactions; they just ensure every block is accompanied by enough data that anyone could later execute those transactions if needed.
技術上，Avail 嘅設計理念同 Celestia 有大量交集。Avail 用咗擦除碼同多項式承諾（KZG）去做數據可用性採樣同證明。2021 年 Polygon 團隊介紹 Avail 架構時就有講：Avail 會將每個區塊數據排做一個二維矩陣，對住每一列做擦除編碼（等於加倍額外冗餘資料），再用 Kate（KZG）承諾每列。Light client 隨機抽驗矩陣中嘅 cell 再用 KZG 證明 check 一致性。如果有啲數據唔見咗，light client 喺隨機抽查時極大概率會捉到唔對路。照呢個原則，只有所有數據真係齊晒，Avail 嘅共識會認可個區塊 – 即係同 Celestia 完全一致。Polygon 團隊自己都講過，「Avail 將區塊驗證問題簡化到做數據可用性驗證，成件事可以用 data availability check 以定額成本高效完成」。即係話，Avail 嘅驗證者本身唔執行交易，只係負責確保每個 block 都有齊夠數據，任何人日後都可以隨時復原、執行返啲交易。
One key objective Avail set forth is to enable “standalone chains or sidechains with arbitrary execution environments to bootstrap validator security without needing their own validator set, by guaranteeing transaction data availability.” This paints a picture of many heterogeneous chains – whether built with Polygon’s SDK, Cosmos SDK, Substrate, or other frameworks – all writing their data to Avail and thereby outsourcing one of the hardest parts of running a chain (data availability and consensus). Like Celestia, Avail is agnostic to the execution logic of those chains. They can be Ethereum-like, or UTXO-based, or anything; Avail doesn’t validate the state, it just holds the data and orders it. The Avail team explicitly mentioned supporting chains built with Polygon SDK, Cosmos SDK, or Substrate, highlighting a cross-ecosystem ambition from the start. If Celestia emerged from the Cosmos world, Avail has roots in the Ethereum world, but both
Avail 明確其中一個目標：希望「獨立鏈或側鏈（任意執行環境）都可以靠 Avail 保證交易數據可用性，從而開展驗證者安全，唔使自己建立一套驗證者」。換言之，唔論用 Polygon SDK、Cosmos SDK、Substrate 或其他框架起出嚟嘅各種異構鏈，都可以一齊寫數據去 Avail，直接將運行一條鏈其中一個最難搞嘅部分（數據可用性同共識）外判咗畀 Avail。好似 Celestia，Avail 對你條鏈點執行完全「中立」：你可以好似 Ethereum、又可以 UTXO 基礎、得你話事；Avail 唔 check 狀態，純粹保存數據同排序。Avail 團隊一開始就講明重點支持 Polygon SDK、Cosmos SDK、Substrate 起出嚟嘅鏈——好明顯橫跨多個生態。Celestia 係 Cosmos 世界出身，Avail 就根植 Ethereum，兩者但係...
（內容未完，如要翻譯餘下部分請補充提供）target a neutral ground where diverse chains meet.
以匯聚多元公鏈的中立平台為目標。
One notable difference in positioning is that Avail has emphasized interoperability and connectivity features alongside data availability. In early 2024, when Avail raised a $27M seed round to accelerate development, it revealed a three-pronged product vision called the “Trinity”. This Trinity consists of: Avail DA (the base data availability layer), Nexus, and Fusion Security.
Avail 強調除了數據可用性之外，互操作性和連接性功能亦是其定位上的一大分別。2024年初，Avail 完成了 2,700 萬美元的種子輪融資，以加速發展同時公布了三大核心產品藍圖 “Trinity”。這個 Trinity 包括三個部分：Avail DA（基本數據可用性層）、Nexus（樞紐層）、以及 Fusion Security（融合安全機制）。
-
Avail DA is the core data layer (“Avail provides data space for auxiliary layer-2 networks or rollups” as Coindesk described it). It’s expected to go live by Q2 2024, providing the fundamental service of ordering blocks and ensuring data availability for client chains.
-
Avail DA 是核心的數據層（正如 Coindesk 形容：「Avail 為輔助 layer-2 網絡或 rollup 提供數據空間」）。預計於 2024 年第二季啟動，為用家鏈提供區塊排序及數據可用性的基礎服務。
-
Nexus is described as a “zero-knowledge, proof-based coordination rollup on Avail DA” that will act as an interoperability hub. Essentially, Nexus will be a layer that connects different rollups and chains running on Avail, allowing them to communicate and transact with each other. According to the team, Nexus will serve as a verification hub unifying a wide array of rollups both inside and outside the Avail ecosystem, using Avail DA as the root of trust. By employing zero-knowledge proofs, Nexus can coordinate cross-rollup operations securely. This addresses a crucial aspect of “connecting blockchains”: it’s not enough to share a data layer; chains also need ways to pass messages or assets trustlessly among themselves. Nexus is Avail’s answer to that, enabling an interoperable multi-chain environment on top of the shared data layer.
-
Nexus 被形容為一個基於零知識證明的協調 rollup（協作層），運行於 Avail DA 之上，是互操作性的中樞。實際上，Nexus 會連結所有在 Avail 運作的不同 rollup 及區塊鏈，令它們可以互相通訊及交易。官方指，Nexus 會成為 驗證中樞，統一 Avail 生態圈內外的眾多 rollup，以 Avail DA 作為信任根基。藉助零知識證明，Nexus 能安全協調跨 rollup 運作。這正回應了「鏈與鏈互聯」的核心痛點：單靠共用數據層並不足夠，各鏈之間還需要能無信任地傳遞訊息或資產的方法。Nexus 正正提供了此功能，令共同數據層之上可以建立真正互聯互通的多鏈環境。
-
Fusion Security is the third component, focusing on shared security. It aims to take crypto assets like Bitcoin and Ether and contribute to the security of the Avail ecosystem. While details are sparse, this seems to hint at a system where major external assets could be staked or utilized to secure Avail and perhaps its connected chains. It may involve something akin to re-staking or providing collateral that backs the validity of Avail’s services – conceptually somewhat like EigenLayer’s restaking, but implemented within Avail’s framework. The goal is likely to enhance Avail’s security by involving more capital and more diverse stakeholders (hence “multi-token staking” as one description suggests). Fusion Security is planned for 2025, indicating a longer-term aspiration to make Avail not only a data provider but also a security provider.
-
Fusion Security 是第三個核心，致力實現共享安全。目標是 令比特幣、以太幣等加密資產也能為 Avail 生態圈提供安全保障。雖然細節仍在醞釀中，但概念上似乎會允許大型外部資產抵押於 Avail，又或連同其關聯的鏈一同提供安全性支援。當中或有類似「再質押」的架構，亦可讓外部資產成為 Avail 服務效力的抵押品——與 EigenLayer 的 restaking 類似，但設於 Avail 之內。其目標，是連結更多資本和多元持份者參與（正如有些介紹提到「多代幣質押」），從而鞏固 Avail 的安全度。Fusion Security 預計 2025 年推出，反映 Avail 不單要做數據供應者，還立志成為安全服務供應者。
By pursuing Nexus and Fusion, Avail is explicitly trying to be the “unifying factor” for disparate rollups. Anurag Arjun, Avail’s founder, noted the fragmented nature of Ethereum’s rollup landscape and the need for a neutral party to help coordinate them: “You really need a credible third party like Avail to come and work with all these teams… We essentially want to be that unifying factor.” This reflects a slightly different philosophy from Celestia’s more minimal approach. Celestia focuses on doing one thing (data consensus) extremely well, and relies on others for settlement/bridge layers. Avail is more holistic, aiming to provide data availability and a built-in interoperability layer (Nexus), and even tie in to security economics (Fusion). One could say Celestia is going for the UNIX philosophy (“do one thing well”) whereas Avail is going for a broader platform play under one umbrella.
透過 Nexus 及 Fusion，Avail 目標是成為不同 rollup 的「統一力量」。Avail 創辦人 Anurag Arjun 曾指出，以太坊 rollup 生態極為分散，需要一個中立方來統籌：「你確實需要一個如 Avail 般有公信力的第三方，去協調眾多團隊……我們希望能成為這個統合因素。」這種思路與 Celestia 的極簡取向有別——Celestia 只專注做一件事（數據共識），其他層面則交由第三方負責。而 Avail 則希望一站式包辦數據可用性、內建互操作層（Nexus），甚至涉足安全經濟（Fusion）。可以說，Celestia 取經 UNIX 哲學（只做好一件事），而 Avail 則傾向建立綜合性平台生態。
Despite these conceptual differences, Avail and Celestia are often seen as direct competitors – “rival” data availability networks in the words of the press. Indeed, industry observers have framed it as a race. Avail’s backers include heavyweights like Founders Fund and Dragonfly Capital, signaling confidence that more than one data network might thrive. Coindesk in early 2024 described the emergence of these data layers as “one of the most hotly discussed trends in crypto,” noting they could transform blockchain architecture into a modular design. Celestia’s mainnet arrival in October 2023 “brought the concept to limelight,” and not far behind was EigenDA – an Ethereum-aligned effort we’ll discuss soon – and Avail itself. In this modular blockchain trifecta, each project takes a different path: Celestia via a new L1, Avail via a dedicated layer initially incubated by Polygon, and EigenDA via Ethereum’s restaking ecosystem.
雖然理念上存有差異，外界常將 Avail 與 Celestia 視為直接對手，媒體亦稱之為「數據可用性網絡之戰」。行業觀察員更將這場發展視為一場競賽。Avail 背後獲得 Founders Fund、Dragonfly Capital 等重磅機構支持，顯示市場相信數據網絡不會只有一個贏家。Coindesk 在 2024 初形容這些新數據層的興起為 「加密產業內最熱門討論現象之一」，指它們可望推動區塊鏈進入模組化架構。Celestia 於 2023 年 10 月主網上線「讓此概念一躍成焦點」，之後 EigenDA（將於下文討論）、及 Avail 等亦緊接有進展。在這場模組化公鏈三強中，Celestia 自建 L1，Avail 經 Polygon 孵化專用層，而 EigenDA 則以以太坊再質押生態切入，各自走出獨特路線。
Where does Avail stand currently? As of mid-2025, Avail is transitioning from development into deployment. A public testnet was launched (in mid-2022, Anurag Arjun released an early testnet for Avail), and the project has been refining its protocol with input from the Ethereum and Polygon communities. The fundraising and spin-out in 2023–2024 provided resources to push towards mainnet. Avail’s seed round of $27M (Feb 2024) was earmarked to finish the core products – DA, Nexus, Fusion – with Avail DA expected live by Q2 2024. If that timeline held, Avail’s data layer may already be live or imminently launching as we write. One tangible sign of Avail’s progress is its partnerships: in December 2023, Avail announced an agreement with StarkWare, the developer of the popular StarkNet Layer-2, to collaborate on new “appchains” using StarkWare’s technology and Avail’s data layer. Under this deal, Avail will provide data availability for application-specific chains built with StarkWare’s Madara sequencer (a decentralized sequencer for Layer-3 chains), effectively integrating Avail into StarkWare’s ecosystem for scaling Ethereum. This kind of partnership is significant – it shows Avail positioning itself as the go-to data layer even for advanced use cases like Layer-3 custom chains, potentially in competition or parallel to Celestia. If StarkNet’s ecosystem can spin out appchains that use Avail for data, it validates Avail’s model and brings real usage. Avail has also joined discussions in the broader Ethereum community on how off-chain data availability can complement on-chain scaling. Its documentation emphasizes support for Validiums (off-chain data, zk-proof-secured rollups) and other L2 models that want more throughput by moving data off Ethereum. In essence, Avail wants to be the scalable data hosting layer that all these solutions plug into.
那麼 Avail 現時發展到哪一步？截至 2025 年中，Avail 正由開發逐步過渡至部署階段。其公開測試網早在 2022 年中由創辦人 Anurag Arjun 上線，之後團隊透過以太坊及 Polygon 社群持續優化協議。2023–2024 年的融資及獨立化給予更多資源推進主網。Avail 的 2,700 萬美元種子輪（2024 年 2 月）明確指向完成 DA、Nexus、Fusion 三大核心，並預期 Avail DA 於 2024 年第二季推出。如果一切順利，本文發佈時其數據層或者已經上線，抑或已準備就緒。一個具體進展是其策略夥伴關係：2023 年 12 月，Avail 宣布夥拍人氣 Layer 2 StarkNet 的開發商 StarkWare，利用 StarkWare 技術連同 Avail 數據層合作打造新「應用鏈」（appchains）。在這協議下，Avail 將為擁有 StarkWare Madara 協調器（適用於 Layer-3 之分散式 sequencer）的應用鏈提供數據可用性，實質將 Avail 融入 StarkWare 擴容以太坊的生態。這類合作十分具代表性——代表 Avail 力爭成為進階新應用（如 Layer-3 appchain）的數據首選，可能與 Celestia 并駕齊驅。如 StarkNet 生態能衍生出以 Avail 為數據層的 appchains，將證明 Avail 的模式具有實際價值。Avail 亦活躍參與更廣泛以太坊社群關於「鏈下數據可用性與鏈上擴容互補」的討論，官方文件重點支援 Validium（鏈下數據、由 zk-proof 保安的 rollup）以及其他有意從以太坊移除數據以提升吞吐量的 L2 模型。本質上，Avail 希望成為 所有擴容方案對接的可擴展數據託管層。
From a technical governance standpoint, it’s worth noting that Avail is implementing a Proof-of-Stake network with its own token (likely used for validator staking and possibly for cross-chain fees). Avail’s design mentions “multi-token staking” for decentralization – potentially meaning that validators might need to stake a basket of assets or that multiple assets can be used, but details remain to be seen. The consensus protocol is likely Tendermint-style (Polygon’s teams are experienced in Tendermint and Substrate) but customized for the data availability checks. As Avail becomes independent, it will have to build its own validator community. Interestingly, the Fusion concept hints that Avail might attract security from other chains (via BTC, ETH collateral), which could differentiate it from Celestia’s more self-contained staking of its native token.
從技術治理角度，值得一提 Avail 已採用權益證明（Proof-of-Stake）網絡並有自家代幣（預計用於驗證者質押，亦可能應用於跨鏈付費）。設計中提到「多代幣質押」來提升去中心化——可能代表驗證者須持有多元資產作抵押或可用不同資產，但細節仍有待揭示。共識協議推測會用 Tendermint 類型（Polygon 團隊對 Tendermint 及 Substrate 有豐富經驗），但針對數據可用性作自訂。隨著 Avail 走向獨立，必須發展專屬驗證者社群。有趣的是，Fusion 概念暗助 Avail 可能吸引其他鏈資產（如 BTC、ETH）來增強安全，令其路線有別於 Celestia 主打原生代幣自我質押的作法。
In summary, Avail represents another strong push toward a modular, interconnected blockchain world. It shares the same fundamental insight as Celestia – that separating data availability and consensus into a specialized layer can supercharge scalability – but it wraps that in a vision of cross-chain connectivity and shared security. Avail wants to be the glue that holds many chains together: the common data layer, the bridge hub (Nexus), and even a platform that leverages liquidity from major coins to secure new networks (Fusion). If successful, Avail could enable a flourishing of new appchains and rollups that are all mutually interoperable and easy to spin up, accelerating Web3 innovation across the board. Of course, Avail will also have to compete for adoption: developers may choose Celestia, Avail, or even Ethereum’s in-protocol data solutions depending on various factors (costs, trust assumptions, community). This brings us to the third major player in the race – one that comes at the problem from Ethereum’s angle and focuses on reusing Ethereum’s security for modular projects.
總結來說，Avail 推動區塊鏈世界朝著更模組化、互聯方向邁進。理念上與 Celestia 視野一致——即將數據可用性及共識分拆專屬層級以提升擴容效率——但 Avail 結合跨鏈連通、共享安全等目標。Avail 想成為多鏈的連結「黏合劑」：既是共通數據層，又是橋樑樞紐（Nexus），甚至能挪用主流幣流動性保障新網絡安全（Fusion）。若能成功，有機會催生大量互操作 facilement、易於啟動的新 appchain 和 rollup，推動 Web3 創新遍地開花。當然，Avail 亦須競逐主導權：開發者可根據成本、信任與社群等因素，自行選擇用 Celestia、Avail 或以太坊原生數據方案。由此引伸出第三家主要參與者——他們以以太坊本身為根基，著眼於如何將 ETH 的安全性重用於模組化項目。
EigenLayer: Re-Staking Ethereum’s Security for New Modules
EigenLayer：將以太坊安全再次質押推進新模組
While Celestia and Avail build new base layers to share data and consensus, EigenLayer takes a different approach to connecting blockchains: it extends the security of an existing blockchain (Ethereum) to new use-cases. In essence, EigenLayer is a “re-staking” protocol that allows Ethereum stakers and validators to opt-in to securing additional networks or modules on top of Ethereum. By doing so, it creates a marketplace of shared security – a pool of capital (staked ETH) and trust that other projects can tap into, without needing to launch their own token or validator set from scratch.
Celestia 和 Avail 都是自建新的底層區塊鏈，提供數據和共識層服務。EigenLayer 則另闢蹊徑，以提升區塊鏈互聯：它直接延展了現有區塊鏈（即以太坊）的安全性，讓其應用於全新場景。本質上，EigenLayer 屬「再質押」（re-staking）協議，讓以太坊質押者及驗證者自願選擇保護以太坊以外的網絡或模組。如此一來，打造出一個共享安全性市場——一個大型資本池（已質押的 ETH）及信任基礎，其他項目無需自建代幣或驗證者，就能借力用上。
Imagine you’re an Ethereum validator with 32 ETH staked and securing the Ethereum 2.0 consensus. With EigenLayer, you could “re-stake” that same 32 ETH into EigenLayer’s smart contracts, which then permit you to validate other services known as Active Validated Services (AVS). These AVSs could be anything: an oracle network, a cross-chain bridge, a new sidechain, or even a data availability layer (EigenLayer’s team themselves have a module called EigenDA). When you opt-in, you agree that if you misbehave in any of those services (as determined by their own slashing conditions), your staked ETH can be slashed as penalty. This opt-in “security sharing” is entirely voluntary and modular – each validator can choose which services to support, and each service can set its own requirements and rewards.
假設你是以太坊驗證者，質押了 32 ETH 參與 Ethereum 2.0 的共識。透過 EigenLayer，你可將同一筆 32 ETH「再質押」進入 EigenLayer 智能合約，這樣你便能參與負責其他稱為 AVS（Active Validated Services，主動驗證服務）的驗證。AVS 服務可包括：預言機網絡、跨鏈橋、新側鏈，甚至是數據可用性層（例如官方的 EigenDA 模組）。加入 AVS 等同於同意如有違規（每個 AVS 有自己的懲罰條件），你質押的 ETH 可遭懲罰扣減。此種 opt-in 式「共享安全」完全自願且模組化——每位驗證者可自由選擇參與哪些服務，每個服務又可以自設要求及獎勵。
EigenLayer’s core idea was born from the observation that Ethereum’s proof-of-stake has amassed a large, economically secure validator set (over $40B in staked ETH by 2025), which is a form of “cryptoeconomic energy” that could secure more than just the Ethereum blockchain itself. There is a lot of redundant security in the crypto ecosystem – hundreds of new projects launch their own token and mini-validators, often
EigenLayer 的核心思想，是觀察到以太坊權益證明機制已聚集一個龐大且經濟上極具安全性的驗證者網絡（截至 2025 年，質押 ETH 超過 400 億美元），這股「加密經濟能量」完全可用於保障以太坊以外的更多新用途。放眼整個加密生態，安全性高度重覆——數以百計新項目都要自建代幣和微型驗證者，大多數時候……struggling to get enough honest participants. EigenLayer says: why not recycle the security from Ethereum and allocate it to these new projects, so you don’t need a new token and you get instant decentralization from Ethereum’s thousands of validators? In doing so, it hopes to accelerate innovation by lowering the barrier to launching new infrastructure. Developers can focus on the unique logic of their service, while trusting Ethereum’s validator community (via EigenLayer) to run the nodes and enforce the rules.
難以招募足夠誠實嘅參與者。EigenLayer 話：點解唔重用 Ethereum 嘅安全性，分配畀啲新項目，就唔使再搞新 token，而又即時獲得 Ethereum 成千上萬個驗證者嗰種去中心化？咁做可以降低新基建項目嘅門檻，加快創新。開發者可以專注做服務本身嘅邏輯，而節點運作同規則執行就交俾 Ethereum 嘅驗證者社群（透過 EigenLayer）處理。
Launched by the startup EigenLabs (founded by researchers including Sreeram Kannan), EigenLayer began testing in 2023 and quickly captured attention as a potential game-changer for the Ethereum ecosystem. Vitalik Buterin and other Ethereum core developers have acknowledged both the promise and the risks of restaking. On one hand, reusing Ethereum’s security can make the whole ecosystem more robust and interconnected. On the other, there are caution flags: if too many external services piggyback on Ethereum’s stakers, a failure or exploit in one could theoretically cascade and slash a lot of ETH, potentially threatening Ethereum itself. Buterin in mid-2023 warned against overloading Ethereum’s consensus with excessive extra duties (like watching other chains via restaking), lest it increase complexity and attack surface. EigenLayer’s design tries to mitigate risks by using a “opt-in and opt-out” model: only validators who explicitly choose to restake are affected by EigenLayer, and slashing conditions are isolated per service (so an incident on one AVS would only slash those who opted into that AVS, not the entire Ethereum set). Essentially, Ethereum’s base consensus remains unchanged; EigenLayer operates in smart contracts on Ethereum that a subset of validators interact with.
EigenLayer 係由初創公司 EigenLabs（有 Sreeram Kannan 等學者創立）推出，2023 年開始測試，之後好快就引起圈內注意，被視為有機會改變 Ethereum 生態格局嘅新元素。Vitalik Buterin 同其他以太坊核心開發者都承認 restaking 有巨大潛力，但都有風險。一方面，重用 Ethereum 嘅安全性，令整個生態更加穩陣又有連接。但係，專家有警告：如果太多外部服務依靠 Ethereum 嘅 staker，萬一某個出事或被攻擊，理論上會引發連鎖效應，大量 ETH 被 slashing，最壞情況會威脅埋 Ethereum 本身。Buterin 喺 2023 年中就警告，不要過度「加重」Ethereum 共識層嘅責任（例如用 restaking 睇住其他鏈），否則會令複雜度、攻擊面飆升。EigenLayer 設計用 opt-in/opt-out（自願參與或退出）冇咁高風險：只有自願 restake 嘅驗證者會受影響，而每個 AVS（Actively Validated Service）嘅 slashing 條件都獨立，就算有 AVS 出事，都只會影響參與咗嗰個 AVS 嘅驗證者，而唔會斬晒全以太坊。簡單講，Ethereum 基本共識冇變，EigenLayer 只係用智能合約運行，淨係部分驗證者會互動。
By early 2024, the excitement around EigenLayer translated into major funding: EigenLabs raised $50 million in a Series A, and shortly after secured an additional $100 million investment from Andreessen Horowitz (a16z). This war chest (over $150M total reportedly) indicates high confidence that restaking will become an integral part of Ethereum’s roadmap. Indeed, in April 2025, Nansen reported that EigenLayer had about $8 billion TVL in restaked assets – effectively a measure of ETH participating in restaking. If accurate, that suggests a significant portion of Ethereum stakers are on board, attracted by the promise of earning additional yields from securing multiple services. EigenLayer has even conducted an airdrop and launched its own token ($EIGEN) incentives for early adopters, distributing 15% of its token supply to early restakers in a Season 1 airdrop. This was likely done to decentralize governance and reward those taking on the early risk of restaking.
去到 2024 年頭，EigenLayer 已經吸引到大量資金：EigenLabs A 輪融資籌到五千萬美元，之後仲有 a16z（Andreessen Horowitz）加碼一億美元。總計據報已經有超過一億五千萬美金現金庫，顯示大家對 restaking 作為以太坊未來路線圖組成部分信心好大。到 2025 年 4 月，Nansen 報告指 EigenLayer 已經有約 80 億美金 TVL 參與 restake，等於好多 ETH 持有者同時信任呢個模式，為賺取額外收益而幫手鎖定多個服務安全。EigenLayer 之前亦都做咗空投，而且出咗自己 token（$EIGEN），Early adopter 喺 Season 1 空投分咗全供應量 15%，目的都係分散治理權力，獎勵早期承擔風險嘅 restaker。
So what new capabilities does EigenLayer actually unlock? We can break it down by looking at some example AVS (Actively Validated Services) that either already exist or are envisioned:
咁 EigenLayer 到底解鎖到啲咩新功能？可以以現有或者計劃中嘅 AVS（Actively Validated Services）為例：
-
EigenDA (Data Availability): As mentioned, one of EigenLayer’s flagship modules is EigenDA, a data availability layer similar in goal to Celestia/Avail but secured by Ethereum restakers. If EigenDA comes online, a project launching a rollup could choose to use EigenDA for data availability, effectively leveraging Ethereum’s validator set (via restakers) to guarantee data is published. This provides an Ethereum-aligned alternative to Celestia/Avail. Coindesk noted that EigenDA was “in development” as of late 2023. By tying into EigenLayer, EigenDA wouldn’t need its own separate token or large validator recruitment; it inherits security from re-staked ETH. This underscores how EigenLayer and Celestia/Avail could become competitors: a rollup could either post data to Celestia (with Celestia’s own token and validators) or post to EigenDA (with ETH validators via restaking).
-
EigenDA（Data Availability）：正如上面講過，EigenLayer 其中一個旗艦模組叫 EigenDA，係個資料可用性層，目標同 Celestia/Avail 差唔多，但係以太坊 restaker 提供安全。如果 EigenDA 成功上線，日後有 rollup 項目可以直接選擇用 EigenDA 做資料層，等於即時用到 Ethereum 一大班驗證者（透過 restaker）保證數據有公開。咁對於 Celestia/Avail 就係另一種以太坊陣營嘅選擇。Coindesk 指 EigenDA 去到 2023 年尾都仲係「研發中」。配合 EigenLayer，EigenDA 唔駛自己出 token，也唔駛請大班驗證者——靠 restake ETH 就有保安力。咁解釋到 EigenLayer 有機會直接同 Celestia/Avail 競爭：rollup 可以揀將數據丟去 Celestia（用 Celestia token 同 validator）或 EigenDA（用 ETH validator + restaking）。
-
Bridges and Cross-Chain Services: Cross-chain bridges have notoriously been weak points, often secured by a limited set of validators or multi-sigs leading to hacks. With EigenLayer, a bridge could be built as an AVS that uses dozens or hundreds of Ethereum validators (those who opt-in) to validate cross-chain transfers, vastly increasing its trustlessness. Because those validators have skin in the game (their ETH stake), attacking or bribing them is far more expensive than attacking a typical standalone bridge. Several teams have expressed interest in building EigenLayer-secured bridges and messaging protocols, which could enhance interoperability across blockchains.
-
跨鏈橋同跨鏈服務：跨鏈橋過去成日出事，因為多數淨係靠好少 validator 或 multi-sig，容易中招。有咗 EigenLayer，啲橋就可以升級做 AVS，搵幾十甚至幾百個自願 opt-in 嘅 Ethereum validator 來審批跨鏈操作，安全性提升好多。因為啲 validator 本身有 ETH 壓住，所以攻擊或者收買佢地個成本遠高於傳統獨立橋。依家已有團隊想建基於 EigenLayer 嘅跨鏈橋同消息協定，幫區塊鏈之間互通更容易。
-
Oracles: Services like Chainlink provide off-chain data to blockchains and are critical infrastructure. An oracle network could use EigenLayer to get a ready-made set of staked ETH validators to report data and get slashed if they lie. This could either complement or compete with existing oracle providers, bringing more decentralization. For example, an oracle AVS might have EigenLayer stakers collectively sign price feeds or random beacons.
-
預言機（Oracles）：好似 Chainlink 嘅服務咁，預言機為鏈提供鏈外資訊，係區塊鏈基礎設施。EigenLayer 俾預言機網絡用現有 staking ETH validator 報數，如果有亂咁報會被 slash。呢個未來可以同現有預言機方案競爭或補足，令整個機制更去中心化。舉例話，預言機 AVS 可以由 EigenLayer staker 集體簽名報價 or 隨機數據。
-
New Consensus or Research Chains: A project inventing a novel blockchain or sharding mechanism could launch it as an AVS, essentially piggybacking on Ethereum’s validator set for security while running its own consensus rules in parallel. This is a bit like Polkadot’s model (parachains reusing relay chain validators), but EigenLayer does it on Ethereum in a permissionless, opt-in way. It creates a sandbox for consensus experimentation where the economic security is there from day one. We see early hints of this in collaborations like Espresso Systems (behind the Espresso sequencer for rollups) working with EigenLayer – they can focus on their rollup/sequencer tech and rely on restaked ETH for security.
-
新共識鏈／研究性鏈：有啲搞新型區塊鏈或分片嘅研究項目，可以直接做成 AVS，等於攞 Ethereum 嘅 validator 帶保安，但行自己條共識規則，同步跑。呢個有啲似 Polkadot 嘅平行鏈攞中繼鏈安全，但 EigenLayer 係 permissionless、opt-in，係以太坊內進行。咁即日可以做共識創新沙盒，一開始都有經濟安全保護。例如 Espresso Systems 同 EigenLayer 合作，Espresso 專注做 rollup/sequencer，安全方面就靠 restake ETH。
EigenLayer operates entirely on Ethereum’s Layer 1 via smart contracts. Validators who join must run additional software (for each AVS they support) and stake ETH into EigenLayer contracts, but they still perform their normal duties for Ethereum itself. EigenLayer introduces an operator-delegation model: not every ETH holder who restakes needs to run the new services themselves; they can delegate to qualified operators who run nodes for the AVS. This means even if you’re staking via a liquid staking token (LST) or through an exchange, you could potentially opt those into EigenLayer and have some professional node operator handle the actual validation tasks. It’s a two-sided market – AVS creators want as many validators as possible to opt-in, and validators want worthwhile AVS with good rewards. EigenLayer’s contracts mediate this, and they enforce slashing across all opted-in assets if a validator is proven to misbehave in an AVS. By April 2025, the ecosystem was taking shape: Binance’s staking service integrated EigenLayer options, and projects like Renzo (a liquid restaking solution) launched to make restaking accessible.
EigenLayer 係完全運作喺 Ethereum L1 嘅智能合約上。想參加嘅 validator 需要多跑多啲軟件（每個 AVS 要對應個 client），並且將 ETH lock 入 EigenLayer 合約，但都仍然做緊自己本業以太坊 validator 嘅工作。EigenLayer 採用 operator 委託制：唔一定每個 restake ETH 嘅人都自己跑新服務，可以委託俾專業 operator 幫手。而且就算用流動質押代幣（LST）或者經交易所質押，都可以拎去參與 EigenLayer，再俾專業節點操盤管理 validation。市場係雙邊—AVS 建立者想越多 validator 參加越好，validator 自己又想揀有好回報嘅 AVS。EigenLayer 合約會協調同管理，發現 AVS 有 validator 出錯就自動 slash 所有參與咗嗰個 AVS 嘅資金。去到 2025 年 4 月個生態都愈嚟愈齊：幣安質押已經整合 EigenLayer 選項，Renzo（流動 restaking solution）等項目都已上線，令 restaking 更易參與。
Of course, with power comes risk. EigenLayer’s team and community are very aware of the “contagion” risk – if one AVS has a flawed slashing condition or is hostile, it could grief Ethereum validators. To mitigate this, EigenLayer is rolling out in phases, initially with whitelisted, vetted AVS and limited scope, and plans for more permissionless addition of AVS later. They also emphasize “attributable slashing” – only slash those who actually did wrong, and design AVS so that any slash is provably linked to malicious action by a specific subset of validators. This avoids scenarios where honest validators get unfairly slashed due to others’ misdeeds. The contract and crypto-economic design is complex, but it’s undergoing audits and scrutiny by the Ethereum research community. If successful, EigenLayer could make Ethereum not just a base layer for rollups, but the security bedrock for myriad modules – effectively turning Ethereum into a decentralized trust services platform.
當然，有權就有風險。EigenLayer 團隊同社群好清楚「傳染性」危機——如果有 AVS 割錯、或者有惡意，可能會影響 Ethereum validator。為咗降低風險，EigenLayer 係分階段推，初時只准入白名單、審查過啲 AVS，規模都有限，遲啲先開放更多 permissionless AVS。佢地特別強調「可追溯 slashing」——淨係割真犯錯嗰啲，AVS 嘅設計要確保 slash 一定能夠證明係特定 validator 惡意行為，唔可以搞到冇事嗰啲 validator 一齊捱刀。呢套合約同加密經濟設計好複雜，依家都俾以太坊研究社群審計緊。如果搞得掂，EigenLayer 能夠令 Ethereum 唔只係 rollup base layer，仲係周邊模組嘅安全根底——等於令 Ethereum 晉身成去中心化信任服務平台。
To put it simply, EigenLayer connects blockchains by connecting their security. Instead of every new chain or service floating alone with its own small life raft of security, EigenLayer ties many to the big aircraft carrier of Ethereum. A common set of validators (ETH stakers) can verify events on multiple chains, creating natural interoperability. For instance, if the same validator set secures Chain A and Chain B (via EigenLayer), then a transaction proven on Chain A can be recognized on Chain B without need for an external bridge, since those validators witnessed it directly. This could streamline cross-chain functionality and reduce reliance on third-party bridges. It’s somewhat analogous to how Polkadot’s parachains all share one validator set and thus can communicate under a unified trust model – but here it’s happening on Ethereum in a voluntary fashion.
講到尾，EigenLayer 係將鏈同鏈嘅「安全性」連接埋一齊，從而連結區塊鏈。以後每條新鏈或者新服務唔使各自揸住細細舢舨漂泊，可以直接綁上 Ethereum 呢艘航空母艦。有共同一班 validator（ETH 持有者），可以同時驗證多條鏈，自然實現互操作。例如同一 set validator 經 EigenLayer 同時保 Chain A 同 Chain B，A 鏈嘅交易只要被呢班人驗證咗，B 鏈都即時識得到，唔使靠第三方跨鏈橋，因啲 validator 親眼睇住結果。咁可以簡化跨鏈操作、減少對第三方橋信任需求。少少類似 Polkadot 嘅平行鏈共用 validator，不過呢個係喺以太坊、而且自願 opt-in 方式發生。
EigenLayer’s rise also raises interesting governance questions. Ethereum’s ethos has been cautious about too much complexity at the base layer. EigenLayer builds on top as a layer-2-like solution (though it’s not about scaling, but about extended functionality). If a large portion of ETH becomes re-staked, one could argue EigenLayer becomes an extension of Ethereum’s consensus. The community will need to watch whether any adverse incentives or centralization creep in (e.g., will large staking pools dominate certain AVS? Will restaking disproportionately benefit big players?). So far, the approach has been relatively decentralized – a report noted concerns about large pools, but also that EigenLayer’s model aims to mitigate systemic failures by isolating risks. The fact that a16z and others have poured funding suggests they see restaking as a pillar of the future crypto infrastructure.
EigenLayer 快速崛起都帶嚟咗治理方面多啲問題。以太坊文化本身一向戒慎恐懼、唔想 base layer 太複雜。EigenLayer 係「疊」上去，類似 layer-2 （雖然唔係為咗擴容，而係增加功能）。如果太多 ETH 都拎去 restake，有啲人會話 EigenLayer 等於以太坊共識嘅延伸。社群要睇住會唔會有反效果或者中心化街主導（例如大 pool 壟斷某啲 AVS？restake 利益會唔會集中俾大戶？）。目前步伐都算去中心，不過都有人擔心大 pool 問題，而 EigenLayer 嘅模式又係設計咗 risk isolation，想避系統性失敗。大機構（如 a16z）咁落重注，亦都證明佢地睇 restake 係未來加密基建大柱石。
Complementary or Competitive? The New Modular Landscape
Having explored Celestia, Avail, and EigenLayer individually, it’s clear they share a common vision: a multi-chain crypto ecosystem that is more scalable and interconnected than the siloed, monolithic chains of yesterday. Yet they approach it from different angles and will likely coexist with both cooperation and
經過探索 Celestia、Avail、EigenLayer 後，好明顯佢地其實都想推一個更可擴展、連接性高嘅多鏈加密生態，取代以往各自為政、孤島式 monolithic 鏈。不過大家切入角度唔同，未來好可能會一邊競爭一邊合作並存。competition.
Celestia vs. Avail: 兩者都是純粹的數據可用性網絡，為執行層服務，難免會被拿來比較。Celestia 具備先行者優勢，較早推出、獲得更多公眾關注（亦有發行代幣）。Avail 則有 Polygon 背景，在以太坊擴容社群內紮根深厚，有望吸引已經在 Polygon／zk-rollup 生態的項目。技術上，兩者十分相近——同樣採用數據抽樣、糾刪碼、PoS 驗證人等技術。有一個如前所述的策略性分野：Celestia 堅持極簡主義，而 Avail 則加入互操作性（Nexus）及潛在的多資產安全性（Fusion）等功能。根據 Lithium Finance 的分析，「Celestia 是首個將數據可用性和共識從執行層分離的網絡……Avail 則採取稍稍不同的路線，設計上支援多鏈運作，並通過多代幣質押來加強去中心化。它亦容許應用鏈之間互動，而不用嚴格同步。」 換言之，Celestia 讓人能靈活建立與其他生態連接的獨立 rollup，而 Avail 強調跨鏈整合和多元安全輸入。實際上，項目會根據效能、成本及生態契合度來選擇數據層。將來或會出現分工：如 Avail 若獲 StarkWare 等 rollup 團隊整合，或可在以太坊 Layer-2 領域佔優，而 Celestia 可能吸引更多主權鏈及 Cosmos 風格的 appchain。亦有可能相反，端看網絡效應及穩定性。可以肯定的是：兩個網絡都押注會有大量新鏈誕生並需要他們的服務——這幾乎可以確信，因行業正多元化至遊戲、社交媒體、企業等專屬公鏈。
EigenLayer vs. Celestia/Avail: 乍看之下，EigenLayer 屬於另一種類——它本身並非數據網絡。然而，其下的 EigenDA 讓它直接與 Celestia／Avail 競爭數據可用性服務這一角色。若 EigenDA 正式上線，某個 rollup 可能會考慮用 EigenDA（享有以太坊安全性保障）還是 Celestia／Avail。EigenDA 若 rollup 本身以太坊為中心，或許能帶來更低信任假設（以太坊的經濟安全超強）和便利。Celestia 則可能成本更低或更主權（毋需依賴以太坊）。最終還是取決於經濟利益：每個數據層的手續費有多高，整合難易程度如何？亦有可能這些方案各自取長補短：例如，一個 optimistic rollup 可以把 fraud proof 發去以太坊（用上以太坊安全性），而大量交易數據則放在 Celestia（利用 Celestia 高吞吐量）。事實上，這一類設計已有討論（用 Celestia 做數據層、以太坊做結算／最終確定）。Avail 的 Nexus 以及 EigenLayer 的互操作努力同樣可協作——例如，EigenLayer 的 oracle 資料流入已連接 Avail 的 appchain。
EigenLayer 特別之處是可支援多元用途，絕不限於數據層本身。理論上，甚至有朝一日 Celestia 或 Avail 都能成為 EigenLayer 的 AVS，與以太坊合併驗證人集。雖然因自有代幣及社群而較難成真，卻顯示 EigenLayer 更像是一個平台，而非單一服務。我們或許也會見到 Celestia 和 Avail 借鏡 EigenLayer 的玩法：例如 Celestia 日後可否讓 $TIA 在多個 Celestia 實例間重質押，或讓其他鏈借用 Celestia 的驗證人集？現時已將共識和執行分開，日後透過 IBC 等手段實現 Celestia 與其他 zone 間的共享安全也不無可能。
互操作性與橋接：三個方案的共同目標，都是讓區塊鏈互動更無縫。Avail 的 Nexus 將連通 Avail 上的 rollup。Celestia 則依賴 IBC 及外部橋（如 Hyperlane）接通生態。EigenLayer 可以強化跨鏈 oracle 與橋接網絡。最終，使用者並不在意一個應用運行在哪條鏈上——他們只求資產或數據方便流動、最終結果可信。這些創新正趨於融合，未來一個用家或會用到限定用途的 rollup：數據儲存在 Celestia，證明發送到以太坊（可能經 EigenLayer），而又能用 Avail 的 Nexus 橋與其他 rollup 原生交換資產。背後運作複雜，但如設計得當，這些複雜性會被抽象掉，最終使用者只見交易更快更平更統一的多鏈錢包體驗。
歷史脈絡與展望：值得回顧一下我們是如何走到這一步。2010 年代後期，擴容焦點還是鏈上分片（以太坊 2.0 最初藍圖，後來轉向）和多鏈網絡如 Polkadot（2020 推出）及 Cosmos（2019 上線、2021 啟用 IBC）。Polkadot 帶來跨 parachain 的共享安全構想；Cosmos 則開創 IBC 無縫互通但安全性各自為政。今日模組化路線，可以視為綜合這些思路：Celestia 和 Avail 成為數據／共識的共享安全層，讓多條鏈借用（有點像 Polkadot 的 relay chain，但不負責狀態執行、也不那麼緊耦合），而 EigenLayer、Nexus 等協議則強調跨鏈通訊（類似 Cosmos 的橋接理念）。有趣的是，以太坊自己也轉向「rollup 為中心」路線，把自己定位為 rollup 的結算及數據層。2023 年的 Proto-danksharding（EIP-4844） 首次加入便宜的 blob 空間讓 rollup 用。未來如果完成全面 danksharding，以太坊本身都會變成高產能的數據可用性層。即是說，以太坊 L1 也漸趨模組化（專注於共識和數據，把執行交給 L2）。這種路線圖與 Celestia／Avail 等項目的理念暗合，說明模組化設計已被廣泛視為未來趨勢。
然而，獨立安全層的出現也帶來治理和激勵方面的新問題。若大部分活動都移到這些層面，價值和手續費怎分配？例如，Celestia 代幣會否捕捉到所有用它的 rollup 的費用？以太坊又會否因為作為 EigenLayer AVS 保安獲得新價值？未來可能會出現跨多層的 MEV（礦工／最大可提取價值）市場——例如 Celestia 區塊生產者競投 rollup 排序權等。多層間的協調（如如何確保 rollup 狀態必須等 Celestia 數據先決定）必須審慎處理，以免同步問題或成為攻擊向量。這些都是現正積極研究中的議題。
挑戰與批評
雖然模組化方案展現強大潛力，但同時帶來一系列新挑戰：
-
複雜度：加入多層（數據層、結算層、執行層…）令整體架構更複雜，潛在故障點更多，同步假設增加。如何確保所有層相互配合並不容易。新型失效模式可能會冒出——例如若數據可用性層掛掉或嚴重延遲會怎樣？依賴它的 rollup 可能即使執行層一切正常，也會被迫停機。
-
延遲：多了一層等於交易潛在延遲。假如 rollup 需要等 Celestia 數據 final，再等 EigenLayer oracle 更新，整個流程會慢下來。現時設計正努力優化減低這個問題（Celestia 已推近即時 finality，有幫助）。
-
經濟激勵對齊：每層都有自己的代幣（Celestia 的 TIA、Avail 可能有自家幣、EigenLayer 用 ETH 同時有 EIGEN 管治／獎勵用途）。要令多方激勵對齊絕不容易。例如，如果 Celestia 持幣人決議大幅加價，會否迫走 rollup 選擇 Avail？又或 EigenLayer 的治理和以太坊核心開發在什麼服務該開放上出現分歧？多重社群之間需要持續溝通協調。
-
安全假設：Celestia 和 Avail 的安全取決於其驗證人組成以及諸如抽樣需有至少 20% honest 节點等假設。若這些網絡分佈度不夠或抽樣邏輯有漏洞，使用者恐有滅頂之災。EigenLayer 則依賴以太坊的安全性，但同時疊加自身風險（合約漏洞、連鎖 slashing 爭議等）。共享安全，即是共享風險——既是賣點，也是隱憂。若某個被廣泛應用的層遭攻破，可能波及大量鏈。例如 Celestia 共識若出大漏洞，數百條依賴的 rollup 全部受挫。說回頭，也因為拆分，即使某執行層爆 bug，也不會拖垮整個生態，只影響該 rollup。
-
合規及社會因素：系統愈互聯，或愈易受監管機構關注，因它們愈來愈像大型網絡而非獨立小社群。另一方面，社會面會有既得利益阻力（例如你本來是 L1 項目，接納 Celestia 等同放棄自家驗證人社群）。現有項目真轉型上模組化會須時日。
因此，雖然區塊鏈行業以模組化連接之戰已正式展開，這不是短跑，而是馬拉松。上述三大方案仍處於積極開發或早期部署階段。Celestia 已進入主網測試期，生態正在成形；Avail 準備主網和相關模塊推出；EigenLayer 則在 2024 年逐步解鎖更多以太坊主網服務。我們大機會會見到平行試驗——比如有大熱遊戲／社交 app 在 Celestia 啟動自家鏈的案例等等。Certainly! Here is your requested translation, following your instructions (skipping translation for markdown links):
或者一個主要 DeFi 協議採用 EigenLayer 預言機）同時可能經歷一啲挫折（例如喺呢啲新系統入面發現早期漏洞或者經濟攻擊）。
Conclusion: Toward a Modular, Interconnected Blockchain Ecosystem
Celestia、Avail 與 EigenLayer 嘅出現，代表住區塊鏈設計嘅範式轉移。加密社群唔再一味建立越嚟越龐大、一鏈稱霸嘅系統，而係開始喺唔同層面追求專業化同跨層協作。呢個模組化願景帶嚟前所未見嘅可擴展性——每秒數以百萬計嘅交易，分散喺唔同嘅執行分片之間——同時透過共享驗證同抽樣技術，保留甚至提升安全性。模組化同時俾咗開發者更大嘅創新自由：可以自由組合唔同組件（例如喺呢度攞安全性、嗰度攞數據、執行用自己鍾意嘅方式），建立針對應用需要度身訂造嘅平台。
未來幾年，我哋好可能會見到大批新區塊鏈唔再孤立存在，而係插入一張基礎層同服務嘅網絡。一個 DeFi 交易所可以作為 rollup 喺某個數據網絡運行，一個遊戲宇宙可以用以太坊 restaker 作安全保障自己開一條鏈，佢哋通過標準化嘅橋樑或者樞紐互通有無。用戶甚至可能唔知道自己究竟喺邊條鏈——好似用互聯網應用嘅人唔會知自己數據包打咗邊個 data center——佢哋只係信賴底層模組化基建做緊應份嘅嘢。
值得留意，呢種模組化方式並唔係零和遊戲。Celestia、Avail、同 EigenLayer 各自專注唔同領域，可以專心發展其自身優勢，又可以喺邊緣互相協作。例如可以想像一個 Celestia rollup 用到 EigenLayer 提供嘅預言機服務，或者有 Avail 的 appchain 喺以太坊結算重要證明。最終目標都係打造一個更互聯嘅區塊鏈宇宙，使價值同資訊可以更加自由、安全地流動。
過程中會遇到唔少挑戰。各項目需要長時間證明自己安全可靠。經濟模型要成立——用戶負擔到兩三層基礎設施嘅費用嗎？初步跡象係有利嘅，因為專業化往往帶嚟效率提升（例如 Celestia 高吞吐可以令單位數據成本降低）。開發者都要適應設計模組化應用嘅新學習曲線，但 Optimism 推出嘅 OP Stack、Cosmos SDK 等框架已經開始化，方便大家插拔唔同數據可用性後端或者結算層。工具同標準都要同步成熟（例如：點樣喺以太坊內部驗證 Celestia 數據可用性，或者標準化 EigenLayer 嘅 slashing 條件）。
去到 2025 年，賽跑已經展開。Celestia 團隊話自己將「blockspace 由撥號升級至寬頻」，而家瞄準「光纖級」；Avail 創辦人冀望做「各種 rollup 嘅統一點」；EigenLayer 嘅創辦人預計隨住以太坊安全可重用，「創新效率可提升百倍」。依家係區塊鏈基建最令人興奮嘅時代——呢啲雄心勃勃嘅專案已唔係白紙一張，而係守護緊真金白銀嘅網絡。對加密社群及更廣闊世界，模組化區塊鏈代表技術發展已經準備好發展到可支援十億用戶而唔需要犧牲去中心化或者互通性。
呢場賽跑最終嘅終點——一個全面互聯、模組化嘅加密生態系統——仍然喺前方。但隨住 Celestia、Avail、同 EigenLayer 不斷突破界限，我哋正一步步邁向一個如網絡咁彈性同互聯嘅區塊鏈互聯網，同時擁有區塊鏈應許嘅可證明信任同安全。到最後，呢場賽嘅最大贏家好可能係用戶同開發者——佢哋將可以享受一個更快、更平、又完全無縫銜接嘅區塊鏈體驗，真正實踐當初點燃呢個行業嘅各種理想。