AO如何打造适用于AI Agent的去中心化计算网络

Odaily星球日报Published on 2024-06-20Last updated on 2024-06-20

Abstract

本文从网络架构以及特性方面,解读AO如何打造适用于AI Agent的去中心化计算网络。

无需信任地执行任意代码,并足以共享给全世界使用,世界计算机的梦想深深地扎根于去中心化网络。在以太坊之后,许多基础设施项目都作出了尝试,Arweave 即将推出的 AO 网络也是这些尝试之一。

对于一个“世界计算机”来说,可以粗略地分为数据的计算、访问与存储三个部分,Arweave 过去一直在充当“世界硬盘”的角色,AO 网络(Actor Oriented)则引入了通用计算的能力,并提供了智能合约。

AO:基于 Actor 的通用计算网络

目前的主流去中心化计算平台分为两类,即智能合约平台与通用计算平台;智能合约平台以 Ethereum 为代表,网络共享全局的状态内存,对改变状态的运算过程进行共识,因为共识要求大量的重复运算,因此高成本下仅用于处理高价值业务;通用计算网络不对运算过程本身进行共识,而是根据业务验证计算结果,处理请求顺序,不存在共享的状态内存,这降低了成本,允许网络扩展到更多领域的计算,这类以 Akash 等算力网络为代表。

当然,还有一些项目基于虚拟机安全的安全假设,将通用计算与智能合约融合。即共识只处理交易的顺序,并对计算结果进行验证,多个状态变化计算在网络节点中并行处理,计算的环境虚拟机保证了确定性结果,因此只要交易顺序一致,最终状态也将一致。

这类网络由于不共享状态内存,扩容成本很低,多个任务可以并行计算且互不影响。这类项目往往基于 Actor 编程模型,代表就是 ICP,而 AO 也属于此类。Actor 下每一个计算单元被视为单独的智能独立处理事务,计算单元之间通过通讯交互(Actor 是传统 Web2 服务中非常常见的架构)。AO 标准化了 Actor 的消息传递,实现了一个去中心化的计算网络。

与传统被动触发的智能合约(如 Ethereum/Solana 智能合约)不同,具备通用计算 Actor 下的 AO 可以通过一致固定时间循环触发的“cron”方式,来实现智能合约的主动运行,例如一个持续监控套利空间的交易程序。

可快速扩容的去中心化计算能力,Arweave 的超大数据存储能力,Actor 的编程模型,与主动触发交易的能力,这让 AO 网络非常适合托管 AI Agent。AO 也支持将 AI 大模型引入区块链的智能合约中运行。

AO 网络特性

上文介绍过 AO 与智能合约网络的区别,AO 不对计算过程进行共识,对交易顺序进行共识,并默认虚拟机的运行结果是确定性的,从而实现最终状态的一致性。

AO 还具备一定的灵活性,网络以模块化方式设计。网络中存在三种基本单元,调度单元 SU、计算单元 CU 与信使单元 MU。

一个交易被发出,作为通讯层的信使单元接受交易,验证签名,转发给调度单元;调度单元可以看做 AO 与 AR 链的连接点,帮助网络对交易顺序进行排序,并上传至 AR 链完成共识,目前的共识方式是 POA(权限证明);对交易顺序的共识完成后,任务被分配给计算单元,CU 负责处理具体计算,结果返回 MU 转发给用户。

CU 集可以看做是一个去中心化的算力网络,在完整的经济学规划下,CU 节点需要质押一定资产,通过计算性能、价格等因素竞争,提供算力赚取收益,如果出现计算错误的情况,会被罚没资产。这是一个标准的经济学保障。

AO如何打造适用于AI Agent的去中心化计算网络

AO 与其他网络的区别

AO 做为通用计算平台,与 Ethereum 等智能合约平台的区别显而易见。与 AR 同为“世界硬盘”的 Filecoin 也推出了自己的智能合约平台 FVM,但这是一种等效于 EVM 的状态共识机架构,且在体验上不及 Ethereum 等传统智能合约平台。

与 Akash、io.net 等去中心化计算网络不同,AO 依然保留了智能合约能力,AO 最终也在 AR 存储上维护了一个全局的状态。

实际上,与 AO 在架构上最相似的是 ICP。ICP 创造了异步计算区块链网络的最早范式,AO 在很大程度上延续了 ICP 的设计,比如仅对交易顺序排序、相信虚拟机确定性计算、Actor 模型异步处理等。

最大的不同点在于,ICP 是基于容器维护状态,即每个智能合约容器可以只维护自己的 private 状态,或者对状态读取设置条件;而 AO 具备一个共享的状态层,即 AR,任何人都可以通过交易顺序与状态证明复原全网状态,这一定程度上增加了网络的去中心化能力,但也丧失了 ICP 中特殊隐私业务的实现可能(比如客户有隐藏套利路径的需求)。

在经济与设计层面,ICP 为了保证网络性能,对参与节点作出了较高的硬件要求,这造成了较高的门槛,而 AO 相对以公平发射、无准入的方式运行,质押即可参与竞争挖矿。ICP 网络选择了大堆栈的实现方式,为了性能牺牲了灵活性,而 AO 使用了模块化的设计,MU、CU、SU 分离,用户也可以自选虚拟机的实现方式,这也降低了一些开发者进入的成本。

当然 AO 也可能存在与 ICP 一样的系统缺点,比如 Actor 异步模型下跨合约交易缺乏原子性,这会导致 DeFi 类应用发展困难,AgentFi 的构想似乎很难在短时间内实现;脱离传统智能合约范式的新计算模式,也对开发者提出了更高的要求。但 AO 架构下 wasm 虚拟机最高能管理 4 GB 的限制,也导致部分复杂模型无法在 AO 上使用。由此看来,AO 选择 AI Agent 的路线实为扬长避短,有趣的是,ICP 也在 2024 年年初宣布专注于 AI 领域。

当然,对比 ICP 50 亿美金的总市值来说,AR 目前总市值 22 亿美金,仍有不小的差距。在 AI 大发展的背景下, AO 可能仍然存在较大潜力。

Trending Cryptos

Related Reads

Standard Chartered Bank’s 50-Fold Fantasy: Predicting AAVE to Reach $3,500

Standard Chartered Bank has issued an optimistic research report predicting that the AAVE token could surge 50-fold to $3,500 by 2030. This forecast is based on the projection that the total value locked (TVL) in DeFi will grow 37x to approximately $2.7 trillion, driven by stablecoin expansion and the tokenization of real-world assets (RWA). The bank's model links Aave's potential valuation directly to its protocol revenue, which is primarily driven by net interest margins. The report highlights Aave's current dominant position, noting it captures over 80% of the net earnings ("protocol retained earnings") in the lending sector while holding only about half of its TVL. It also points to the recent launch of the Aave V4 architecture and a healthy revenue stream of $142 million in 2025 as positive fundamentals. Grayscale's separate analysis, applying traditional valuation metrics like DCF, concluded AAVE is currently undervalued. However, the article notes significant challenges. Aave's peer-to-pool lending model suffers from inherent capital inefficiency, with an estimated $52 million annual "deadweight loss" due to idle funds needed for liquidity buffers. This structural flaw was exposed during the April KelpDAO exploit, which locked a WETH pool at 100% utilization for days. Emerging protocols like Morpho, with more efficient point-to-point models, are cited as growing competitive threats. In summary, while institutional forecasts paint a macro picture of massive growth fueled by RWA adoption, Aave's path forward hinges on addressing its core structural limitations and competitive pressures within the evolving DeFi lending landscape.

链捕手14m ago

Standard Chartered Bank’s 50-Fold Fantasy: Predicting AAVE to Reach $3,500

链捕手14m ago

Tidal Investment: We Remain Bullish on the AI Industry Chain, But the Reasons Have Changed

Tidal Investment remains optimistic about the AI industry chain, but the rationale has shifted. The market narrative has changed. While recent large-scale IPOs (e.g., SpaceX) and major fundraising plans by tech giants like Alphabet and Meta have caused some nervousness, this isn't a sign of an AI peak. The focus has moved from the initial question of AI's viability to the sustainability of massive investment cycles. The key players—primarily the major cloud providers—are not slowing down; their capital expenditure (Capex) guidance for 2026 has been increased across the board (e.g., Alphabet to $180B, Amazon to $200B). This investment cycle is proving resilient and difficult to stop. Unlike traditional hardware cycles, current AI Capex is distributed across multiple physical layers—computing, memory, networking, and critically, power infrastructure. Bottlenecks are shifting from chips to elements like electricity, transformers, and cooling systems, which have much longer lead times and cannot be easily pre-built like fiber optics during the dot-com bubble. Supply chain data (e.g., Eaton's 240% YoY data center orders) confirms this broad-based, project-driven expansion. Market concerns are acknowledged but viewed differently. First, while Capex growth currently outpaces revenue growth, raising ROI questions, this mirrors the early scaling phase of cloud computing itself. A change in view would require concrete signals like downward Capex revisions or missed AI product targets, which haven't materialized by mid-2026. Second, comparisons to the 2000 dot-com bust are flawed. That crash was driven by a massive, parallel oversupply of cheap capacity (fiber). The current cycle faces *supply constraints* in critical, capital-intensive physical infrastructure that cannot be overbuilt as easily. In conclusion, the wave of fundraising reflects the next, more complex act of the AI story. Physical bottlenecks and sustained high Capex plans suggest this is not the finale but an ongoing, capital-intensive build-out phase. The script has changed, but the play is far from over.

marsbit1h ago

Tidal Investment: We Remain Bullish on the AI Industry Chain, But the Reasons Have Changed

marsbit1h ago

Tidal Investment: We Remain Bullish on the AI Industry Chain, But for Different Reasons Now

Tidal Investments remains optimistic about the AI industry chain, but the rationale has shifted. The market is concerned about massive concurrent fundraising by tech giants like SpaceX, OpenAI, Alphabet, and Meta, fearing an AI peak. However, the authors argue this signals the next act of AI development, not its end. Capital expenditure (Capex) from major cloud providers (Alphabet, Amazon, Meta, Microsoft, Oracle) continues to surge aggressively into 2026. This investment cycle is more resilient than past hardware cycles due to its scale and complexity. Bottlenecks have shifted from chips to critical physical infrastructure like power grids, transformers, cooling, and data center construction—areas with long lead times and limited capacity for rapid expansion. Supply chain data (e.g., Eaton's orders) confirms substantial, tangible progress. Key market concerns are addressed: 1. **ROI vs. Capex Growth**: While Capex growth outpaces revenue, the authors note cloud giants have historically overcome similar phases through scale. The cycle will only be in danger if Capex guidance is cut, orders are canceled, or AI product demand falters—none of which are currently observed. 2. **Comparison to the 2000 Dot-com Bubble**: Unlike the telecom bubble, where cheap, oversupplied fiber crashed prices, AI infrastructure (especially power) is constrained, customized, and subject to lengthy approvals, making a similar supply glut and crash unlikely. In conclusion, the wave of fundraising reflects the immense, ongoing capital needs for AI's next phase, constrained by slow-moving physical bottlenecks. The AI cycle is not over; the script has simply changed.

链捕手1h ago

Tidal Investment: We Remain Bullish on the AI Industry Chain, But for Different Reasons Now

链捕手1h ago

Grayscale: These 15 Profitable Crypto Protocols Are Severely Undervalued

Grayscale Research identifies 15 top-revenue crypto protocols trading at significant valuation discounts, with many at single-digit or even 1x revenue multiples. Protocols like Pump.fun, PancakeSwap, and Meteora have market capitalizations roughly equal to their annual revenue. The report argues these financially-focused protocols (DEXs, lending, staking) are fundamentally undervalued and could benefit from the potential passage of the CLARITY Act, expected as soon as next month. This legislation aims to clarify digital asset regulation, potentially reducing institutional barriers and driving on-chain activity. The analysis breaks down the protocols into three groups: the "1x Club" (market cap ≈ revenue), mid-tier protocols with 3-9x multiples (e.g., Aave, Lido, Jupiter), and high-multiple protocols like Hyperliquid (15x) and Uniswap (37x), where valuation reflects future potential rather than current cash flows. Grayscale applies a traditional DCF model to Aave, suggesting a one-year price target of ~$175, representing ~130% upside from current levels. The report notes a risk-off macro environment since the Iran conflict has further compressed valuations, creating a potential entry window. The conclusion highlights that while the valuation data presents an intriguing opportunity, the investment thesis is contingent on the CLARITY Act's passage and subsequent institutional capital flows. Investors are cautioned to consider Grayscale's inherent conflict of interest as a crypto asset manager with products tied to these assets.

marsbit1h ago

Grayscale: These 15 Profitable Crypto Protocols Are Severely Undervalued

marsbit1h ago

Trading

Spot
Futures

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of AI (AI) are presented below.

活动图片