Circle Debuts Arc Chain to Settle Autonomous Machine Economy

Circle推出Arc链以结算自主机器经济 Jeremy Allaire认为，由于自主AI智能体很快将主导商业活动，金融网络必须从投机性代币转向受监管的、以稳定币结算的操作系统。 Jeremy Allaire与Chris Dixon审视了数字货币的发展轨迹，回顾了Circle从早期比特币探索到推出Arc的历程；Arc是一条专为机构结算和自主机器智能体打造的全新Layer 1区块链。他们认为，加密货币正在经历类似于前iPhone移动时代的15年孕育周期，正从投机性交易走向非拟物化、AI驱动的金融原语以及算法信用市场。为支持这一规模，Arc将开源、无许可的开发与注重问责的验证节点集以及以稳定币计价的gas费相结合，在密码学纯粹主义与严格的监管要求之间找到了务实的平衡点。最终，该讨论将公共区块链定性为抗衡科技垄断的关键去中心化制衡力量，并将链上轨道定位为呈指数级增长的智能体经济的结算层。 Circle推出了其专为特定目的构建的Layer 1区块链Arc，旨在满足现有网络未能解决的审慎监管和央行市场基础设施标准。 (6:55) Arc瞄准新兴的机器对机器经济，预计自主AI智能体（而非人类终端用户）将成为链上的主导交易主体。 (10:00) 该网络采用了一种折中的混合模式：构建依然保持开源和无许可，但将验证者运营限制在分布于全球不同司法管辖区、受法律约束并经过审查的机构实体。 (40:00) Circle的跨链传输协议（CCTP）目前每季度结算约1500亿美元的铸造与赎回流动性，该协议正拓展为Arc上发行所有资产的通用传输层。 (15:00)

Circle CEO Jeremy Allaire and a16z crypto general partner Chris Dixon argue that blockchains are entering an institutional, machine-driven era where public ledgers will function as economic operating systems for autonomous software agents and mainstream financial markets rather than speculative trading venues. Arc and the Institutional "Middle Way" (6:55) Allaire explains that Circle developed its clean-slate Layer 1 blockchain, Arc, because existing networks failed to satisfy prudential regulatory mandates as global financial institutions moved from experimentation to production. While traditional blockchains depend on anonymous validator sets, Arc utilizes legally identifiable node operators bound by enforceable service-level agreements. This framework enables sub-second deterministic settlement finality without probabilistic rollback risk, providing the predictable assurances required by central banks and institutional asset managers. Circle models Arc’s governance after Google's open-source Android platform, creating a shared-ownership model for a broad coalition of stakeholders while anchoring compute and gas fees to stablecoins rather than volatile native tokens. To preserve geoeconomic neutrality, Arc launched with an initial cohort of 10 to 11 institutional validators across Asia, Europe, and the United States, designed to scale up to roughly 40 slots through decentralized governance elections. Expanding Moats Beyond Stablecoin Issuance (15:00) Addressing competition from Wall Street entrants and the deflation of software creation costs driven by generative AI, Allaire points to Circle's multi-layered network effects. In addition to issuing USDC and EURC, Circle facilitates approximately $150 billion per quarter in primary and secondary mint-and-redeem liquidity directly connected to international commercial banking and clearing rails, giving it capital efficiencies that competing tokens cannot easily replicate. Circle is also turning its Cross-Chain Transfer Protocol (CCTP)—which frequently routes up to 70% of cross-network stablecoin flows—into an open transport layer on Arc. Rather than carrying only USDC, CCTP allows third parties to distribute tokenized credit, equities, and alternative stablecoins across networks like Ethereum and Solana. This infrastructure is paired with the Circle Payments Network (CPN), an institutional messaging and clearing system connecting roughly 200 financial institutions globally to execute cross-border settlements. The Mobile Analogy and Native Financial Primitives (20:00) Allaire and Dixon draw an extended historical parallel to the mobile internet, noting that 15 to 20 years of experimentation across platforms like Palm OS, BlackBerry, Symbian, and Windows Mobile preceded the mainstream breakthrough of the iPhone and Android. They argue that crypto remains in a similar mid-stage transition, where developer mindshare is still small and user abstraction is only beginning through fintech applications and stablecoins. As enterprise adoption progresses, Dixon and Allaire expect on-chain finance to transition away from digital "brochureware"—applications that merely digitize legacy processes like cross-border wires. Instead, combining fully reserved digital currencies with artificial intelligence will unlock non-skeuomorphic products, including algorithmic, "AdWords-style" credit markets where long-tail human and machine participants underwrite, demand, and clear credit directly without traditional bond-packaging intermediaries. Economic Rails for Autonomous AI Agents (35:46) With enterprise AI adoption estimated at around two percent, Dixon and Allaire predict that autonomous software agents will become the primary transacting entities on distributed ledgers. Because cognitive labor constitutes economic work, blockchains will provide the coordination, settlement, and verification rails for machine-to-machine commerce, including automated dispute resolution systems built to arbitrate agreements when AI agents hire other agents or human contractors. Dixon further frames decentralized, open-source networks as an essential structural counterweight to the centralizing dynamics of massive corporate AI compute platforms. Early activity on Arc already reflects this shift toward hybrid experimentation, ranging from bio-computational bytecode projects like Flybrain to sandboxed deployment environments like Arc Studio designed for rapid AI generation. While Circle anticipates an exponential surge in transaction velocity akin to the commercial internet's expansion in the 1990s, the precise trajectory of an agent-driven machine economy remains unpredictable. Arc's long-term viability depends on whether a hybrid governance model can satisfy rigorous central-bank compliance standards without compromising the open, permissionless characteristics necessary to attract developers.

Circle 首席执行官 Jeremy Allaire 与 a16z crypto 的 Chris Dixon 认为，区块链正从投机性工具演变为承载万亿美元规模的互联网经济操作系统；面对人工智能推动的自主代理经济，链上底层需要融合确定性结算与合规监管。 十五年周期与经济操作系统演进 (20:00) Jeremy Allaire 将区块链当下的发展阶段与移动互联网早期进行了类比。在 iPhone 和 Android 推动移动生态普及之前，移动行业历经了 Palm OS、BlackBerry、Windows Mobile 及 Symbian 等系统长达 15 到 20 年的试错期，早期大量应用不仅体验粗糙，且资本投入未能产生持续商业回报。Allaire 认为，自比特币诞生至今的十余年里，加密行业同样处于该漫长周期的中段，目前大多数去中心化应用仍未准备好迎接主流消费者，全行业活跃开发者基数与传统平台相比依然较小，因此尚无任何一条公链建立起绝对的护城河。 Chris Dixon 回忆称，2007 年初代 iPhone 发布时频繁掉线，许多人需同时携带备用功能机，科技行业直到 2012 年 Facebook 启动移动转型才普遍确认智能手机的主导地位。目前企业级 AI 采用率估计仅约 2%，正如 1990 年代万维网早期将传统纸质手册照搬上网一样，区块链早期也只是在复刻传统金融逻辑。真正的拐点将发生在开发者利用原生可编程货币基础元，创造出传统银行体系无法实现的全新金融行为。 机构级合规与 Arc 的折中架构 (6:55) Circle 决定开发自主 Layer 1 区块链 Arc，源于其观察到的两大结构性拐点：一是主流金融机构开始部署非投机性的真实资金结算、国库管理与证券代币化；二是央行与金融审慎监管机构明确要求，若将核心资本市场接入公链，网络必须达到严格的金融市场基础设施标准。由于既有公链在治理机制与运行保障上难以满足央行级别的合规审计，Arc 借鉴了 Android 的开放生态模式，在保持开源、公开发布与无需许可部署的同时，对底层验证节点设立严格准入标准。 与以太坊或 Solana 的匿名验证节点不同，Arc 采用已知且受法律服务水平协议（SLA）约束的实体运行质押验证节点，从而实现无概率性回滚风险的亚秒级确定性结算。主网启动时设有 10 至 11 个机构验证节点，规划通过去中心化治理选举扩展至约 40 个席位，涵盖亚洲、欧洲及美国的做市商、资管机构与全球性银行。Allaire 将这种设计定义为“中间道路”：既非完全封闭的私有许可链，也摒弃了极端的技术意识形态，以换取主流金融机构和大型企业长期部署合约所需的确定性与地缘经济中立性。 跨链基建与对抗复制的核心壁垒 (15:00) 面对人工智能大幅降低软件编写门槛以及华尔街机构尝试自建稳定币联盟的竞争环境，Allaire 将 Circle 的防御壁垒划分为三大支柱：合规资产网络、深度流动性枢纽与互操作基础设施。除 USDC 外，Circle 产品线已扩展至数字欧元 EURC、代币化比特币以及代币化美债基金。在流动性层面，Circle 每季度在传统银行与链上清算系统间执行约 1,500 亿美元的铸造与赎回，这种资金吞吐效率是普通仿造代币难以复制的。 在底层管道方面，Circle 的跨链传输协议（CCTP）占据了部分跨链稳定币流动性高达 70% 的份额。Arc 将 CCTP 升级为开放传输层，使该网络上发行的代币化股票、信贷或其他第三方代币均可原生借道 CCTP 进行跨链流转。此外，连接近 200 家全球金融机构的 Circle Payments Network（CPN）正扮演类似区块链原生 Visa 或 Swift 的角色，通过稳定币协调跨境机构清算，巩固其基础设施网络效应。 机器经济时代的自主金融与中枢平台 (30:00) Allaire 与 Dixon 一致预测，未来链上主要参与主体将从点击钱包的人类用户转变为自主运行的 AI 代理软件，认知劳动在本质上即是经济工作。针对这一“代理人经济”，Arc 直接在底层计费与 Gas 机制上围绕稳定币展开设计，以消除原生加密资产价格波动对计算成本造成的干扰。在该架构下，传统打包评级后售出固定债券的静态金融模式，预计将转向由人与机器共同供需的动态算法信贷市场，其撮合机制可能类似于数字广告竞价模式（如 AdWords），从而激活海量长尾资产与信贷发行方。 在应用试验层面，Arc 上已出现基于 16MB 真实果蝇神经计算规则构建的代码实验 Flybrain，以及针对 AI 代理之间劳务分包的去中心化自动化仲裁系统。为了防范 AI 生成代码带来的安全隐患，Circle 提供了名为 Arc Studio 的沙盒工具以规避漏洞。Dixon 指出，尽管 AI 具有向科技巨头高度集中的趋势，但具备去中心化、可组合性与开源特性的公链可作为防止平台垄断的抗衡力量；Allaire 则举例称，其内部测试网上一款轻量级应用在未公开宣传的情况下，一周内便遭到自动化机器人每日约 15,000 次探测敏感凭证与私钥的扫描，表明机器经济的自动化交互已在真实网络中发生。 该讨论表明 Circle 正试图超越单一稳定币发行方的定位，通过 Arc 和 CCTP 构建兼顾传统金融合规与亚秒级结算的经济操作系统。然而，在现有区块链开发者群体基数有限、主流机构仍处早期尝试的背景下，这套旨在连接大型金融实体与自主 AI 代理的混合治理架构能否真正克服行业意识形态分歧并主导下一阶段生态，仍有待时间验证。

## Background 本期节目由 Andreessen Horowitz（a16z）于 2026 年 10 月 16 日在播客及网络平台发布。节目主要围绕 Circle 联合创始人兼 CEO Jeremy Allaire 与 a16z crypto 创始人兼创始合伙人 Chris Dixon 的深度对话展开，探讨了稳定币的未来演进、Circle 推出的全新 Arc 区块链网络，以及当可编程货币与日益强大的 AI 机器智能相遇时所带来的范式转变。 ## TL;DR 对话从 Circle 历经十余年从比特币早期探索、以太坊 USDC 爆发到多链扩张的技术路线切入，深入拆解了其推出面向机构合规与金融基础设施的 Arc 区块链的战略逻辑。Jeremy Allaire 详细阐述了 Circle 依托季度结算规模达 1,500 亿美元的跨链传输协议（CCTP）与深厚流动性所构筑的三大支柱护城河，并类比移动操作系统 15 年演进史，强调公链当前正处于迈向隐形“经济操作系统”的极早期阶段。Chris Dixon 与 Jeremy Allaire 一致认为，未来互联网的主要参与主体将由人类全面转向 AI 代理，智能合约的可组合性与去中心化网络将成为承载认知劳动、去中心化长尾信贷及抗衡科技巨头 AI 垄断的关键基座。Arc 采取了“验证节点身份已知受约、应用层完全开源免许可”的务实折中设计，旨在为主流金融、央行机构以及未来数以亿计的机器经济提供具备确定性结算终局的全球价值中枢。 ## Atmosphere & style 整场对话洋溢着硅谷资深技术老兵与顶尖风投领袖之间的默契与笃定，叙事风格宏大而严密，既有回望科技浪潮的历史纵深感，又不乏对工程实现与金融监管的务实洞察。两位嘉宾在探讨技术哲学与未来图景时充满前瞻性的兴奋，谈及多代理 AI 仲裁、链上生物神经模拟等前沿实验时碰撞出极强的极客思维火花。节目后半段穿插着关于 Bot 自动扫描测试网、全网潜伏抓取私钥的幽默调侃，让严肃的金融基础设施探讨在略带赛博朋克现实感的轻松欢笑中圆满落幕，听感既扎实硬核又极具启发性。 ## Bulletin - `05:00` Circle 正式上线自研区块链 Arc 主网，定位于支持央行与主流机构合规标准的“经济操作系统”，早期由来自欧美和亚洲的 10 至 11 家机构节点共同启动，设计上限约为 40 个验证节点。 - `15:00` Circle 首席执行官 Jeremy Allaire 披露，其跨链传输协议 CCTP 每季度原生铸造与赎回规模达 1500 亿美元，部分季度占据全球跨链流动的 70%，如今 Arc 上的任意资产均可作为 CCTP 有效载荷全链分发。 - `30:00` 针对 AI 时代的商业重构，对话指出可编程货币正推动信贷机制从传统中心化债券打包转向“Google AdWords 式”的长尾实时竞价撮合，使人类与机器都能作为长尾发行人直接发行金融资产。 - `37:25` Chris Dixon 与 Jeremy Allaire 强调，区块链是对抗 AI 基础模型中心化垄断的关键解药，而未来的金融基础设施必须承载由自主机器代理主导的认知劳动清算与毫秒级经济交换。 ## Key takeaways - `00:46` Circle 创办之初即聚焦“可编程货币”愿景，但早期因比特币社区功能封闭转向以太坊推出 USDC，随后因单链拥堵逐步演进为多链分发与跨链互操作架构。 - `05:00` 随着机构资金与监管机构严肃入场，现有公链无法原生满足金融市场基础设施的严苛标准，促使 Circle 推出兼顾开放部署与合规审查的 Arc 区块链。 - `10:00` 未来经济网络的核心交互主体将从人类转变为 AI 代理，全球级经济操作系统必须具备地缘政治与地缘经济的多元中立性，以保障亚秒级结算终局。 - `15:00` Circle 的核心护城河建立在三大支柱之上：以 USDC 为核心的多元资产矩阵、承载超大规模流动性的跨链基础设施（CCTP 及 Arc），以及连接近 200 家金融机构的 Circle 支付网络（CPN）。 - `20:00` 区块链当前正处于类似移动时代爆发前 15 年的漫长摸索期，目前尚无任何平台建立绝对护城河，随着 AI 降低编程门槛与无感应用涌现，真正的质变突破正在临近。 - `25:00` 新技术落地普遍存在认知滞后，主流商业采纳会经历“把传统业务搬到链上跑得更快”的映射阶段，最终迈向基于去中心化原语的原生创新爆发期。 - `30:00` AI 与全额储备数字美元的结合，彻底打破了传统金融依赖中心化打包的信贷分销逻辑，开创了机器主导的长尾信贷自执行机制。 - `32:37` Arc 上的“Flybrain”果蝇大脑模拟项目与 Arc Studio 氛围编程工具表明，确定性区块链正与计算生物学和 AI 碰撞出高容错、全新维度的沙盒实验。 - `35:10` 面对 AI 代理跨机器协同完成知识产权与代码开发的现状，链上自动化仲裁协议展示了多代理裁决和资金解锁的可行性，释放了认知劳动的组织潜力。 - `37:25` 顶尖 AI 模型高度依赖算力、数据与资本，天然带来中心化倾向，而开源无许可的区块链网络正是防止巨头封锁资源、维持开放计算环境的关键平衡制度。 - `40:00` Arc 选择了一条“中庸之道”的设计路线：应用层对全球开发者完全开源和无许可，而底层的验证节点则必须由受法律契约约束的已知合规实体运营。 - `46:48` 在指数级加速的机器时代，网络大部分流量与交互已被自动化 Bot 占据，为这些自主机器赋予合规、可编程的链上经济结算轨道是技术发展的必然趋势。 ## Detailed walkthrough ### `00:00` 互联网经济操作系统与 Circle 的链上护城河 节目的开篇探讨了互联网规模的金融基础设施应具备怎样的特征。Jeremy Allaire 提出，构建一个运行在互联网上的经济操作系统，其内涵早已远超单纯的支付结算，它需要承载各类复杂的金融合同、经济协议乃至治理决策。如果这种操作系统要真正实现全球互联网级的规模，就必须在地理经济（geo-economic）和地缘政治（geopolitical）上具备充分的多元性。 面对当前一边是 AI 极大地降低了软件代码编写门槛、另一边是传统金融巨头纷纷入场推出自营产品的竞争态势，Jeremy Allaire 重点谈及了 Circle 的核心护城河——跨链网络与流动性。 “The most widely used interoperability in the world right now on all blockchain networks is Circle's cross-chain transfer protocol, CCTP. We mint and redeem $150 billion a quarter of this and that liquidity that exists in on a primary basis, on a secondary basis, in on-chain markets and on-chain protocols. That's very powerful.” —— Jeremy Allaire 随着金融科技的交汇融合，全球金融系统正在经历一次底层结构的彻底变轨，五到十年后现有的全球金融版图将被深度重塑。 - **季度结算规模**：Circle 的跨链传输协议（CCTP）每个季度铸造与赎回的规模高达 1,500 亿美元。 - **流动性深度**：该协议不仅打通了链上一级发行与二级交易市场，还与各大去中心化协议形成深度嵌套，构成极高的网络效应壁垒。 --- ### `00:46` 从比特币协议到 USDC：可编程货币的演进历程 主持人 Robert Hackett 随后正式引入话题。Circle 最近刚将其自研的区块链网络 Arc 推上主网，而作为曾在《财富》（Fortune）杂志长期追踪 Circle 的记者，Robert 回顾了 Circle 漫长的探索周期：从最早的面向消费者的比特币业务，到类似 Venmo 的 Circle Pay，再到场外交易与收购交易所，最终凭借稳定币 USDC 取得巨大成功。针对“为什么这个世界还需要另一条区块链”的疑问，Jeremy 回溯了公司在 2012 年底至 2013 年成立之初的技术愿景。 当时吸引 Jeremy 进入该领域的核心概念是“可编程货币”（programmable money），即将分布式账本视为能够运行应用代码的计算机，旨在公共环境下提供确定性数据、确定性交易和确定性计算的保障。尽管团队当时就想发行数字美元（digital dollar）——Jeremy 特意强调团队当时只使用“数字货币”（digital currency）一词，从不使用“加密货币”（cryptocurrency）——但在比特币网络上进行高层协议创新的尝试很快遭遇挫折。 “Well, that didn't work out very well because it became a very kind of closed community in the sense of tightly controlled.” —— Jeremy Allaire 比特币开源社区在功能扩展上逐渐狭隘，并陷入激烈的意识形态派系争论，限制了其图灵完备性和编程语言能力的扩展。而曾在 Web 1.0、Web 2.0、互联网流媒体和虚拟机基础设施领域深耕的 Jeremy，随后在 2016 年左右将目光转向了以太坊。尽管从技术纯粹主义者的眼光来看，早期的以太坊与 Solidity 甚至显得有些像“口香糖和打包铁丝拼凑起来的产物”（gum and bailing wire），极不具备企业级成熟度，但它验证了虚拟机的可行性。这促使 Circle 放弃在比特币网络上转移美元的方案，转而在以太坊上推出了 USDC。 ```mermaid flowchart LR accTitle: Circle底层技术路线演进 accDescr: 展示从比特币网络探索到以太坊发行USDC，再到多链扩张并最终推出Arc区块链的过程。 subgraph S1["早期探索 (2013-2016)"] BTC["比特币网络"] -->|"意识形态与功能受限"| PIVOT["转向智能合约"] end subgraph S2["单链爆发 (2016-2020)"] ETH["以太坊 (Ethereum)"] -->|"上线运行"| USDC["USDC 诞生"] end subgraph S3["多链与拓展 (2020-至今)"] USDC -->|"拥堵/高Gas费推动"| MULTI["多链拓展<br/>(Solana / L1s / L2s)"] MULTI -->|"金融基础设施合规需求"| ARC["自研区块链 Arc"] end ``` --- ### `05:00` 基础设施瓶颈、金融监管与 Arc 区块链的诞生 进入 2020 至 2021 年周期后，以太坊严重的性能瓶颈、网络拥堵与高昂的 Gas 费用暴露无遗。Circle 采取了类似流媒体平台的内容分发策略，即“让数字美元在任何设备和网络上都能无缝运行”——USDC 就像 Spotify 或 Netflix 一样，不论底层是 Solana、新兴公链 L1 还是以太坊 L2，用户随时触手可及，并配套推出跨链协议保证流转安全。 大约在两年半前，行业出现了关键拐点：传统主流金融机构开始严肃考虑利用区块链承载真正的法定货币与证券资产，非投机性用例快速崛起。与此同时，央行与资本市场监管机构开始介入，明确要求任何承载核心金融体系的区块链底层都必须符合严苛的基础设施标准。市场上现有的公链无一能够原生满足这种主流合规标准，Circle 由此决定从零开始构建 Arc。 “And what we really found was that there were not blockchain networks and computers that were purpose-built for that what we saw as like a new era that was coming, this mainstream era.” —— Jeremy Allaire Jeremy 强调，Arc 的组织与治理模式借鉴了开源社区、代币激励网络以及类似 Google 对 Android 的平台生态扶持经验，确保所有权分散在广泛的利益相关者和最终用户手中，同时在技术上达标央行对于大额真实资金和投资合同清算的审慎要求。在设计 Arc 的过程中，团队进一步预见到，这一新一代经济操作系统的主要交互主体将不再仅仅是人类，而是以指数级速度演进的自主软件与机器智能。 - **多链扩张策略**：2020 年率先与 Solana 展开深度合作，随后广泛集成其他主流 L1 和 L2 网络，保持中立的资产覆盖面。 - **机构关注转移**：稳定币（尤其是 USDC）成功验证了低成本、高效率的资金转移能力，驱动链上生态从纯投机场景向实质金融场景演变。 - **监管合规门槛**：全球央行及证券监管机构对金融市场基础设施（Financial Market Infrastructures）设有强制性合规标准，这是既有公链此前普遍欠缺的维度。 ### `10:00` 代理经济的崛起与为什么还需要一条新公链 Jeremy Allaire 回顾了早期构想在互联网上运行去中心化系统时的设想，坦言当时虽然完全预料不到生成式人工智能（Generative AI）的爆发，但后来局势变得极其明朗：未来软件的主要用户、主要部署者以及在链上运行的核心实体，极大概率会是 AI。“智能体经济”（agentic economy）的到来，成了 Circle 架构设计中最核心的前提与背景。 面对市场上关于“为什么还需要又一条新区块链”的疑问，Chris Dixon 解释了他支持 Circle 推出 Arc 的原因。Chris Dixon 指出，投资中最关键的始终是团队，而他长期了解 Jeremy Allaire 并对其执行力深感佩服。区块链不仅是拥有社会性社区的载体，更是一种新型的全球金融计算机，需要深入理解操作系统和编程语言设计的团队来打造。此外，Chris Dixon 强调 Jeremy Allaire 在应对全球各司法辖区复杂的监管与国际化问题上拥有极深远且务实的眼光；货币天然具备最强烈的网络效应，USDC 与美元在全球广受欢迎，Arc 有机会成为这层全球货币基础设施（Global Money Layer）。 针对国际化与监管落地，Jeremy Allaire 详细阐述了 Arc 的验证节点准入门槛及其治理逻辑。Arc 要求节点运营方和质押验证者必须符合严苛的营运标准，包括信息安全合规、运营规范以及服务等级协议（SLA）。正是通过确保节点运营者满足这一高门槛，网络才能在处理真实资金时实现亚秒级的确定性结算终局（deterministic settlement finality）。 Jeremy Allaire 进一步强调，一个支持支付、金融合约乃至治理决策的全球经济操作系统，不能只是一套由美国主导的系统，而必须具备可信中立性以及地缘经济与地缘政治的多样性： > “If you want that to be something that's Internet scale, it has to have geo-economic and geopolitical diversity. It actually needs to be something that is credibly neutral vis-a-vis the stakeholders that run the infrastructure and that ultimately govern the infrastructure.” —— Jeremy Allaire - **多区域机构参与**：Arc 的早期验证节点网络已引入资本市场、资产管理机构、全球支付公司及大型跨国银行，覆盖亚洲、欧洲和美国市场。 - **主网上线与规模扩展**：主网刚上线时有 10 到 11 家机构共同启动，网络设计上限约 40 个验证节点席位，未来将通过质押验证者投票选举的方式逐步向全球基础设施伙伴开放扩展。 --- ### `15:00` Circle 的护城河：三大支柱与跨链互操作底座 主持人提出疑问：在 AI 极大降低从零编写软件的门槛、同时花旗（Citi）、高盛（Goldman Sachs）等传统金融巨头纷纷结盟推出新型稳定币（如各大银行联合项目）的背景下，Circle 和 USDC 的防守壁垒（护城河）究竟何在。 Jeremy Allaire 将 Circle 的防御体系归纳为三大支柱，其中最核心的是“稳定币网络”（Stablecoin Network）。除了旗舰产品 USDC 之外，Circle 的资产矩阵还涵盖数字欧元（EURC）、代币化比特币产品，以及目前市场上规模极大的代币化国债与货币市场基金等。稳定币网络本质上是平台型与网络效应业务：连接的应用程序越多，网络的效用就越高，进而形成吸引下一位开发者的飞轮效应。 更为关键的是流动性网络效应与基础设施壁垒。Jeremy Allaire 透露，Circle 每个季度在原生层面上铸造与赎回的规模高达 1500 亿美元，与全球银行体系实现了深度接轨。链上初级与二级市场极深的流动性带来了极高的资本效率，构成了极强的转换壁垒。在基础设施层面，Circle 的跨链传输协议（CCTP）已成为全球使用最广的跨链协议，部分季度占据了跨链转移活动高达 70% 的份额。通过与 Arc 结合，Arc 上的任何代币化资产都能直接作为 CCTP 的有效载荷进行全链流动。 > “So in a sense like we built the highways to move value around all these networks and the only car that could drive on the highway was USDC until now. Now anyone that issues an asset... can ride those same highways and get that interoperability liquidity distribution that flows, that's really powerful.” —— Jeremy Allaire ```mermaid flowchart TD accTitle: Circle 三大支柱与价值流动架构 accDescr: 展示 Circle 的稳定币网络、底层跨链与合规基础设施以及 Circle Payments Network 如何服务机构并连接全链资产 subgraph Pillar1["支柱一：稳定币网络与资产矩阵"] USDC["USDC (核心美元稳定币)"] OtherAssets["EURC / 代币化国债 / 代币化 BTC"] Flywheel["应用接入飞轮效应\n(季度铸造赎回 1500 亿美元)"] USDC --- Flywheel OtherAssets --- Flywheel end subgraph Pillar2["支柱二：基础设施与互操作层"] ArcNet["Arc 区块链\n(合规节点验证)"] CCTP["CCTP 跨链传输协议\n(覆盖高达 70% 跨链流动)"] Payload["Arc 上资产直接作为\nCCTP 传输 Payload"] ArcNet --> Payload --> CCTP end subgraph Pillar3["支柱三：Circle 支付网络 (CPN)"] FIs["近 200 家跨国金融机构"] Coordination["区块链与稳定币作为\n清结算协调轨道"] FIs --> Coordination end Pillar1 -->|"提供全网深度流动性"| Pillar2 Pillar2 -->|"提供跨链流通高速公路"| Pillar3 ``` - **流动性规模**：Circle 季度铸造与赎回金额达 1500 亿美元，深度链接全球传统银行系统。 - **CCTP 协议开放**：以往 CCTP 高速公路主要承载 USDC，现在 Arc 上发行的代币化股票、信贷产品或新型稳定币均可直接借由 CCTP 穿透多链。 - **市场份额预期**：稳定币市场未来将由数千亿美元迈向数万亿美元，Jeremy Allaire 明确表示“不认为这是一个赢者通吃达 90% 的市场”（"We don't assume this is a 90% market"），未来必然会存在多张主流网络共存的格局。 - **Circle Payments Network (CPN)**：作为第三大支柱，CPN 面向金融机构（FI），形态类似于 Visa 或 SWIFT，以区块链和稳定币作为价值交换与对账机制，目前已有近 200 家金融机构加入该网络。 ### `20:00` 移动时代的15年长征与公链作为“经济操作系统” Jeremy Allaire 提出了一个深度的历史类比来审视当前区块链的发展阶段。他回顾道，人们今天谈到移动时代，总会下意识地直接联想到 iPhone 和 Android，但现实是移动时代的探索在此之前就已经经历了长达15到20年的漫长岁月。从开源开发者到微软、康柏（Compaq）这样的跨国巨头，曾向其中倾注了数十亿美元；市场上也曾涌现出 Palm Pilot、黑莓（BlackBerry）、Windows Phone、三星早期手机、诺基亚、Symbian 以及日本的 NTT 等层出不穷的形态。在那个漫长的周期里，业界同样经历过早期的兴奋，惊叹于“手机上居然能跑应用了”，但对主流普通用户而言，那些初代产品体验极其糟糕。海量资金砸出的 App 根本无法支撑起真正的商业模式，大量基础设施最终被弃用，沦为散落世界各地的电子垃圾。 直到硬件设备、软件操作系统、用户体验、开发者激励机制以及经济闭环等多重要素在长期的演变中产生交汇与共鸣，Apple 和 Google 才几乎在同一时期迎来了移动时代的质变突破。Jeremy Allaire 认为，区块链网络在底层心智模型上就是一种“经济操作系统”：它们沉淀资产、吸引开发者构建。当前的加密世界正处在这场15年演进长征的相似节点上。现在根本无法轻言哪条链是这个生态里的 iOS 还是 Android，因为绝大部分链上应用依然缺乏面向主流大众的可行性。真正的质变才刚露端倪——一些主打无感交互的消费级应用（如 Fomo）以及新一代 Neobank 让用户在完全感知不到加密底层的情况下享受服务，而 AI 的加入更是大幅加速了这一进程。 从数百万人为了交易投机或意识形态信念而坚守的核心圈层，跨越到数十亿人将区块链作为日常经济活动隐性基础设施的未来，整个行业仍然处于极其早期的起跑阶段。Jeremy Allaire 强调，当前没有任何平台真正建立起了不可撼动的护城河，即使放眼整个行业，活跃链上开发者的数量若与传统开发平台相比依然少得可怜。与此相应的是，软件的开发迭代速度正在被 AI 极速拔高，Circle 内部的软件交付效率也在成倍攀升。在这一开放竞争的大格局下，Circle 推出了 Arc 区块链。Jeremy 将 Arc 比作 Google 当初打造的 Android：Circle 作为发起方和主要构建者，注入自身的机构信誉、庞大的稳定币网络，并将计算层与 Gas 费机制深度绑定稳定币特性；但其治理和经济所有权则是彻底分布式的，Arc 的代币白皮书已经公开，众多大型机构均作为关键利益相关者参与其中。 “And so my view is like this is a wide open space. No one has won significant developer mindshare. The number of active developers on any of these networks is incredibly small when you think about developer activations in traditional platforms. And so it's super early. And so no one has a moat is, I guess, my point.” —— Jeremy Allaire ```mermaid flowchart TD accTitle: 移动时代与区块链作为经济操作系统的演进对照 accDescr: 对比移动操作系统从长期混乱探索到整合突破的历史，展示区块链当前所处的早期阶段及未来与AI的融合路径。 subgraph MobileHistory["移动时代演进（15-20年）"] M1["早期探索与混乱尝试<br/>(Palm, BlackBerry, Windows Phone, Symbian)"] --> M2["基础设施沉淀与体验摸索<br/>(早期App缺乏商业模式，大量硬件沦为垃圾)"] M2 --> M3["要素交汇质变<br/>(硬件/操作系统/经济激励成熟，iOS与Android爆发)"] end subgraph BlockchainEvolution["区块链经济操作系统演进"] B1["早期极客与投机驱动<br/>(数亿核心参与者，投机/意识形态驱动)"] --> B2["经济操作系统竞争期 (当前阶段)<br/>(开发者基数仍小，无绝对护城河，Arc等新型L1/L2展开布局)"] B2 --> B3["隐形基础设施与主流采用<br/>(数十亿用户无感使用，AI代理驱动认知劳动与链上协作)"] end M3 -.->|"历史路径映射"| B3 ``` - **移动生态先驱列表**：在 iPhone 和 Android 统一市场前，包括 Palm Pilot、黑莓、Windows Phone、三星早期系统、Symbian、诺基亚、NTT 在内的众多平台均经历了长期的试错与退场。 - **Arc 的定位设计**：效仿 Android 模式，由 Circle 担当核心发起者并提供以稳定币为中心的结算底座，但通过公开代币白皮书与多元机构利益相关方实现去中心化治理架构。 --- ### `25:00` 认知劳动链上化、技术落地的滞后性与主流机构的务实转向 随着讨论深入，Jeremy Allaire 指出未来三到五年行业将迎来根本性重塑。最核心的驱动力在于未来的主要参与主体将由人类转变为 AI——无论是人类指导下的 AI 编写代码，还是 AI 自主生成代码并直接在链上开展经济协作。由于认知劳动（Cognitive Labor）本身就是一种高价值的经济活动，这类计算工作量、协作流与价值结算，最自然的归宿就是直接在去中心化网络上承载与清算。 Chris Dixon 对此表示认同，并补充了技术爆发期常见的“认知滞后”现象。他回忆道，即使是 2007 年初代 iPhone 发布时，其体验实际上也相当简陋，掉话严重到他必须随身多带一部功能机或黑莓备用。彼时大众普遍将手机视为电脑之外的“边缘辅助配件”，硅谷也花了整整 5 年时间——直到 2012 年 Facebook 顶着铺天盖地的质疑发起移动端转型（Mobile Pivot），整个科技界才彻底惊醒。这种怀疑与挫折在每一轮技术范式转移早期都是常态。 面对当前加密圈部分意识形态派参与者的低迷情绪，Chris 认为客观数据展现的反而是前所未有的繁荣与落地动能。尤其在支付与金融领域，伴随监管环境的逐步明朗，稳定币技术正在被主流世界迅速接受。包括像 Stripe 这样以极度务实、唯工程与产品效果是瞻的技术公司，已经全面拥抱底层区块链网络进行支付结算。虽然目前企业级 AI 采纳率可能仅有 2% 左右，AI Agent 在消费端的渗透也刚刚起步，但它们必然会打破既有的 Web 互联网商业模式。未来 10 到 20 年数字化经济版图的重构，恰好与 Arc 这样高性能区块链的兴起形成完美契合。 这种演进逻辑同样复刻了早期万维网的发展历程：在 1990 年代，企业所能做的只是把纸质黄页和商品名录照搬上网（即所谓的宣传品网页 Brochureware）；直到 2000 年代，真正的社交网络、动态互动和原生数字创意才大举涌现。Chris Dixon 预计，未来几年行业会先经历将现有金融事务“搬到链上跑得更快、更全球化、更具可编程性”的阶段，随后更具颠覆性的链上原生创意才会全面爆发。Jeremy 对此表示高度赞同，指出底层的意义在于提供资金流和金融合约的原语（Primitives），比如信贷体系与可编程抵押借贷的重构，至于未来最伟大的应用形态究竟是什么，唯有留给拥有全新生产力工具的开发者们去探索与创造。 “It was still – people were still anchored on the framing that the phone is the side gadget, not the primary gadget. And it took a long time for people to kind of realize that.” —— Chris Dixon - **移动转型的标志性事件**：2007 年 iPhone 问世，但直到 2012 年 Facebook 艰难完成全公司层面的移动优先战略转型，市场才真正确立移动端作为第一终端的地位。 - **AI 普及的现状参照**：当前企业级 AI 采纳率大约仅在 2% 左右，自主性代理（Agents）在消费端普及度依旧极低，整体仍处于重构商业模式的前夜。 - **技术落地周期规律**： 1. **搬迁与映射阶段**：将既有成熟业务迁移至新载体（如 90 年代的黄页上云、当前的跨境稳定币支付与现成资产抵押）； 2. **原生创意爆发阶段**：诞生基于全新底层原语（金融智能合约、AI 认知劳动清算）的内生型颠覆应用。 ### `30:00` 机器与长尾信贷：从打包债券到“Google AdWords式”信贷市场 全额储备的数字美元与人工智能的交汇，正在催生截然不同的信用机制与市场范式。传统的信贷模式依赖于中心化机构将债务打包成债券分销给投资者，而当人和机器同时既是资金的供给方也是需求方时，信贷市场的底层结构将彻底重塑。这种去中心化信贷撮合机制不仅重构了单位经济模型与分发渠道，更首次将科技界熟知的“长尾理论”真正引入金融市场。 当前的真实世界资产（RWA）代币化和股票代币化，仅仅是长尾“市场参与者”层面的早期试验；而真正的范式转变在于迎来股票、信贷及代币等金融工具的“长尾发行人”。软件基础设施、API以及区块链计算机天然具备确定性（deterministic），而 AI 正在以惊人的速度理解并实现这些底层逻辑，使得软件构建的门槛被彻底抹平。正如 Jeremy Allaire 预测的那样：“AI is able to learn and implement at an extraordinary pace. And so what that means is that the creative output that we're going to see in terms of what software people create is just going to... blossom like crazy.” 到 2027 年左右，新一代区块链应用将全面打破拟物化（skeuomorphism）的思维定势。这些应用不再是把传统线下金融合同机械地搬到链上，而是将展现出从未设想过的交互模式。在 Arc 等新兴链上生态中，用户甚至仅凭 USDC 就能直接进入复杂的动态模拟和游戏经济系统，无需经历以往晦涩繁琐的操作门槛。 ```mermaid flowchart TD accTitle: AI与可编程货币重塑信贷市场 accDescr: 展示从传统中心化债券打包模式演变为机器驱动的长尾信贷市场的过程 subgraph TradFin["传统金融模式"] A["借款人 / 资产发起方"] --> B["中心化打包机构"] B --> C["分级债券产品"] C --> D["机构投资者"] end subgraph NewCrypto["长尾可编程信贷模式"] E["长尾发行人<br>(人类与机器)"] --> F["去中心化信贷撮合协议<br>(类似 AdWords 实时竞价)"] G["长尾资本提供方<br>(全额储备数字美元)"] --> F F --> H["自执行微型信贷合约"] end ``` --- ### `32:37` Arc 上的果蝇大脑与“氛围编程”的安全软垫房 在 Arc 区块链上，最突破想象力的实验莫过于名为“Flybrain”的项目。匿名开发团队将果蝇大脑真实的生物神经学计算规则转换为约 16 MB 的字节码并部署在 Arc 链上，随后利用 AI 生成前端用户交互界面，让这些果蝇神经元构成的链上生物体在确定性环境中自由交互与演化。虽然其科学探索的真实意图与代币经济激励尚未完全明朗，但这生动揭示了“计算生物学/神经科学模拟 + 人工智能 + 区块链网络 + 经济机制”碰撞时激发的荒谬而迷人的创造力。 为了降低开发者直接编写智能合约的高门槛与安全隐患，Circle 推出了名为 Arc Studio 的“氛围编程”（vibe coding）工具。由于智能合约直接关涉资产安全，代码容错率极低，Arc Studio 扮演了类似“安全软垫房”（rubber room）的角色，允许开发者在严密的安全防护内快速试验、部署原生智能合约。这一工具极大地拉动了开发者的参与度，社区中涌现出的既有严谨的金融协议，也有嵌入经济激励机制的沙盒模拟游戏。面对链上蜂拥而至的异质实验，Jeremy Allaire 谈到 Flybrain 时感慨道：“It's indicative of like, you know, what AI combined with these computer networks combined with economics, people are starting to try really different things.” - **Flybrain 项目核心构成**：利用 16 MB 链上字节码映射果蝇大脑神经计算规则，由 AI 驱动界面，实现在区块链网络中的动态演化。 - **Arc Studio 的核心定位**：专为智能合约开发者设计的安全沙盒与 vibe coding 环境，在保障确定性执行安全的同时加速产品构建。 --- ### `35:10` 玩具假象、AI 劳工仲裁与指数级叠加效应 科技史上一再重现的定律是：“下一件大事一开始往往看起来像个玩具”（The next big thing starts out looking like a toy）。正如 Chris Dixon 在其著作《Read Write Own》中所阐释的，区块链的核心价值在于让全球参与者能够围绕去中心化生态建立统一的经济契约，而软件交互界面的革命正在把这种理论模型变为现实。 一个正在 Arc 上落地的颠覆性用例是针对“AI 劳动力”的自动化仲裁系统。随着人类雇佣 AI、AI 雇佣其他 AI 协同完成代码编写、设计或知识产权创作成为常态，如何判断任务完成度以及如何释放资金成为痛点。该协议通过构建从多代理 AI 仲裁员到最终人类陪审员的分层判定体系，实现了去中心化的劳工与资金交付裁决。Jeremy Allaire 对此深感震撼：“That kind of thing, like automated arbitration of work and economic work and creative work and intellectual property work. Like that was inconceivable a few years ago, now it's actually very conceivable and could be very scalable and result in some really powerful unlocks for the labor economy and work in general.” 此类突破之所以能够呈爆炸式增长，根源在于技术底层存在多重指数级动力曲线的叠加： - **AI 扩展定律（Scaling Laws）**：算力与模型性能的非线性提升。 - **区块链网络效应（Network Effects）**：用户与资本聚集产生的飞轮。 - **可组合性（Composability）**：智能合约如同乐高积木般自由组合调用，免去重复造轮子的成本。 --- ### `37:25` 区块链：抗衡 AI 中心化趋势的技术解药 当被问及在生成式 AI 爆发后其核心理念是否发生修正时，Chris Dixon 指出，AI 的高速演进完全符合扩展定律的预期，但他依然坚信区块链是对抗科技垄断的关键防线。最先进的 AI 基础模型对资本、尖端算力集群和海量私有数据的依赖极其沉重，这天然赋予了拥有最多资源的大型科技巨头无与伦比的竞争壁垒，形成了极度中心化的产业格局。 区块链在此充当了抗衡这种垄断趋势的重要平衡力量。通过为开放数据、去中心化协议及社区治理提供确定性的规则保障，区块链能有效防止网络资源被少数科技寡头筑墙闭环。正如 Chris Dixon 明确总结的那样：“Part of my argument was that that blockchains in some ways can be an antidote to what I believe is a centralizing tendency of AI.”区块链与 AI 的深层结合并非噱头，而是构建开放、免许可数字世界所不可或缺的制度性对冲。 ### `40:00` Arc 的“中间路线”：在去中心化与合规现实之间寻求平衡 讨论从政策与技术哲学的交汇点切入。Chris Dixon 指出，开源是技术史上最具民主化力量的浪潮（例如 Linux 的崛起），而区块链本质上就是一种“开放服务”，为全球开发者提供了低费率、无门槛的公平竞争环境。他认为在政策层面，监管者若能允许这种去中心化力量与 AI 可能带来的中心化趋势抗衡并共同繁荣，将具有重大意义。紧接着，主持人抛出核心问题：面对以太坊（Ethereum）、Solana 等完全开放的公链，以及当前部分应用走向许可制网络的趋势，Circle 打造的 Arc 区块链究竟处于什么位置？它该如何在中心化与去中心化之间权衡？ Jeremy Allaire 明确指出，Arc 在软件使用和部署层面上是完全开放、公共且无须许可的（public and permissionless），而且整体开源。任何人都可以连接到 Arc、在其上构建并部署应用，甚至审查其底层代码。然而，Arc 与传统无许可公链的根本差异在于验证者节点的运作方式：其验证基础设施的运行方不是匿名实体，而是经过明确遴选、身份已知的专业机构。这些节点运营方在法律和运营层面均受到严格的契约（covenants）约束。 ```mermaid flowchart TD accTitle: Arc 网络的双层混合架构 accDescr: 展示 Arc 在验证者层和应用构建层之间的分工：受法律与运营契约约束的已知验证者，与面向全球开放无许可的开发者生态。 subgraph Validators["验证基础设施层（已知与受托）"] V1["已知机构节点 A"] V2["已知机构节点 B"] V3["运营与法律契约 (Operational & Legal Covenants)"] V1 --- V3 V2 --- V3 end subgraph Developers["应用生态层（完全开放与无许可）"] D1["全球开发者 / 智能合约"] D2["企业与主流金融机构"] D3["AI 代理与机器网络"] end Validators -->|"保障合规、抗审查与业务确定性"| Developers ``` Allaire 解释说，加密原生领域存在一种原教旨主义观点，认为一切运营治理必须完全内化在网络协议逻辑与密码学证明中，运行节点的匿名性是抵御政府审查的关键底线。然而，Arc 的目标是让主流社会、大型跨国企业、商业银行以及各国央行都能放心地将核心业务、商业契约与治理逻辑迁移到链上。大企业和主流机构在将业务接入全球网络前，必须清楚谁在运行底层网络、节点是如何选举产生的、出现故障时有怎样的法律和运营保障。Arc 走的正是一条“中庸之道”（the happy middle）——它兼具分布式、开源和免许可的接入层，同时具备合规透明的机构级运营治理。正如 USDC 的成功逻辑一样：虽然发行方是中心化的，但正因其合规、具备 KYC 机制、深度融入传统金融体系且提供直观的美元界面，才真正使数字美元进入了全球主流金融视野。 - Jeremy Allaire 针对 Arc 的治理定位强调：“世界上的政府和央行——因为我切实关心他们，我希望主流社会运行在上面，我希望主流金融系统运行在上面，我希望大型上市公司将其合约与治理搬到互联网上。我希望他们这样做，而他们会需要围绕网络如何运营的契约。”（“The world's governments and central banks, because I actually care about them, I want mainstream society running on this, and I want the mainstream financial system running on this, and I want companies, big public companies, putting their contracts and their governance on the internet. I want them to do that, and they're going to want covenants around how it's operated.”） - Allaire 总结其混合设计哲学指出：“它不是完全去中心化的。它是开放的，它是开源的，它是开放且公共的。在构建和使用上是免许可的。我们正在尝试寻找——就像 USDC 一样——什么是中间的平衡点。”（“It is not fully decentralized. It is open. It's open sourced. It is open in public. It is permissionless to build on and use. We're trying to find, and this is like USDC, what's the happy middle?”） --- ### `43:55` 五年后的经济图景：机器经济时代的价值指数级爆发 当被问及五年后在 Arc、加密基础设施与自主代理群（agent swarms）的交汇下全球金融系统会发生何种变化时，Allaire 坦言：“此时此刻的诚实答案是，没有任何人能真正知道，因为在我看来，技术加速的步伐比人类历史上任何时刻都要快。”（“The honest answer is no one has any idea at this point because the pace in my view the pace of technology acceleration is faster than at any point in human history.”）如果相信 AI 的复利定律，未来展现出的能力将难以用现有认知去推演；一旦通用超级智能（ASI）出现，其协调与推理能力在定义上就已经超越了人类大脑的理解范畴——在超智面前，人类的智力甚至就像苍蝇大脑一样有限。 尽管不可测，但其演变趋势是确定的：在未来几年内，全球相当大一部分创造性工作、认知劳动以及实质性的经济价值交换，都将转由 AI 机器全权主导和运行。Allaire 引用了 1995 至 1998 年互联网早期的经典类比——当年如果有人根据梅特卡夫定律（Metcalfe's law）、网络效应和带宽增长断言，未来的全球通信量和公开发布的信息量将比当时暴增十亿倍，绝大多数人只会嘲笑他完全疯了。然而事实证明，指数级增长曲线确实兑现了这一切。同样，伴随金融机器与机器经济基础设施的完善，未来经济活动的周转速度、价值转移的频率以及全球可获取的产品与服务总量，都将呈现出跨越多个数量级的爆发式增长。 ```mermaid flowchart LR accTitle: 信息经济与机器经济指数级跃迁对比 accDescr: 对比 1995 年互联网信息通信爆发与当前 AI 驱动下金融经济周转速度指数级增长的历史演变。 subgraph InternetEra["互联网革命 (1995+)"] I1["带宽与网络效应"] --> I2["信息量与通信周转速度<br/>呈现 10 亿倍级增长"] end subgraph MachineEra["机器经济革命 (当前及未来 5 年)"] M1["AI 代理与金融区块链 (Arc)"] --> M2["经济活动与价值流转周转率<br/>呈现跨数量级爆发"] end InternetEra -.->|"历史规律映射"| MachineEra ``` - 在 Allaire 看来，由 AI 主导的经济系统将极大释放人类社会的总体产能，使全球广泛分布的产品与服务供给规模以指数级扩张。 - 整个经济操作系统的本质演变为：高吞吐、自动化执行的金融基础设施，专门承载数十亿智能体之间毫秒级、不间断的经济交换。 --- ### `46:48` 机器时代网络：幽灵流量、Vibe Coding 趣事与未来落幕 Chris Dixon 接续讨论并展开阐述，他将这一即将到来的新范式概括为“机器时代网络”（machine age networks）。他指出，互联网绝大多数参与者转变为机器已成定局，而赋予这些机器经济属性、让它们拥有自主处理价值与支付的能力，是整个技术演化的自然必然。现场嘉宾相视而笑地补充道，事实上人类当下已经跨过了临界点——当前互联网上的大部分活动早就已经由各类机器程序（Bot）占据，人类在网络流量中的占比正急剧萎缩。 为了形象地说明这一现实，Allaire 分享了一段幽默而真实的个人经历。他最近尝试用“Vibe Coding”（氛围式编码）在 Arc 测试网上顺手搭建了一个实验性应用，并未对外公布域名，仅仅在 Circle 内部与同事做了小范围分享。一周后他查看后台数据，惊人地发现这个几乎只在 Whois 数据库或极隐蔽网络记录中存在的站点，每天竟然持续涌入约 15,000 次 Bot 访问。这些潜伏在暗网与互联网底层的自动化脚本集群全天候扫描全网，拼命探测抓取环境变量文件（.env）、SSH 登录凭证以及任何可能暴露的加密资产私钥。 Allaire 笑着调侃，当时内部团队可能就一位同事真正试用了一下测试网应用，剩下的几万次交互全是来找漏洞、搜刮加密资产的自动化代理。这个略带惊悚却又荒诞的插曲，恰恰无可辩驳地证明了机器大军早已遍布整个数字世界。正如对话尾声总结的那样：“它们已经在这里了，而我们需要给它们正经的事情去做。”对话在主持人对 Circle 多年探索历程与 Arc 愿景的致谢中圆满结束。 - Chris Dixon 总结指出，互联网参与主体正在发生不可逆的结构性转变：“机器将成为互联网绝大多数参与者，这是极度合理的，而且这些机器在执行动作时自然会伴随着经济属性。”（“And I think, to Jeremy's point, it's very plausible that you'll have the vast majority of Internet participants will be machines. And it seems natural that those machines would have economic components to what they do.”） - Allaire 的实验应用实测数据：该隐藏项目上线仅一周，即录得每日约 15,000 次机器人扫描请求，其搜刮目标主要集中在暴露的配置文件、SSH 凭据与加密钱包私钥。 ## Notable quotes > `37:25` Part of my argument was that that blockchains in some ways can be an antidote to what I believe is a centralizing tendency of AI. — Chris Dixon > `46:48` And I think, to Jeremy's point, it's very plausible that you'll have the vast majority of Internet participants will be machines. And it seems natural that those machines would have economic components to what they do. — Chris Dixon > `30:00` AI is able to learn and implement at an extraordinary pace. And so what that means is that the creative output that we're going to see in terms of what software people create is just going to... blossom like crazy. — Jeremy Allaire ## Mentioned Circle, Jeremy Allaire, Chris Dixon, Robert Hackett, Andreessen Horowitz (a16z), USDC, Arc, Arc Studio, CCTP (Cross-Chain Transfer Protocol), Circle Payments Network (CPN), EURC, Bitcoin, Ethereum, Solana, Solidity, Stripe, Visa, SWIFT, Citi, Goldman Sachs, Apple, Google, Android, iPhone, Facebook, Palm Pilot, BlackBerry, Windows Phone, Symbian, NTT, Fomo, Flybrain, Read Write Own ## Video description Circle co-founder and CEO Jeremy Allaire joins a16z crypto founder and founding partner Chris Dixon for a conversation about stablecoins, Circle’s Arc blockchain, and what happens when programmable money meets increasingly capable machines. [Source](https://mgln.ai/e/1344/afp-848985-injected.calisto.simplecastaudio.com/bf52be5a-9faf-45fb-9bc3-bb4270ec53d9/episodes/d132d72e-fec5-4e78-bf77-800207c63789/audio/128/default.mp3?aid=rss_feed&awCollectionId=bf52be5a-9faf-45fb-9bc3-bb4270ec53d9&awEpisodeId=d132d72e-fec5-4e78-bf77-800207c63789&feed=XPOpH7r4)

[37:25] Chris Dixon: Part of my argument was that that blockchains in some ways can be an antidote to what I believe is a centralizing tendency of AI. [46:48] Chris Dixon: And I think, to Jeremy's point, it's very plausible that you'll have the vast majority of Internet participants will be machines. And it seems natural that those machines would have economic components to what they do. [30:00] Jeremy Allaire: AI is able to learn and implement at an extraordinary pace. And so what that means is that the creative output that we're going to see in terms of what software people create is just going to... blossom like crazy.

[35:46] Chris Dixon: Part of my argument was that blockchains in some ways can be an antidote to what I believe is a centralizing tendency of AI. [23:30] Jeremy Allaire: And it's going to be AI intermediating and interacting and performing economic work. And in fact, cognitive labor is economic work.

[0:00] When we think about an economic operating system that runs on the internet, which goes way beyond payment, every kind of financial contract and economic contract and governance decisions, if you want that to be something that's internet scale, it has to have geo-economic and geopolitical diversity. How are you thinking about your moat? On the one side, you've got AI now makes it so much easier to code up software. And then on the other hand, you have legacy financial institutions coming on board with new sorts of products of their own. The most widely used interoperability in the world right now on all blockchain networks is Circle's cross-chain transfer protocol, CCTP. We mint and redeem $150 billion a quarter of this and that liquidity that exists in on a primary basis, on a secondary basis, in on-chain markets and on-chain protocols. That's very powerful. Five years from now, what has changed about the global financial system as a result of the convergence of these new technologies? We're going to go through a shift cords. Okay, so I'll kick things off. Why does the world need... another blockchain. What is missing from the infrastructure we already have? And What's made possible by launching a new one? I'm Robert Hackett. This is the A16Z Crypto Show. And joining me are Jeremy Allaire, co-founder and CEO of Circle, which just brought a chain of its own to Mainnet called ARK. And Chris Dixon, founder and managing partner of A16Z Crypto. Thanks, guys, for being on. Awesome. Thank you. So Jeremy, you and I, we go back to my days when I was a journalist at Fortune. I've covered Circle through many of its iterations, its journey from the Bitcoin consumer business to Circle Pay, the Venmo-like product. You had a trading business, bought an exchange. Ultimately, you guys struck gold with one of the world's most dominant stablecoins today, USDC. And now you're following that up with ARK. Tell me, what did you learn along the way that made you say... You need to build your own blockchain. So I guess maybe stepping back a couple of things, it actually goes back to like when we started Circle and. in 2013, late 2012, 2013, when we were getting started, I think the thing that drew me to this whole space was sort of this idea of programmable money and this idea of these networks. At the time, there's blockchain network, etc. The Bitcoin network, excuse me. But these networks actually becoming computers that you could run applications on. And the founding vision of the company was, you know what? I think that these networks are going to become programmable. like new distributed compute engines. I think that they're gonna, they're very purpose built for things where you wanna have like, assured data and transactions and compute. That's the problem that they can solve that hasn't been solved. But doing that in a public environment where you have essentially a public computer and that you could issue things on top of that. You could issue a digital dollar. At the time, we always use the phrase digital currency. We never said cryptocurrency, but you could have a dollar digital currency. And the idea was first you could build like a protocol for dollars on In order to do that, you needed these computer networks to actually let you build protocols, stuff that Chris and I go way back in terms of the Internet protocol. You need to be able to build these higher-level protocols on these computers, and then you need to be able to scale them and make them user-friendly and all this stuff. In a former life, you were working on video on the Internet, right? I worked on early web protocols. I worked on Web 2.0 protocols. I worked on video protocols, communication protocols. programming languages, virtual machines, like all the infrastructure stuff of the first and second generation of the Internet. And so to me, this was like, this is the next logical infrastructure layer of the Internet. And, you know, at the time, our view was, well, you know what? Bitcoin has early mover advantage. It's a good open source project. Let's get involved with that open source community and let's let's add the things that let this be more capable. Let's add more programming language capabilities. Let's let's let's add the things that let us realize our vision. Well, that didn't work out very well because it became a very kind of closed community in the sense of tightly controlled. The open source community, like the contributors, it was more narrow in terms of what it was trying to do. And there was a lot of ideological battles about that, et cetera. And we were on one side of that and there were others on the other side of that. And we don't need to repeat all that history. But I think Sean and I were always thinking, like, should we build, should we just build one of these computer networks ourselves? we had going on, as you just noted, like we've had a lot of stuff that's going on in our history. It never felt right. You know, interestingly, when the second generation of these computer networks came out, Ethereum, a lot of those ideas, programmability, a virtual machine, [5:00] issuing assets, like it actually worked. And, you know, I think you could be skeptical of it technically, because if you looked at it as a technical purist, you'd say, wow, this is like gum and bailing wire. I don't know what this solidity thing is like. This is like it's not it's not like an enterprise grade thing or however you want to characterize it. By 2016, like it was clear to us it was good enough. And so, as you recall and probably wrote about, like we abandoned the kind of layer of moving dollars over Bitcoin and instead said, let's Let's actually build a protocol for dollars that runs on Ethereum. And that's that led us to the development and launch of USDC. And so basically, like by 2017, 2018, we were like, OK, we can now do this. We can do that that that layer. And it wasn't again, wasn't clear to us like we should we should build our own computer network and our own operating system. And then you kind of fast forward to like 2020, 2021. It was really clear that the infrastructure model of these networks was not adequate. Like Ethereum itself was slow. It was it was expensive. The gas fees were really a problem until a lot of people started saying there an opportunity to like have a clean room and build like de novo networks And and we started to say well let let align with those because we need them to realize our vision of like scaled applications and programmability and velocity of money Like we need those. So we partnered with the Solana team in 2020. We partnered with other L1s that were emerging. And then we started to work with the L2s. And so then our philosophy really became like, well, let's just make sure that digital dollars that we make and the protocols that support them run everywhere. I kind of make it akin, like USDC is like Spotify or Netflix. Like you just buy a device, you know, it works. We want to make sure that if you have a blockchain, like you have these digital dollars and they work and you can integrate them. And then we built the interoperability layer to connect it all so that you could safely move them around. Through all of this, like I think our view is like these networks keep improving and, you know, solving for different things. But about two and a half years ago, I think we started to see a couple of things happen. I think one was major institutions were starting to say, hey, actually, we might use this technology for actual money, actual securities. There's real promise here. What changed from the early days that suddenly got institutions on board and seeing this as a viable path forward for them When I think about it something like USDC was really important Stablecoins was like a thing that people were like oh wow this actually that seems to solve a problem for moving money And so people started to look at like I could I could actually you know, I could actually start to really move money better with this. And that's when we really started to see growth in non speculative uses of this. Right. And that that that was important. And then I think the other change was, you know, regulators actually started to sort of look at this and say, you know, hey, if like the world's financial institutions are going to connect to blockchains, we expect certain things of those blockchains. We expect them to be kind of meet certain standards. And without boring you, there are standards for infrastructure in the financial system. And those standards are actually enforced. And they're enforced by central banks. They're enforced by the regulators that oversee our capital markets. And those are infrastructure standards. And so we started hearing that. And we started meeting with more and more of these institutions who are interested in stablecoins and the applications of stablecoins. And what we really found was that there were not blockchain networks and computers that were purpose for that what we saw as like a new era that was coming this mainstream era And so we took the decision to de novo build a new operating system and a new network with ARK And I think we designed it in a model, which I think it inherits a lot from what we've learned from how blockchain networks are incentivized, grown, and governed. It inherits a lot from other big platform ecosystem initiatives that we've seen historically, like Google with Android. It inherits a lot from open source. And we've tried to bring those together in a way where this is a network that is owned by a really broad set of stakeholders, ultimately the users on the network as well, but that the infrastructure can actually be held to standards that the world's central banks would say, yeah, you're good to go with putting real money and real investment contracts to perform on this thing. And then I think the last thing I'll say, and we can move off this, is we really saw as we were building this that the primary actors that were going to be using these new types of operating systems were increasingly going to be machines. We always thought that, like for a long time, we always thought that kind of autonomous software [10:00] running on the internet could do this stuff. No idea about generative AI, obviously, when we thought about that a long time ago. But then it became very clear that the primary users, the primary deployers of software and the primary entities that were going to be running on this was going to very likely be AI. And we started to think about a lot of the issues that go into that, and we continue to as well. It's like a very central issue. set of design features for what we're doing is the shift to this agentic economy as well. But anyway, that's like the backdrop. That's the long-winded backdrop. I want to drill down on some of those design decisions you made. But before we do, Chris, you've seen a lot of pitches for new blockchains. We've backed a number of them. What convinced you about? that there's room for yet another. Yeah, I think I mean, like a big part of this for us is as I've known Jeremy a long time and been incredibly impressed by what he's built. And so, you know, a lot of our most of the kind of the most important thing we make an investment is the team and also the team he's built around this. I share his sort of vision, I guess, of. of you know that this is a And I've written and talked about this extensively, but this is a new type of computer, a global kind of finance computer that has all sorts of platform possibilities, new applications and so forth. Blockchains are funny, right, because there's an aspect in which they're kind of a, you know, there's sort of a social social community around them, but also their computers. And you want people who who think deeply about, you know, sort of operating system and programming language design and so forth. I also think and maybe Jeremy can talk about this. He has a unusually disengaged. a deep view on the kind of regulatory side and on the international side. Money has the most obvious network effect. USDC is extremely popular. Dollars are extremely popular around the world. I think we have an opportunity to have a system like ARK be kind of a global money layer, which is a kind of critical idea. So that's one thing I would love to hear from you about, Jeremy, is just sort of the, I guess, the kind of international angle and particularly with all the complexity of the different governments and regulatory bodies. Obviously, one of the things arc is that in order to operate the infrastructure of the network and and to run ultimately what will be like staking validation nodes on the network you actually have to you have to meet a high bar of of operating standards that are operational and information security and compliance and other stuff and by knowing that the people doing that are meeting those standards and upholding them and have service level agreements and other things part of that actually allows the network to deliver what we call deterministic settlement finality in a fraction of a second. And that's actually very, very important with real money. But the other thing about that is that when we think about an economic operating system that runs on the internet, that can support The full vernacular of economic activity, which goes way beyond payments, like every kind of financial contract and economic contract and governance decisions, all this. If you want that to be something that's Internet scale, it has to have geo-economic and geopolitical diversity. It actually needs to be something that is credibly neutral vis-a-vis the stakeholders that run the infrastructure and that ultimately govern the infrastructure. of the kind of covenants that we've made. And it shows up in how this has stood up. There are quite a few of the largest global players in capital markets, in asset management, in payments, as well as many global banks that cover huge amounts of international markets, Asian markets, European markets, US markets. And so we've started, and this is really just the very beginning, by trying in the world, I understand that this is run by in this way. And I think we would like to see that go much, much further. We would like to see kind of every major geoeconomic block in the world running the infrastructure, having significant economic stake in the infrastructure, having their local champions in their markets using the infrastructure. And that's critical. If you don't have that, it's going to be very, very hard to get Internet scale. If it's an American thing or what have you. And so blockchain networks and distributed governance and all these things give us some really interesting tools for that. And so that's a big motivation. And again, we're early in that. Like we had the first 10 or 11 folks like launch a few weeks ago with us with Mainnet. That will obviously grow. Ultimately, there can be like 40 slots on the network. And so we really think about that. And then ultimately, those staking validators and operators will vote in elections on the expansion of the membership of the infrastructure partners from around the world. But I think we're trying to steer it. [15:00] from the start to have that strong basis behind it. Jeremy, how are you thinking about your moat and the points of defensibility that you have? You could look at what's going on from two directions. On the one side, you've got AI now makes it so much easier to code up software from nothing. You can create software that's competitive with incumbent software. And then on the other hand, you have legacy financial institutions coming on board with new sorts of products of their own. We're seeing coalitions come together around their own sorts of stable coins, OUSD, Coin, and Citi and Goldman have something coming out perhaps next year, along with a number of other banks and partners. How are you thinking about the defensibility of what Circle is doing? in terms of USDC, but also this new infrastructure that you've created? There's sort of like three pillars across what we're doing. There's the stablecoin network, which is USDC and also our euro stablecoin and other digital assets that we issue. We have a tokenized Bitcoin product. We have the digital euro. We have the largest, I believe, currently tokenized treasury and money market product. And so there's like a whole family there. But USDC is Stablecoin networks themselves, their platform and network affects businesses. The more applications that are connected to the network, the more utility that the network has, which means that the next developer that comes along and says, hey, what network am I going to support? Oh, I should support this because everyone else supports it. You have this really powerful utility flywheel. And that's there. And so primary liquidity, the ability to seamlessly kind of move this in markets globally and financial market centers globally is huge. So we we mint and redeem, you know, one hundred and fifty billion dollars a quarter of this. And it's plugged into all of these banking systems globally. And that liquidity that exists on a primary basis, on a secondary basis in on chain markets and on chain protocols, that's very powerful. deep, deep fundamental liquidity on it means that its capital efficiency is greater and therefore other products that need the most efficient capital benefit from that as well. So you have utility network effects and you have liquidity network effects. And then you also have like these, you know, what I'll broadly call like infrastructure modes, which is to say, you know, we've built regulated infrastructure in all these parts of the world, which is non-trivial. It's a of approved infrastructure in these markets. And we generally have been first in that. And then there's actual infrastructure infrastructure, which is interoperability infrastructure, for example. So the most widely used interoperability protocol in the world right now on all blockchain networks across all assets is Circle's cross-chain transfer protocol, CCTP. And that protocol is the most widely used because USDC is the most transferred digital currency across these networks. chains. Yeah, and that bounces around. It's sort of been as high as 70% in recent quarters. And so it's significant, right? And one of the – actually, it ties back to ARK. One of the big things that we've done with ARK is if you issue an asset – on ARK, a tokenized equity or a stablecoin or a tokenized credit product or whatever it is, your asset actually becomes eligible to run on the payload of cctp and so in a sense like we built the highways to move value around all these networks and the only car that could drive on the highway was usdc until now now anyone that issues an asset so if you're an asset issuer you can ride those same highways and get that interoperability liquidity distribution that flows that's really powerful so the stablecoin network has those attributes and i think um we've what we like to say is like the stablecoin market structure and there's going to be, over time as this goes from several hundred billion to hopefully several trillion, there's going to be multiple networks, right? And so we assume that. We don't assume this is a 90% market, right? It's a winner, take most. And then I think the other piece is we have CPN, which is our circle payments network. That's also a network model. It's built for financial institutions. It is a network not dissimilar from like a Visa or a Swift, meaning FIs join, and then they can use it to move value around globally. using stablecoin rails and blockchains as the as the coordination mechanism for all that value exchange that's growing really fast. We keep reporting every quarter how that's growing and we've got almost a couple hundred financial institutions on it that in turn also supports [20:00] or demand for the stablecoin network. But at the level of ARK, what I would say is it relates to where I started here, which is in this conversation, which is, you know, we think we're going into a new chapter in this space. And in my mind, and I think this will resonate for both of you, hopefully, but, you know, we went through different eras in the Internet's development. And, you know, One era was the mobile era. And we all think of the mobile era as like iPhones and Androids. But the reality is the mobile era started like 15 or 20 years before that. And there was an unbelievable number of attempts at building the mobile operating systems and mobile platforms that we'd all use. And it attracted billions and billions of investment from everything from open source developers to the biggest companies in the world, like Microsoft and Compaq. And so we had the Palm Pilot and we had the BlackBerry and we had the Windows phone and we had the Samsung phones and we had Symbian and Nokia and NTT and many, many, many others. And we all thought at the time, like, wow, mobile's big. Look at all this stuff. Look, you can do an app on a phone. And it was all terrible It actually didn work for mainstream users And so it was only the early adopters that were like leaning in And there was a lot of money that went in to build apps Those apps never turned into real businesses. There was a lot of infrastructure that got built. A lot of that infrastructure just is dead. It's in boxes and garbage sites all around the world, physical products, et cetera. But it was like a long journey. It was like 15 years. And there was a confluence of things and a convergence of things that actually made it possible to actually create both the right kind of device and the right software operating system and experience and the right set of developer incentives and economic incentives to actually make it really work. And Apple and Google both kind of arrived there roughly the same time. You could argue about which one really drove it the most. But when I look at blockchain networks, my mental model is these are operating systems. They attract developers. They attract assets. They're economic operating systems. And I think we've been in that 15-year journey. And it's very hard to say today, like, oh, this is the iOS of that or the Android of that. It's very hard to see that. And it's actually, I think most of us would agree like most of the apps that we have that sit on these are still not like viable to mainstream users We only just starting to see those apps emerge We just beginning like FOMO or you know or like these these apps where you don't even know you're using crypto. And same thing with stable coins. Like you've got these money apps, these neobanking apps where you don't even know you're using crypto. And AI makes that even easier. So I think my view is like going from early adopter to mainstream scale, going from the few hundred million diehards who are there to trade and make money or because they believe in it to the few billion people where this is just this infrastructure that's behind the scenes powering economic activity, that's all still in front of us. And so my view is like this is a wide open space. No one has won significant developer mindshare. The number of active developers on any of these networks is incredibly small when you think about developer activations in traditional platforms. And so it's super early. And so no one has a moat is, I guess, my point. And I think you're exactly right, which is that software velocity is increasing extraordinarily fast. Our own software velocity as a company, we're shipping a lot more software than we ever met. You might not need as many developers after all I mean it phenomenal It phenomenal what we seeing And so I believe that it is a very wide open space right now And I think Circle as a sponsor on this the institutional credibility the stablecoin network that we bring, the stablecoin centricity of the compute layer, the gas layer, other things, and the designing it for this kind of future space. But it's also, we're building this to be, you know, basically distributed in its ownership and its governance. And that's all public and out there. You know, we've ARCToken and ARCToken White Papers out there. Many major firms are stakeholders. That will just continue to be the case, and that will become much more distributed. And so this is why I sort of think of like Google's Android. Android is an open source operating system that Google is the primary author of. They bundle apps that most people include, but they don't have to. And then device makers and hardware manufacturers and telephone carriers and retailers and others have a role to play and value to get out of that ecosystem. And it's 85% of the market or whatever it is. I believe it's early. And I believe the next three to five years will really redefine this. And it comes back to the other comment, which is the primary actors in all of this are going to be AI creating the software. Maybe there's people behind the AI creating the software. [25:00] but it'll also be AI creating its own software. And it's going to be AI... intermediating and interacting and performing economic work. And in fact, cognitive labor is economic work. And so the cognitive labor itself will get performed and manifest and coordinated. and exchanged on these networks as well. Things are going to get weird. And as you said, you're drawing these comps to what has happened in earlier Internet history and particularly the mobile revolution. You know, a point, Chris, that I know you like to make is even with the launch of the iPhone, it was kind of. crappy at first. The first model actually wasn't even very good. And people forget this. They think the iPhone came out and boom, the whole world changed overnight. But there was this whole stop and go sort of development in the early stages of all of this to get to this kind of mobile first world that we all live in and enjoy today. Yeah, I used to carry two phones around when the iPhone first came out because it dropped all so much. Blackberry. Yeah, or just a feature phone or whatever. And it wasn't until, I think 2007 and then I think it was 2012 that Facebook did their mobile pivot, which was sort of symbolic of Silicon Valley realizing this is a really big deal. Which people questioned whether they could pull off. Yeah. There were so many doubts. I mean, I think, you know, it was still – people were still anchored on the framing that the phone is the side gadget, not the primary gadget. And it took a long time for people to kind of realize that. Yeah, no, I think – I mean, I agree with everything Jeremy said. who seem a little bit downcast, I guess. Some of the more ideological actors maybe aren't as excited, but at the same time, just... empirically there's far more traction in use cases than there ever has been before. I mean, the level of adoption on, I mean, Jeremy, maybe I don't know what you can talk about, but like, the, uh, level of adoption for financial institutions. It's just a lot of this is regulation, genius passing, I think just sort of instantly normalized stable coins in the technology world. And you see these companies like Stripe and others who, you know, are not, who are very pragmatic and they're picking the best technology and they're, they're using blockchains. Right. So we're seeing that across the board. Um, And I think just the combination of the fact that crypto has been around a while, gone through cycles, all the focus on AI right now, I think in my mind, it's sort of an overlooked thing. important trend. And well, as Jeremy says, we'll converge with AI. I think it's, you know, we're gonna, we're still relatively early in the AI revolution. If you look at the data, there's like 2% enterprise adoption. Things like agents have very low consumer adoption so far. Obviously, that will grow. That will... put pressure on existing... internet business models as an example, and you'll need new ones. So I think there's sort of this, you know, 10 to 20 year period where you're going to sort of reconfigure the digital economic landscape. And that's dovetailing, I think, timing wise very nicely with the with the rise of highly performant Blockchains like ARK? making catalogs and essentially brochure where and then the 2000s you had this whole new wave of creative you know social media and so forth so i don't know when that will happen like that you know the kind of the like i expect a couple of years i don't know what you think jeremy we'll see kind of more you know taking existing things and making them more performant and and global and so forth programmable but then i think at some point i think people will start to really kind of you know, dig in and get creative. I totally agree with that. And I like I like to say, like, you know, You want to create something where developers and entrepreneurs can can go do stuff. And when you create like these new open platforms with new primitives and in this case, these are like. money primitives, financial contract primitives, other, you know, other software primitives that are there, like the creativity of the developers is what's going to ultimately show us the way. And so people say, like, what are going to be the most important apps? It's very easy to go to our priors and say, oh, it's going to be you're going to be able to do cross-border money movement better, or you're going to be able to, like, pledge your stock as collateral more easily and borrow against it in more places. And like, it's very it's very easy to take your priors and just say, is probably also very true. I use the example of credit, and I think credit is something where the combination of these new blockchain network operating systems and [30:00] This new base layer of fully reserved digital dollar money and ai like these things combined are going to allow us to create completely new types of credit instruments with people who supply money and people who demand money and machines that supply and demand money and are going to be able to create credit marketplaces that probably look more like AdWords than like someone who packages up a bunch of credit and sells it as a bond, right? So you're going to end up with really, really different market structures, unit economics, distribution models, reach the long tail thesis, which we all know really well, but applied to financial markets, which has never really happened. And we're getting early indicators, early views into this, which is like tokenization of RWAs, opening up equities to more people and you can do more with those equities. That's like an early view into long tail, long tail market participants, as opposed to long tail issuers. Well, what if we get long tail issuers of credit and equity and tokens and those economic instruments? So my view is, yes, I'm very aligned with yours. And I think it's going to be very, very interesting. And so the two-year thing, I think that's probably right. I mean, my view is – There are a bunch of different things happening, whether it be like technologies like ARK, like stable coins becoming legal forms of money that people can kind of depend on and then people can rely upon. And actually, what's been amazing, I've experienced this firsthand because I do a lot of I build a lot of stuff with AI now just for fun. Software infrastructure and APIs, whether it's a blockchain computer or a software application, a client application, it's all deterministic. AI is able to learn and implement at an extraordinary pace. And so what that means is that the creative output that we're going to see in terms of what software people create is just going to... blossom like crazy. I'm actually seeing some interesting stuff just in the last week starting to launch on ARK, which are like these crazy games and simulations. They have economic incentives, and they have stablecoin money. And the user experience for using them is you just show up with USDC. You don't need anything else. And it's stuff I've never imagined. And I don't know, is that going to be the killer apps? We don't know. But I think the simplicity with which is and deploy them and make them usable, it's arriving right now. And I think AI and AI software engineering is accelerating that. So I actually expect, we're looking forward 2027, just 'cause we'll use round numbers, I actually expect to see a big proliferation in new apps on these computers that violate the priors rule. Like they violate the skeuomorphism, as Chris was saying, they start to be performing and doing things that were just like, wow, I can't believe I never thought that was possible. So it's an exciting time from that perspective. Jeremy, what's the weirdest thing you've seen people build on Ark at this point? I mean, I just literally like saw this in the last couple of days. There's this thing called Flybrain. Oh, my God. The Flybrain. Yeah. Yeah. And so this, I don't know who the team is because they're just out there. they basically took all of this neurological information about the brains of flies, the real actual computational rules of fly brains. They basically converted it into like 16 megabytes of bytecode. that runs on Arc. They clearly use AI to create a user experience layer. where the fly brains, they've been set loose. And they are actually evolving and engaging and interacting. in this environment. And so I haven't looked at like the token incentives and all like I don't know what the deal is like the deal deal. Like I haven't looked at that, but I'm like, there's like a computational biology slash neuroscience experiment that's taking real world stuff. And now it's creating simulations and they've got some way for you to, I guess, participate in the fly world. But like, I mean, I don't know. I literally I was like, what is this? I wrote on Twitter. I'm like, what is this for? these long explanations of what this could mean for scientific discovery. I'm like, I don't actually fully understand it. And maybe it's just a scam. I don't really know. But it's sort of like it's indicative of like, you know, what AI combined with these computer networks combined with economics, people are starting to try really different things. So we launched a vibe coding app called Arc Studio, which basically like can safely deploy smart contracts. Smart contracts are hard to deploy, because the safety issues are real. So we kind of give it like a smart contract publishing rubber room. And so that's been interesting. [35:00] to like really good developer engagement and people are building all kinds of stuff. They're building, they're building obvious like financial apps, but they're also building like games with economics and other things as well. And so it's been interesting, really interesting to watch. It's actually quite shocking how well many smart contracts have held up, despite all these amazing advances in AI, all these crazy papers coming out and just the abilities constantly improving rapidly. Flybrain network on ARK giving a whole new meaning to swarms of agents. And going back to something that it reminds me very closely of something, Chris, that you like to say, which is that the next big thing starts out looking like a toy. And I'm not saying that FlyNet is the next big thing. But you start seeing all these weird experiments take shape and you never know what direction they're going to they're going to go in and what actually has something of substance behind it that that people are just writing off because it looks super weird. Totally agree. I mean, I think Chris wrote a lot about this and it's fundamental to your rewrite own thesis, which is. You know, when you can start to to create economic incentives for participants in applications and those can be kind of globally distributed and the application is itself like an economically incentivized ecosystem like that, that model, we can really start seeing people bring life to those. because of these advances in software user experience, now you can actually dial it in. And so I'm really interested in people who are doing really novel things with governance, participation, economic contracts, much more complex forms of economic arrangements. that are running in these deterministic you know, systems. And so I feel like we're just because of these advances, we're coming into an era where creativity in that in that arena is going to really flourish. Like, I'll give you another example of a project coming to ARC, which is essentially arbitration for for the work of AI. And that scales from AI jurors or AI, AI arbitrators that kind of scales up into human with the fact that AI will hire other AI and humans will hire other AI for work, and you need arbitration, and you need mechanisms for the money and the release of money around that. That's all now there. You can actually start to do these things. That kind of thing, like automated arbitration of work and economic work and creative work and intellectual property work. Like that was inconceivable. a few years ago, now it's actually very conceivable and could be very scalable and result in some really powerful unlocks for the labor economy and work in general. So I look at things like that and go, wow, again, entrepreneurs are starting to work on problems that couldn't be worked on before, and now they're starting to. The observation, the next big thing looks like a toy, is related to an observation, and the most important things in tech tend to have exponential curves under them. of the improvement of performance of ships, price performance of ships. Right now, there's a lot of focus on AI scaling laws, which are very important. But there are other exponential processes, including network effects, which blockchains have, and also composability, which is composability is the idea that blockchains enable applications to sort of interoperate and be built together like Lego bricks. That's sort of because of these exponential curves, things can, you know, when And then as they get better and the same thing here, like maybe, you know, at first it's just a collective of people doing some kind of economic governance thing. But, you know, if it makes sense and it scales, it can it can. grow very, very rapidly because of these underlying exponential forces of composability network effects. Chris, I mean, you wrote rewrite on a couple of years ago. And at that time, the AI stuff was very nascent. It was only just kind of getting underway. How has your thinking about the future changed? What would you do differently? Or how would you say your thesis has has evolved since then? Yeah. Well, I actually talked about AI a lot in the book and had it as one of the core areas, I think. And part of the positioning was, and look, at the time you could see the scaling laws and I've written about this before and it seemed like it would get much, much better and more important. Part of my argument was that that blockchains in some ways can be an antidote to What I believe is a centralizing tendency of AI. I think AI is amazing. I love AI. But there is this... potential effect where companies with the most capital and other resources, you know, can have an advantage. And so [40:00] Part of my argument is part of the policy argument we make, too, is that it's, you know, that one open source is very important to counter. You know, that's open source has been the most democratizing trend in the history of technology. If you look at things like Linux and then, you know, blockchains are like open source. They're open services. Right. They provide kind of a level playing field and, you know, low fees, global access and so forth for blockchain. all sorts of developers. So, you know, part of my core argument is, And particularly on the policy side, it would be a shame if policymakers didn't allow this sort of countervailing, decentralizing force to flourish alongside policymakers. what might be the sort of centralizing tendency of AI. Yeah, Jeremy, I mean, one question I have for you is how you navigate how much centralization versus decentralization to have on a network like ARK. You know, obviously there are sort of fully public open blockchains that you referenced, Ethereum, Solana, et cetera. And this shift now toward more permissioned. settings. How do you think about where Arc fits and where it is now, where it might go, how much openness can there be on the network or should there be? Well, a couple of things. I think the first is Arc is a public and permissionless network. Anyone in the world can connect to it and build software on it and deploy software on it. It is completely open, public and permissionless. So it has those attributes that are really, really key. And obviously it's all open source as well. So you can investigate it, you could hard whatever you want. So it meets those set of criteria. I think the critical difference, though, is the actual operators of the validation infrastructure are known. They're not anonymous and they are elected and selected on the basis of who they are and the covenants that they make operationally. and legally. So that is different. And so our view is that if you're going to have an economic operating system, that the world's financial institutions and corporations and ultimately households are going to depend on, that actually those operational covenants, are actually very important. Now, there is a there's a purist view that says, no, the operational governance should just be implicit in the performance of the network. And the promise of anonymity and who runs the network is critical. We need to be able to have it be censorship resistant from government from governments. And so anonymous, public, et cetera, and just using crypto to prove it. I think is going to be needed. The world's governments and central banks, because I actually care about them, I want mainstream society running on this, and I want the mainstream financial system running on this, and I want companies, big public companies, putting their contracts and their governance on the internet. I want them to do that, and they're going to want covenants around how it's operated. They're going to want to understand what that is, and they're going to want to understand the election process and the governance process, and that's what we're building. And so it's distributed. It is not fully decentralized. It is open. It's open sourced. It is open in public. It is permissionless to build on and use. We're trying to find, and this is like USDC, what's the happy middle? We're trying to find the happy middle. One could say, you know, USDC is centralized. It's a centralized issuer. Well, if you want a safe, actually liquid integrated into the real world financial system, digital dollar, that's actually how you have to do it. And so we've It was compliant. It was KYC'd. And we actually built a UI where it was like dollars. It wasn't like Bitcoin in your face. It was like, no, we actually think people want the benefits of this technology, but they want it with dollars. And it's going to have to be compliant and all this stuff. So I think it's just sort of this next phase, the middle way. What is the middle way? I think we're trying to... find that balance and strike that balance. Paint a picture for me five years from now. What has changed about the global financial system as a result of the convergence of these new technologies, things like ARK, but also in tandem with fly agent swarms? The honest answer is no one has any idea. at this point because the pace in my view the pace of of technology acceleration. is faster than at any point. in human history. If you believe in the compounding laws that are happening in AI right now, which I do, the capabilities that we'll see are literally unimaginable. It's difficult for the human brain to reason about when you're talking about something like ASI, where the reasoning and coordination capabilities of ASI, they actually, by definition, go beyond what a human brain can even reason about. So fly brains, we become the fly brains. [45:00] I guess the reality is like that that is there and that is happening and so the results of that is like it is very very difficult to know but what what I do think is is we're going to go over the coming years, we're going to go through a shift in towards just an extraordinary amount of the work that happens, the creative work and the cognitive work and all the economic substance and exchange, like where that is primarily operated by these AI machines. And I think what that likely means is sort of like in 1995 or 1998, I like to use these old Internet analogies, like you sort of said, like, hey, Because of this, you know, the compounding laws of the Internet and Metcalfe's law and network effects and better bandwidth, if you had said, like, the amount of communications in the world is going to be a billion times larger than the amount of communications that we have in the world today. And the amount of information published in the world can be a billion times larger than the amount of information that's currently published. Most people would say, like, you're out of your fucking mind. You have no idea what you're talking about. It's just not going to happen. It doesn't make any sense. It's just hard to wrap your head around. But that's exactly what happened. velocity of communications, information output, like these extraordinary curves. And so my view is the velocity of economic activity, the velocity of value exchange, the velocity of the use of financial machines, right, will be like these multiple orders of magnitude larger. And so... How does that translate? I mean, I think it sort of goes hand in hand with the machine economy and this infrastructure, and it should be a world where the creative output of the economy is. is significantly larger than the creative output of the economy is today. The amount of the volume of products and services – and other things that are available broadly distributed around the world should be orders of magnitude larger as well. And so that's what I think things look like in five years. compelling vision. Chris, let's get your view on this also. I mean, I've heard you use this phrase machine age networks, too, to describe what's coming. As Jeremy said very eloquently, I think some of this will be very hard to kind of imagine, but it's pretty clear where we're headed based on a lot of these trends. And I think, to Jeremy's point, it's very plausible that you'll have the vast majority of Internet participants will be machines. And it seems natural that those machines would have economic problems. components to what they do. And so, you know, that's a will be fascinating, interesting kind of new world totally we've already crossed the uh the the majority of activity online now is bots so the humans are a dwindling dwindling portion of the yeah i have a funny story which is very quick which is i i like vibe coded an app to build on arc i won't say what it is i won't say what the domain was because i was just playing around and i don't want to i want to do that i i i like built this thing and i shared it inside my company and then i went i went to it about a week later and I literally never, Like it maybe existed in a Whois database or whatever it was. And it was like it was getting like 15,000 visits a day from bots. who were looking for, you know, environment variable files, SSH credentials and crypto. And like it basically said, like, hey, you're getting, you know, there's like all of these bot swarms, basically, that are just coming looking to find holes and crypto. And so I'm like, OK, well, I think there was one testnet user in my company that tried the thing. And there's all these all these things. to figure out where the exploits are, which is kind of terrifying as well. They're here, and we need to give them things to do. Jeremy, it's a pleasure checking in with you and hearing about the journey Circle's been on. Always keen to learn the latest, so thank you so much for joining us. My pleasure. Thank you, Jeremy.