Plaid Cuts Valuation to $6 Billion as Data Analytics Reshapes Growth

Plaid估值降至60亿美元，数据分析重塑增长格局 Visa并购案告吹迫使Plaid从人工银行数据抓取转型为年复合增长率达93%的AI驱动型金融分析平台。 本期内容回顾了Plaid的演进历程：从2012年通过屏幕抓取技术为Venmo等早期客户提供支持的初创企业，逐步发展为主导性的金融科技基础设施网络。CEO Zach Perret详细讲述了这段充满高风险的运营“过山车”历程——从同意接受Visa以53亿美元收购、经历美国司法部（DOJ）长期的反垄断调查，到最终终止交易，并带领公司应对估值从135亿美元大幅回落至60亿美元的巨大波动。尽管经历了这些宏观周期，Plaid仍成功拓展了核心业务，推出了聚焦实时信用评分、欺诈检测和银行支付的三大高增长分析支柱。最后，Perret反思了传统银行关系的动态演化，以及AI时代的到来如何促使竞争优势从单纯的工程产出转向高层次战略。 Plaid的估值在五年间经历剧烈波动：2020年同意以53亿美元被Visa收购，在金融科技繁荣期达到140亿美元峰值，随后在2026年的一轮融资中定格在60亿美元。 (0:00) 2020年Visa拟以53亿美元发起的收购案，在美国司法部提起旨在阻止该交易的反垄断诉讼后最终宣告终止。 (0:00) 关于Visa收购的内部公告在全员大会召开前20分钟于网络泄露，给现场聚集的350名员工带来了混乱的场面。 (2:30) 美国司法部立案调查的前提是认为Visa作为根深蒂固的垄断者，收购Plaid是为了巩固其垄断地位，尽管Plaid与Visa在具体的消费者交易类型上并不存在直接竞争。 (4:57) 在2020年疫情期间，消费者行为大幅转向数字金融应用，推动依托Plaid技术支持的产品交易量显著激增。 (7:31)

Following a terminated $5.3 billion acquisition by Visa, Plaid leveraged its extensive bank-linking network to pivot into higher-margin data analytics and risk products, navigating extreme valuation swings and macro shifts. The Aborted Visa Merger and Valuation Swings (0:00) Plaid signed a definitive agreement to sell to Visa for $5.3 billion in January 2020. However, the transaction faced an antitrust lawsuit from the Department of Justice, which alleged that Visa was attempting to protect a monopoly. Because the initial merger agreement included a one-year exclusivity window, Plaid exercised its right to walk away in January 2021 rather than endure a projected two-year court fight that would freeze the company's valuation during a fintech surge. The decision initially caused significant internal friction among employees who had anticipated immediate liquidity. Following the termination, Plaid's valuation experienced extreme market volatility. The company raised an Altimeter-led primary round at a $13.5 billion to $14 billion valuation in early 2021, providing partial employee liquidity through a secondary offering. As broader market multiples later contracted, Plaid completed a subsequent financing round at a $6 billion valuation. CEO Zach Perret noted that the company weathered the multiple reset by issuing Restricted Stock Units (RSUs) rather than stock options, which prevented underwater strike-price complications for staff. From Screen Scraping to Bank Infrastructure (17:05) Founded in 2012 after early consumer budgeting apps failed to gain traction, Plaid pivoted to business-to-business infrastructure after realizing that developers struggled to ingest consumer bank data. In its early days, before banks provided dedicated developer interfaces, Plaid relied on screen scraping. Engineering teams built tools that logged into thousands of financial institutions using consumer-provided credentials, relying legally on Dodd-Frank data-ownership provisions. The company endured lengthy procurement cycles, including nine months to integrate with Venmo under eBay's corporate umbrella, while simultaneously serving early startups like Robinhood and Coinbase. Over time, traditional financial institutions shifted from adversarial opposition to active collaboration. Catalysts included government validation of cloud storage via Amazon's GovCloud in 2014–2015, rising bank technology budgets (exemplified by JPMorgan's IT spending scaling to $17 billion annually), and fee competition from fintech platforms like Robinhood. Plaid consequently transitioned the majority of its traffic from screen scrapers to secure, direct API integrations with banks. Product Diversification Into Financial Analytics (27:10) Plaid evolved from a basic bank-linking pipeline into an analytics platform spanning fraud detection, real-time credit scoring, and payment infrastructure. Because more than half of U.S. adults with a bank account have linked it through Plaid, the company leverages broad horizontal data to identify patterns across thousands of disparate apps—a cross-platform perspective that walled-garden institutions like JPMorgan or Visa cannot see alone. Its fraud product flags anomalous activity across different applications, such as a user applying for four loans across four platforms within a 12-hour window, while its credit scoring model incorporates real-time income and expense reductions that traditional FICO scores lag in updating. This structural pivot cushioned the company when macroeconomic conditions deteriorated in the second half of 2022. Rising interest rates chilled retail investing, loan origination, and new fintech signups. However, Plaid's recurring fees and existing payment volumes remained stable. By 2024, Plaid's three new analytics business lines accounted for more than 20% of its Annual Recurring Revenue (ARR) and were compounding at an annual rate of 93%, helping the company achieve financial breakeven. Strategic Leverage and Developer Evolution in AI (52:00) Reflecting on organizational advantages, Perret identified network effects and execution speed as Plaid's principal competitive strengths, rather than raw brand power. While Plaid historically distinguished itself by tackling manual operational hurdles—such as grinding out 12,000 separate bank integrations—Perret argues that the returns on sheer labor volume have diminished. As AI agents increasingly automate manual software tasks, competitive differentiation has shifted toward high-level strategic decision-making and product design. To adapt to new development workflows, Plaid released tools like an MCP server compatible with AI development environments such as Cursor and Windsurf. Plaid is designing its infrastructure so that AI-assisted creators or autonomous models can surface, configure, and integrate Plaid's financial APIs through natural-language prompts without requiring traditional manual setup. Plaid transitioned through regulatory scrutiny, interest rate hikes, and a compressed valuation multiple by pivoting into identity, credit, and anti-fraud products. A remaining strategic question is whether developer workflows driven by AI will continue to reinforce its network effects, or if the democratization of software development will alter how financial APIs are selected and integrated.

Plaid联合创始人兼首席执行官Zach Perret阐述了公司如何从险些被Visa收购的困局中突围：尽管经历了反垄断叫停、估值从高位135亿美元回落至60亿美元的起伏，Plaid仍通过从早期屏幕抓取向开放银行API升级，并依托覆盖超半数美国账户的数据网络，成功构建起反欺诈与实时信用等高增长基础设施业务。 并购终止与估值周期的主动调整 (10:06) 2020年1月，Plaid同意接受Visa提出的53亿美元收购要约，随后遭遇美国司法部反垄断诉讼阻挠。司法部主张Visa意在收购潜在威胁以巩固垄断地位，但Perret指出两者并非直接竞争对手，银行转账与信用卡刷卡极少在同类交易场景中直接对冲。因诉讼预计将导致长达两年的业务停滞，且疫情红利促使金融科技使用量激增、公司实际价值远超协议对价，Plaid于2021年1月根据为期一年的排他期条款主动终止了交易。这印证了Intuit创始人Scott Cook此前对Perret提出的预判，即收购被监管叫停反而可能成就独立巨头。 交易告吹后，Plaid迅速在2021年4月完成由Altimeter领投的高达135亿美元估值融资，并组织员工老股回购以缓解流动性预期落差。随着后续宏观加息周期到来，金融科技估值倍数大幅压缩，Plaid在2026年将融资估值回调至60亿美元。Perret强调，公司采用受限股票单位（RSU）而非传统股票期权来激励员工，有效规避了行权价倒挂风险，且各轮融资条款极其清晰。在估值回调期间，底层业务已实现经营盈亏平衡，并恢复强劲的内生增长。 从屏幕抓取到开放银行生态合作 (21:54) Plaid成立于2012年，最初尝试构建面向消费者的个人记账软件，因用户抗拒被要求削减开支而失败，但团队在这一过程中察觉到金融数据接入的底层痛点。在Venmo工程负责人的建议下，Plaid彻底转向API基础设施开发。由于早期美国缺乏成熟的开放银行法规，CFPB尚未出台细则，Plaid基于多德-弗兰克法案（Dodd-Frank）中关于消费者拥有自身财务数据权的法理，开发了针对数千家银行的复杂屏幕抓取系统，以模拟用户登录提取数据。早期客户包括尚未形成庞大规模的Venmo、Robinhood和Coinbase。 屏幕抓取机制最初频繁触发银行IT安全警报，部分银行因忌惮散户流向新兴交易平台而故意设限。然而，随着消费者数字化诉求的爆发，行业格局迎来三大转折点：亚马逊GovCloud获得政府认可推动了银行上云信心；摩根大通将年度IT预算激增至170亿美元倒逼同行跟进；以及传统机构被迫跟进零佣金等金融科技创新。传统银行最终认识到数据互通不可阻挡，推动Plaid从被动的抓取中介升级为API基础设施合作伙伴，如今绝大多数数据交互均已基于银行官方API直连。 数据网络效应与三大新兴分析支柱 (32:18) 2022年下半年利率暴增导致投资与借贷市场骤冷，散户行为呈现出“行情高涨时扎堆注册、暴跌时停止使用”的周期性特征。这导致Plaid的新用户开户收费出现断崖式下滑，但基于存量活跃账户的按月订阅费与每笔支付手续费保持稳健，公司绝对收入在整个下行周期中依然实现扩张。Plaid借此机会推进战略重构，将2021年融得的大量资金投向工程研发，全面重写底层数据后端，实现了从单纯数据通道向金融数据分析商的跨越。 目前Plaid已衍生出反欺诈、实时信用评估和支付分析三大新兴业务。Perret透露，超过50%拥有银行账户的美国人已通过Plaid关联过账户，这种广泛的横向跨平台视野赋予其超越单一银行机构的能力。例如，平台可实时侦测跨多款应用的异常借贷申请欺诈，或通过消费流水与收入变动评估真实偿债能力，以弥补传统静态FICO评分滞后于用户实际财务状况的缺陷。2024年，这三项分析业务已占Plaid总年经常性收入（ARR）的20%以上，年复合增速达到93%。 隐形品牌定位与AI时代开发范式转型 (52:00) 在品牌与竞争优势方面，Plaid将自身定位于“成分品牌”（Ingredient Brand），力求将前端界面压缩在15至30秒的无感连接中，既通过数据透明机制赢得终端用户信任，又绝不越界侵占客户与用户的主账户关系。Perret依据《七大力量》（7 Powers）框架指出，网络效应是公司最核心的护城河：前端不仅充当极低流失率的注册接入服务，后端跨生态的数据网络更构筑了竞品难以复制的风控及信用承销壁垒。 与此同时，Perret反思了创业早期的运营假设。过去Plaid依赖团队的蛮力攻坚，通过高强度重复劳动手动完成了12,000家银行接口的对接；但在AI时代，代理工具正在迅速自动化此类繁重的基础工程工作，纯粹工作量带来的回报显著递减，组织的核心竞争力已彻底转向战略质量与顶层产品决策。为适应这一变迁，Plaid推出了MCP服务器，旨在让Cursor或Windsurf等AI原生开发工具能够通过自然语言提示词自主调用并部署其金融接口。 Perret总结指出，金融服务业天然与宏观金融周期绑定，公司虽通过逆周期分析产品降低冲击，但宏观波动依然无法预测。随着AI工具使软件开发的执行成本大幅摊平，初创公司在极短时间内冲破1亿美元ARR已成现实，Plaid未来能否依托其庞大的数据网络积淀，在自主AI代理深度重构软件与支付体系的全新竞争周期中保持底层核心地位，将是检验其长期壁垒的关键问题。

## Background 本期视频由商业播客频道 Acquired 于 2025 年 5 月 27 日发布在 YouTube 平台。节目对金融科技基础设施巨头 Plaid 的联合创始人兼 CEO Zach Perret 进行了深度专访，聚焦 Plaid 险些被 Visa 以 53 亿美元收购的内幕、疫情与零利率周期下的剧烈市场起伏，以及公司如何通过拓展反欺诈检测、替代信贷评估和银行直连支付等多元业务，从单一产品进化为更具韧性的行业领导者。 ## TL;DR 访谈全面复盘了 Plaid 跌宕起伏的商业演进史：从早期转型攻克 12,000 家银行数据抓取的苦活累活，到 2020 年宣布以 53 亿美元出售给 Visa，后因面临长达两年的反垄断诉讼而果断放弃并购、选择独立。在经历估值冲上 135 亿美元后回调至 60 亿美元的资本过山车中，Plaid 凭借覆盖全美超半数银行账户的用户网络，成功将业务从单一的账户连接工具扩展至反欺诈、实时信贷评估及支付分析三大高附加值支柱，使新业务贡献超 20% ARR 且年增 93%。Zach 深入剖析了宏观强周期下的商业韧性与组织预期管理，并分享了 AI 时代如何从早期的“蛮力苦干”转向依托更高决策杠杆的技术范式思考。 ## Atmosphere & style 整场对话兼具极高的智识密度与坦诚通透的反思色彩，既有严密宏大的商业战略拆解，也穿插着全员会危机、家庭互动等鲜活生动的细节。主持团队与 Zach 的交流默契自然，在幽默调侃与深度追问中层层剥开并购流产与估值重置背后的惊涛骇浪，情绪曲线从早期的动荡与焦虑逐渐收敛为穿越周期后的从容与笃定。听众能够沉浸式体会到一家明星独角兽在资本泡沫与宏观退潮中自我重塑的真实心理历程，兼具商业传奇的戏剧张力与硬核案例复盘的启发感。 ## Bulletin - [38:45](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2325s) Plaid 在终止与 Visa 的 53 亿美元并购案后，估值曾借 ZIRP 狂潮飙升至 135 亿美元，近期回调并在 60 亿美元完成新一轮融资，目前公司已实现收支平衡。 - [36:10](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2170s) 公司已成功开拓第二增长曲线，反欺诈、实时信用评估和银行支付分析三大新业务在 2024 年贡献了超过 20% 的 ARR，且正以 93% 的年复合增长率高速增长。 - [34:25](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2065s) Plaid 覆盖了全美超 50% 的银行账户用户，依托跨应用的横向数据聚合、收支行为沉淀与官方 ID 交叉验证，构筑起传统银行与卡组织无法企及的数据网络壁垒。 - [52:00](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=3120s) 面对 AI 时代软件开发范式的重塑，Plaid 推出了 Model Context Protocol（MCP）服务器，主动融入 Cursor 等 AI 原生编程环境，捕捉泛开发者与 AI Agent 的集成需求。 ## Key takeaways - [11:13](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=673s) 2020 年初 Visa 拟以 53 亿美元收购 Plaid，但遭美国司法部反垄断诉讼阻击，Plaid 最终在一年排他期届满后主动放弃合并，避免沦为 Visa 为期两年的免费看涨期权。 - [15:10](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=910s) Intuit 创始人 Scott Cook 曾致电 Zach Perret 预言：正如微软试图收购 Intuit 被叫停一样，并购告吹往往能成就一家更伟大的独立企业。 - [17:25](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1045s) Plaid 始于 2012 年面向 C 端的个人记账应用（PFM），在经历用户反感“被劝省钱”的失败后，因 Venmo 对其底层银行抓取技术的兴趣而果断转型为 B2B 金融基础设施。 - [22:15](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1335s) 早期在全美 12,000 家银行缺乏现代 API 的环境下，Plaid 凭借多德-弗兰克法案中“数据属于消费者”的原则，靠编写屏幕抓取程序（Screen Scrapers）完成了从 0 到 1 的原始积累。 - [27:48](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1668s) Plaid 在商业模式上组合了单次获客费、单次支付手续费和月度订阅费（PUPM），存量用户还款与账单支付的刚性需求让其在 2022 年美联储激进加息周期中维持了极强韧性。 - [32:38](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1958s) 2021 年完成高额融资后，Plaid 重构了底层数据架构，从单一的银行账户连接升级为反欺诈、实时信用体系与账户直连支付三大分析型产品支柱。 - [34:25](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2065s) 相较于摩根大通的单一机构纵深和 Visa 的卡组织封闭网络，Plaid 的核心护城河在于横跨多个金融科技应用的全局网络洞察与跨机构收支交叉验证能力。 - [41:00](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2460s) 在经历估值从 135 亿美元回调至 60 亿美元的过程中，Plaid 依靠 RSU 股权结构避免了传统期权倒挂，并通过主动下调内部估值与透明沟通稳定了团队心态。 - [43:47](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2627s) AWS GovCloud 带来的云安全破冰、摩根大通百亿美元级的 IT 军备竞赛，以及 Robinhood 零佣金带来的用户流失危机，是倒逼传统保守银行业全面拥抱金融科技的三大催化剂。 - [46:06](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2766s) Plaid 将自身定位为“成分品牌”（Ingredient Brand），尽管在绑卡弹窗中仅占有 15 至 30 秒的极短交互，却成功在开发者、银行与终端消费者之间建立起跨代际的安全信任。 - [48:31](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2911s) 即使银行普遍开放了 API，Plaid 依然凭借“入驻即服务”带来的更高转化率，以及海量沉淀数据形成的多边网络效应，让后来者难以简单复制。 - [54:03](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=3243s) AI 时代终结了软件工程单纯靠“蛮力堆人力”（Grinder Problems）的汗水红利，技术护城河正急剧向高阶架构判断与战略决策质量倾斜。 ## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=0s) 开场与节目背景：Plaid 的过山车之旅 节目主持人首先向观众介绍了本期重磅嘉宾——金融科技基础设施独角兽 Plaid 的联合创始人兼首席执行官 Zach Perret。对于大多数用户而言，Plaid 最为人熟知的身份是那个无缝连接银行账户与各类金融科技应用的数据中枢。然而，这家公司在过去几年的经历堪称一部跌宕起伏的商业戏剧：2020 年初，Plaid 曾与支付巨头 Visa 达成高达 53 亿美元的收购协议，随后美国司法部（DOJ）以反垄断为由提起诉讼试图阻止交易，Plaid 最终选择终止收购；在此后的零利率政策（ZIRP）与科技繁荣期，其估值一度飙升至 140 亿美元的天价，而在最近的一轮融资中估值回调至 60 亿美元——虽然相较于巅峰有所回落，但仍高于当年 Visa 的收购要约。 本期节目的核心主线，正是探讨 Zach 与团队如何穿透资本市场的暴风雨，在外部剧烈波动的同时深耕底层业务，通过多元化扩张将 Plaid 从单一的银行账户连接工具，转型为涵盖反欺诈数据分析、替代性信贷风控以及银行间直接支付等综合金融基础设施。在正式对话开始前，节目组还插播了一则预告：Acquired 即将于 7 月 15 日携手摩根大通支付（J.P. Morgan Payments），在纽约可容纳 6000 人的无线电城音乐厅（Radio City Music Hall）举办大型线下活动。 - **估值演变轨迹**：从 2020 年初 Visa 拟以 53 亿美元收购，到后来的 140 亿美元估值顶峰，再到近期以 60 亿美元估值完成融资。 - **业务多元化方向**：从最初的账户身份认证（Auth）与连接，拓展至欺诈检测、替代信贷评估及账户间转账清算网络。 --- ### [01:24](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=84s) 创业过山车：五年间的戏剧性剧变 对话伊始，主持人调侃道，回顾 Zach 过去五年的经历简直令人眼花缭乱：原本是一家万众瞩目的明星独角兽，突然宣布卖身 Visa，紧接着交易告吹，随后以极高估值巨额融资，再以较低估值追加融资；与此同时，公司业务并未停滞，反而开辟了大量外界尚未完全了解的新业务线，而大众对其认知往往还停留在“在 fintech App 里授权关联银行账户的弹窗”。主持人发问：“这是否算得上对你过去五年生活的一个合理概括？” Zach 笑着全盘认同：“非常合理。”他感叹道，人们常说创业就像坐过山车，“但事实上这趟过山车从来没有终点，它一直在往前开，而且波动的振幅还越来越大”。这一路走来充满了高强度的学习与极具成就感的时刻，但背后也有数不清的挫折、沮丧与艰难抉择。主持人更是打趣道，对于一个名为“Acquired”（被收购）的播客来说，平时其实很少专门讲未达成的并购案，而这期节目恰恰是一次回溯并购话题的“复古回”，只不过结局是收购最终流产了。 --- ### [02:30](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=150s) 2020年1月：53亿美元出售给 Visa 与那场难忘的全员会 时间回拨到 2020 年 1 月 18 日或 19 日前后，Plaid 正式对外宣布 Visa 将以略高于 50 亿美元的价格将其收购。这笔交易并非一时冲动，而是双方在过去三到四个月里密集拉锯谈判的最终结果。此前 Visa 已经是 Plaid 的早期投资人兼业务合作伙伴，彼时不仅 Visa 表达了强烈的收购意向，市场上也出现了其他潜在买家的竞购兴趣。在权衡各种利弊之后，管理层做出了一个在当时看似极度理智、但过程异常艰难的决定：将公司出售给 Visa。 宣布消息的那天成为了 Zach 职业生涯中极具象征意义的一幕。那也是全员转入居家办公前，在公司旧金山办公室（原 Atlassian 旧址）举行的最后一次大规模全员会（All-Hands）。那座办公楼里有一处标志性的双层超宽大台阶，宽达三四十英尺，足以容纳数百人。当时现场聚集了约 350 名员工。然而在会议开始前 20 分钟，收购的消息不幸被提前泄露到了媒体上。那些时刻刷新科技新闻的员工已经震惊不已，而没看新闻的员工则对周遭的骚动感到莫名其妙。Zach 苦笑着推测，泄密者大概率是参与交易的投行人士或外部律师。 面对这起突如其来的泄密，Zach 不得不走上台阶面对全场。他直言这是自己“人生中开过第二艰难的全员大会”： > “你看着台下的人群……能同时看到愤怒、沮丧、震惊，还有喜悦。有些人甚至在哭泣，我一时都分不清他们脸上的泪水到底是因为高兴还是因为难过。” 随后，Visa 团队进场向大家表达了对整合前景的兴奋与期许。对于一家年轻的初创公司而言，处理这种级别的组织变革管理（Change Management）几乎是对创始人的极限考验，但在当时，整个团队最终还是渡过了最初的情绪震荡。 --- ### [04:57](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=297s) Plaid 与 Visa 的真实关系：直接竞对还是新旧范式？ 主持人随后提出了一个行业普遍存在的困惑：在传统认知中，Visa 代表着旧世界的既有金融秩序与银行卡垄断网络，而 Plaid 则致力于构建一个更快、更具现代科技亲和力的数据织网，两者的结合看起来颇具讽刺意味。对此，Zach 做出了严肃澄清，他认为这是外界对 Plaid 和 Visa 关系中最大的误解——两者根本不是直接竞争对手。 Zach 详细解释道，Plaid 从未涉足发卡业务，也没有去搭建一套与 Visa 争夺刷卡清算的卡组织网络。虽然从某种宏观抽象的视角来看，银行账户直接付款（如 ACH 转账）与刷信用卡可以算作广义上的“替代品”，但在现实商业世界的具体交易场景中，几乎没有任何一笔交易会同时将这两种支付手段视为等同且可互相替换的选项。因此，尽管后来美国司法部发起反垄断诉讼，指控 Visa 是垄断者且收购 Plaid 旨在消除潜在竞争对手以扩大垄断版图，Zach 始终坚称两家公司的产品在微观层面并无重叠： > “我们从来没有认为彼此的产品存在直接竞争。如果这场交易的底层逻辑本身是反竞争的，我们当初根本就不会选择推进这项交易。” 当然，Zach 也承认，在更高维度的生态演进中，大量开发者正将 Plaid 作为核心底座，在 Visa 卡网络体系之外搭建新一代支付和交易轨道。从“传统金融体系 vs. 新兴金融科技生态”的宏观范式而言，两者的确站在时代转折点的两端；但在具体产品功能上，双方当时并不存在直接互斥的竞争关系。 ```mermaid flowchart TD accTitle: Plaid与Visa在金融生态中的定位对比 accDescr: 展示Visa与Plaid在微观产品和宏观范式层面的差异与重叠 subgraph Macro["宏观范式视角（外部监管与行业视角）"] Legacy["Visa 传统金融体系<br/>（卡网络垄断、成熟结算渠道）"] Disrupt["Fintech 新一代网络<br/>（以 Plaid 为底层连接的数据与新兴支付轨道）"] Legacy <-->|"生态演进与潜在替代"| Disrupt end subgraph Micro["微观产品视角（Zach Perret 的业务视角）"] VProd["Visa 核心能力<br/>卡组织网络、发卡清算"] PProd["Plaid 核心能力<br/>银行账户连接、身份验证与数据聚合"] VProd -.-|"非直接竞争<br/>（极少在同一场景互为等价支付手段）"| PProd end ``` --- ### [06:36](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=396s) 全员会上的百感交集：老员工的释怀与新员工的不甘 当被问及为什么那场全员大会上员工的情绪会如此剧烈、甚至有人流下悲伤的眼泪时，Zach 剖析了初创公司文化中深刻的人性与心理机制。对于很多人来说，加入一家高速成长的初创公司（尤其是创始人及早期骨干），个人身份认同与公司命运是高度绑定的，工作成为了自我实现的最核心载体。 员工情绪的撕裂主要源于加入时机与预期回报的巨大落差： - **早期老员工（在职 7 至 9 年）**：很多人入职时 Plaid 的估值仅为 5000 万美元左右。面对 53 亿美元的出售案，他们持有的期权和股权实现了近百倍的惊人增值，这不仅是财务自由的释怀，更是多年青春倾注后的一种庞大情绪释放。 - **后期新员工（入职仅 1 年左右）**：这部分员工加入时公司估值已经很高，他们满心期望 Plaid 能够独立走上 IPO 之路，从当前估值再实现 100 倍的增长、成为万亿美元级的巨无霸。对他们而言，53 亿美元被传统巨头收购意味着独立造梦进程的戛然而止，难免感到大失所望与不甘心。 台下员工的表情极其复杂：有人为财务回报和产品未来的背书欢呼雀跃，有人却因未竟的宏图壮志而黯然神伤。Zach 直面这 350 张交织着喜悦与失落的面孔，深切体会到了在重大并购节点上团队预期管理的极端难度。 --- ### [07:31](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=451s) 突如其来的疫情考验与并购协议中的“大流行条款” 宣布收购之后的一两个月里，生活一度按部就班地推进。整个 2 月到 3 月上旬，管理层忙于向各方沟通传达战略意图，有条不紊地推进并购交割流程，向全球各地监管机构递交合规审查文件。由于坚信自身业务与 Visa 并不冲突，Plaid 内部当时丝毫没有预料到反垄断会成为拦路虎，更未曾想到世界格局即将发生翻天覆地的剧变。 随后，2020 年 3 月中旬，COVID-19 疫情在全球爆发。面对突如其来的全员居家隔离与外部混乱，公司全力投入到保障团队安全与业务运转中。到了 4 月中旬，随着资本市场剧烈动荡，管理层开始回过神来思考这笔并购的命运：“世界都变成这样了，这笔交易还能顺利完成吗？” Zach 随即致电参与并购的外部律师询问交易进展，律师给出的答复却出乎意料地淡定：“别担心，我们合同里有‘大流行条款’（Pandemic Clause）。”Zach 当时一头雾水：“世界上怎么可能会有专门的‘大流行条款’？” 事实证明，在并购交易协议的标准格式中，关于“重大不利影响”（Material Adverse Effect, MAE）的免责除外条款里，通常会预设将全球性公用卫生大流行病列为不可抗力除外情形。这意味着，即使全球暴发大流行、即便 Visa 自身的股价在此期间遭遇暴跌，从法律约束力上，Visa 依然有法定义务按原定条款完成对 Plaid 的收购。在 2020 年 4 月的那个节点，手握坚挺收购合同与大流行保护条款的 Plaid 管理层，一度觉得自己做出了无比英明的商业决策。 ### [08:33](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=513s) 疫情反转与意外迎来的 Fintech 大繁荣 在估算 Visa 收购方案的细节时，Plaid 团队意识到这笔以现金为主、辅以部分股票的交易中，员工将获得按美元计价的大量 Visa 股权，直接锚定 Visa 的最新股价。团队当时甚至惊呼：“我的天，这笔交易明天要是交割，我们可就要持有 Visa 相当大比例的股份了。”然而，世界局势随后发生了急剧逆转。随着新冠疫情席卷全球，居家隔离的消费者发现日常金融需求不但没有减少，反而全面转向线上，使用 Plaid 底层服务的产品和应用瞬间被引爆。 与此同时，零利率政策（ZIRP）让资本市场进入“钱是免费的”时代。初创公司如雨后春笋般涌现，Plaid 的客户几乎全部顺利拿到巨额融资，业务全面开花。加密货币迎来大爆发，出入金（on-ramp / off-ramp）通道需求激增，各种挑战者银行（Neobanks）更是层出不穷，Fintech 行业迎来了前所未有的狂暴牛市。Zach 感叹道，在那个阶段看着所有增长曲线垂直拉升，“那段日子真是太好玩了”。 ```mermaid flowchart TD accTitle: 疫情与零利率环境下的 Fintech 繁荣链条 accDescr: 展示了从外部宏观因素到 Plaid 业务爆发及估值提升的连带效应 subgraph Macro["外部宏观催化剂"] A["居家隔离政策"] --> B["线上金融服务需求暴增"] C["零利率政策 (ZIRP)"] --> D["创业资本充裕 / 客户全线融资"] end subgraph PlaidGrowth["Plaid 核心驱动"] B --> E["Plaid 支持产品用量激增"] D --> F["加密货币与 Neobanks 爆发"] E & F --> G["业务飞速增长，超越原收购预期"] end ``` ### [09:30](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=570s) 反垄断阻力与打破收购“待机”常态 在业务狂飙突进的同时，监管审查的乌云开始聚集。Plaid 团队收到了美国司法部（DOJ）的第一轮反垄断问询（First Request），这在大型交易中本属常规流程；但紧接着迎来的第二轮深度问询（Second Request）却完全出乎意料。Plaid 的自我定位始终是认为自身与 Visa 并不存在直接竞争关系，然而司法部似乎借此机会盯上了 Visa，审查中掺杂了诸多甚至与 Plaid 本身并无直接关联的陈年旧怨。 通常情况下，初创公司在被行业巨头收购的漫长审批期内往往会进入“自动巡航”的停滞模式。但 Zach 和团队在最初谈判中就争取到了完全独立的运营权——交割后作为 Visa 旗下子公司，Zach 依然担任 Plaid 的独立 CEO。因此，团队将自己完全当成独立初创公司向前全速推进。虽然在招聘高管时遇到了一些阻碍（候选人对收购后的定位存疑），但凭借业务的高速成长，公司依然大举招兵买马。 - **独家排他期截止日**：交易合同规定，自签署起满一整年（即到 2021 年 1 月），排他协议到期，Plaid 有权直接退出交易。 - **管理层心理博弈**：到了 10 月、11 月，管理层开始互相对视并意识到，如果审查拖满一年，他们完全可以主动终止收购。 ### [11:13](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=673s) 拒绝当“免费期权”：终止 50 亿美元交易 进入 2020 年 12 月与 2021 年 1 月，司法部明确释放出将提起反垄断诉讼以阻止合并的信号。尽管 Zach 自信公司如果应诉很有可能获胜，但这场官司预计要打上整整两年。主持人指出，若在一年排他期之后再陪跑两年官司，本质上等于锁死 50 亿美元的价格，白送给 Visa 一份价值连城的“免费看涨期权”——无论 Plaid 在这两三年内如何翻倍成长，收购价都不会改变。 此时，当初选择出售公司时面临的诸多业务担忧早已烟消云散，Plaid 的实际价值也已经远超原先的价格。于是，在 2021 年 1 月排他期届满之际，Plaid 正式做出决定：放弃与 Visa 的并购，重回独立发展之路。 ### [12:15](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=735s) 艰难的全员信与 Altimeter 领投的 135 亿美元融资 主动放弃已经到手的真金白银，让 2021 年 1 月的全员大会成了 Zach 职业生涯中最难熬的一场“All Hands”。当时 Zach 独自坐在办公室里，面对 Zoom 屏幕另一端的全体员工，不得不戳破大家的财富幻象：“我知道很多人已经在脑海里盘算着怎么花这笔收购款了，甚至打算去买房——但我必须很抱歉地告诉大家，房子买不成了，这笔钱不会按时到账了。”尽管当时已有外部顶级投资人主动找上门希望以更高估值注资，但未尘埃落定前，Zach 绝不向员工空头许诺。 转机来得极快。就在交易告吹后不久的 2021 年 4 月，Plaid 迅速完成了由 Altimeter 领投的新一轮融资： - **估值暴增**：估值从 Visa 收购时的约 50 亿美元直接跃升至约 135 亿美元。 - **员工老股套现（Secondary）**：公司为绝大多数员工安排了老股变现渠道，允许员工以极高的估值卖出较小比例的持股。虽然可能不足以全款买房，但足够员工买辆新车或者套现改善生活，极大缓解了团队的财务与心理压力。 然而，文化层面的余波依然存在。此前一年为了应对爆发式增长，Plaid 招揽了大量以为自己加入的是“大厂安全伞下的初创公司”的员工；随着重回独立，部分不适应高强度创业文化的员工只得选择离开，团队经历了一轮不可避免的阵痛。 ### [15:10](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=910s) 来自 Scott Cook 的神预言：Intuit 与微软往事 大体量并购在最后一刻流产，对绝大多数初创公司而言往往是致命的打击，士气重挫甚至从此一蹶不振；像 Plaid 这样在 50 亿美元并购告吹后反而估值翻倍、势如破竹的案例实属凤毛麟角。Zach 透露，其实早在并购消息刚刚公诸于世的第一天，他就接到了财务软件巨头 Intuit 创始人斯科特·库克（Scott Cook）发来的邮件和电话。 Cook 在电话里首先对并购表示了祝贺，但随即意味深长地补上了一句：“但是 Zach，我只想告诉你，如果这笔收购最终黄了，你们不仅不会完蛋，而且很可能会变得极其伟大。”当 Zach 询问这是否与他的亲身经历有关时，Cook 坦言：“没错，我这辈子经历过最棒的事情，就是当年微软试图收购 Intuit 时，那笔交易被反垄断机构彻底叫停了。” ### [16:28](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=988s) 交易告吹后的清晨与反思：从 Visa 阴影走出的下一步 在 Visa 收购案宣告终止后，Zach Perret 谈到了一段耐人寻味的插曲。早在宣布出售给 Visa 的第二天，就有一位前辈给他泼冷水并埋下了一颗种子：“祝贺你，但如果这笔交易没成，你未来的发展可能会更好——因为 Plaid 有潜力成为一家价值数千亿美元的超级巨头。”这番话宛如预言一般印在了 Zach 心里。当 Plaid 最终正式宣布放弃被 Visa 收购的那一天，Zach 第一时间给对方发去邮件：“嘿，能一起吃个晚饭吗？我想和你聊聊……顺便感谢你‘促成’了这一切。”Zach 幽默地表示，除了请教如何让那个宏伟愿景成真之外，他当时迫切需要向对方请教现实的管理难题：在收购告吹的巨大动荡下，究竟该如何设计员工薪酬体系，才能安抚团队、防止内部发生“叛乱”与动摇。 紧接着，主持人和 Zach 将话题转向那段魔幻的繁荣期及其随后的剧烈转变。在进入资本狂欢与后来的泡沫退潮之前，主持人希望 Zach 能先回溯 Plaid 的起源与本质：Plaid 在那个时期究竟是一家怎样的公司？唯有理解他们从何而来，才能看清他们如何穿越繁荣与萧条的周期。 ```mermaid flowchart TD accTitle: Plaid 从失败的记账应用转向金融基础设施 accDescr: 展示 Plaid 从 2012 年开发记账软件遇到瓶颈，到转向为 Venmo 等产品提供银行 API 的演进过程。 subgraph Phase1 ["2012 消费端探索"] A["个人财务记账 App (PFM)<br/>90% 时间耗费在获取银行数据"] B["产品痛点：用户不喜欢被劝导少花钱<br/>消费级增长遇阻"] A --> B end subgraph Phase2 ["转折与契机"] C["与 Venmo 工程主管交流<br/>'你们的 App 很蠢，但我想要你们的底层技术'"] B --> C end subgraph Phase3 ["基础设施核心定位"] D["放弃面向消费者的应用<br/>全力构建银行账户 API"] E["连接千家万户与金融科技<br/>(Venmo, Robinhood, Coinbase)"] C --> D D --> E end ``` ### [17:25](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1045s) 2012 年的金融荒漠与做“劝人省钱 App”的惨痛教训 Zach 将时间拨回 2012 年。当时的核心创业假说是：整个银行系统和金融服务体系，都是为一个完全没有预见互联网存在的旧时代而搭建的。彼时移动互联网浪潮早已扑面而来——人们口袋里装着智能手机，Instagram 坐拥海量用户并在那一年被 Facebook 收购，数字消费习惯已经普及。然而，传统的金融服务却像停留在石器时代：无论是开办新支票账户、申请贷款，甚至只是挂失补办一张借记卡，用户都被迫亲自走进线下银行网点，找业务员面对面办理。刚上线不久的 Venmo 虽然开始崭露头角，但出入金流程异常艰难，转账需绑定借记卡或信用卡并承担 3% 的高昂手续费；Venmo 虽曾短暂免除手续费以换取增长，却立刻面临巨额亏损，深陷“免费促增长但极度烧钱”的困境。 彼时市场上所谓的“Fintech”屈指可数（甚至连“Fintech”这个名词直到 2016 年才被创造出来）。除了 Intuit 旗下的 Mint、PayPal、刚起步的 Braintree 以及初期的 Stripe，几乎没有现代金融技术生态。在创办 Plaid 之前，Zach 和团队其实是在做面向消费者的移动记账与预算工具（PFM, Personal Financial Management），试图向用户推荐消费建议。但残酷的现实很快给他们浇了一盆冷水：“根本没人想用一个整天教育自己‘少花钱’的应用。”Zach 坦率地自我剖析：虽然如今像 Rocket Money 这样将预算管理与房贷资格审核绑定的应用证明了 PFM 的商业价值，但他自认并非出色的消费级产品设计师；更致命的是，当时团队多达 90% 的精力都在硬磕如何把银行数据扒下来抓进应用里，根本没空打磨出优质的客户端体验。 - 2012 年的金融体验极其割裂：即便移动互联网已经大行其道，更换借记卡和开立账户仍需前往线下实体网点办理。 - 早期 Venmo 每笔交易收取 3% 费用，用户明知昂贵却仍愿意支付，证明了消费者对线上实时转账体验的极端渴望。 - Plaid 初期的 PFM 应用遭遇滑铁卢，也让团队深刻意识到真正卡脖子的是底层数据管道，而非前端记账逻辑。 ### [20:18](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1218s) 意外的 B2B 转型：从 Venmo 的嘲讽到漫长的 eBay 采购链 在消费端碰壁之际，一次与 Venmo 工程负责人的关键对话彻底改变了 Plaid 的命运。那位主管直接对他们说：“听着伙计们，你们做的那些 App 确实挺蠢的，但我对你们搭建的后端底层架构非常感兴趣，我们愿意出钱授权使用它。”Zach 和团队如梦初醒，当即意识到这才是真正的金矿，毫不犹豫地砍掉消费级产品，全面转型做金融基础设施，目标是打造一个“银行账户的统一 API”。 然而，拿下这位大客户的过程却充满了大公司兼并带来的官僚折磨。Venmo 刚被 Braintree 收购，而 Braintree 紧接着被 PayPal 收入囊中，PayPal 背后的母公司则是 eBay。结果，Zach 不得不跟 eBay 的采购人员周旋拉扯了整整九个月。在这漫长又痛苦的采购流程期间，Plaid 没有坐以待毙，而是迅速将这套银行连接 API 卖给了一众饥渴的新兴初创公司，其中就包括早期的 Robinhood 和 Coinbase。在那个阶段，Plaid 扮演了让用户将传统银行账户（例如 Wells Fargo）一键连接到 Venmo 转账、连接到 Robinhood 充值炒股、连接到 Coinbase 买币的核心管道，随后逐步扩展到报税、报销管理及记账系统，乃至成为整个金融科技行业的底层基础设施。 - Venmo 成为 Plaid 首个展现出极强购买意向的大型潜在客户。 - 由于 Venmo 经历了“Braintree → PayPal → eBay”的三重嵌套收购架构，导致商务采购流程被拉长至 9 个月。 - Plaid 利用这一窗口期快速渗透到 Robinhood 和 Coinbase 等早期高增长平台，建立起强大的先发优势。 ### [22:15](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1335s) 屏幕抓取时代的野蛮生长：法规掩护与万所银行的“猫鼠游戏” 主持人抛出了一个尖锐的问题：如今 Plaid 与银行之间是正规的工程团队级深度合作，但在 2012 到 2013 年，传统银行根本没有任何对外暴露的现代 API，Plaid 早期到底是在写什么代码来实现这一切的？Zach 直言不讳地承认，在没有 API 的岁月里，他们全靠编写**屏幕抓取程序（Screen Scrapers）**，以“自外向内”的方式替用户爬取数据。这是一项在工程上极其恐怖而复杂的脏活累活，因为他们需要面对全美多达 **12,000 家**各具特色的银行与金融机构，适配无数脆弱且频繁变动的前端界面。 ```mermaid flowchart LR accTitle: Plaid 银行数据连接方式演进 accDescr: 展示从早期的屏幕抓取与法律探索过渡到如今的规范化 API 直连模式。 subgraph EarlyDays ["早期模式 (2012-2016)"] A["多达 12,000 家银行/金融机构"] -->|无开放 API| B["Plaid 编写屏幕抓取程序<br/>(Screen Scraping)"] B -->|法律依据| C["Dodd-Frank 法案原则：<br/>数据归消费者所有"] end subgraph ModernEra ["现代合作模式"] D["CFPB 明确制定数据细则"] --> E["银行主动开放官方 API"] E -->|工程级直接对接| F["绝大多数数据通过<br/>API 直连传输"] end EarlyDays -.->|监管明确与生态成熟| ModernEra ``` 从法律层面上看，Plaid 并非毫无防备的法外之徒。虽然英国和欧洲出台了明确的 Open Banking 法案强制银行构建开放 API，而美国《多德-弗兰克法案》（Dodd-Frank Act）虽然确立了“消费者对其金融数据享有所有权并有权授权共享”的基本原则，但负责落地细则的消费者金融保护局（CFPB）当时迟迟没有出台具体监管规程。Plaid 团队在咨询大量法律专家后确立了底气：既然消费者拥有自己的数据，银行不主动提供 API，消费者就有权借助技术工具“登录并获取数据”。 在实际执行中，Plaid 与银行的关系呈现出奇特的“爱恨交织”： 一方面，不少银行表现出积极合作的态度，甚至主动对 Plaid 说：“我们的终端客户强烈想要使用 Venmo、Robinhood 和 Coinbase，但我们内部还没把 API 攒出来，请你们先通过屏幕抓取把通路接上”； 另一方面，部分银行则对新技术抱有强烈的敌意和戒心——尤其是 Robinhood 蚕食了大量传统券商与银行的股票交易佣金，银行的 IT 系统不断拉响警报，故意设置障碍限制数据抓取。直到后来 CFPB 开始正式制定规则并明确“银行不得阻止消费者获取交易数据和路由号码（Routing Number）”，双方的关系才彻底走向规范化与全面合作，如今 Plaid 的绝大多数数据流都已切换为正规的银行 API 直连。 ### [24:36](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1476s) 早期“笨拙爬虫”到现代 API 生态的跨越 在早期的银行数据集成领域，各大银行缺乏构建完善 API 的技术储备与精力。Plaid 不仅与银行合作、为其搭建 API 平台，还协助其实际上线接口，如今整个体系已经迈入了更为长期、可持续的技术基础设施阶段。然而，回顾创业初期，这其中充满了极高强度的“重活累活”。主持人们打趣地描绘当年向风险投资人路演时的荒诞情景：“我们现在正在对银行网页进行屏幕抓取（screen scraping），但别担心，未来我们会用上 API 的；我们先拿到用户的账号和密码存起来，然后派一个后台机器人假装是用户本人登录银行，执行转账操作……”听起来简直“浑身写满了危险”。然而，这恰恰展现了一家伟大科技公司的成长路径：起步时虽然笨拙甚至丑陋，但最终抵达的终点，却是整个生态极度渴求且对消费者体验更好的未来。 扎克·佩雷特（Zach Perret）坦言，刚开始创立公司时，团队自己都不确定是否真的有人会用：当用户要把银行账户绑定到 Venmo 时，到底有没有人愿意这么做？底层技术能不能跑通？消费者真愿意输入银行账号密码这样极其敏感的数据吗？事实证明，用户的涌入是“成群结队”的（in droves）。尽管 Plaid 对自己的后台安全、审查机制与隐私保护极有信心，但上线初期依然心里打鼓。最终，由于对消费者的实际价值极高，加上合作伙伴打造了极其出色的产品，整个金融科技生态彻底爆发。扎克认为，这是一个典型的“先有鸡还是先有蛋”的问题：如果早期生态没有借助那些较难规模化的集成手段迅速做大，银行在后期也绝不可能有动力去主动搭建高可用的 API。 ```mermaid flowchart TD accTitle: Plaid 银行连接技术的演进路径 accDescr: 展示 Plaid 从早期屏幕抓取到推动银行建设原生 API 的闭环演进过程。 subgraph Early["早期破局阶段 (0 到 1)"] A["高壁垒/不具扩展性的方案<br/>(网页屏幕抓取 Screen Scraping)"] --> B["获取高价值消费者用例<br/>(如 Venmo 绑卡转账)"] B --> C["用户成群结队涌入<br/>(验证市场爆发性需求)"] end subgraph Scale["生态成熟阶段 (1 到 100)"] C --> D["倒逼银行重视开放连接"] D --> E["Plaid 为银行赋能<br/>(合作搭建与上线 API 平台)"] E --> F["现代化原生 API 基础设施<br/>(长效可持续、安全、合规)"] end ``` --- ### [26:28](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1588s) 从单一“账户绑定”走向多元产品矩阵 在 2022 年金融科技盛极而衰的转折点到来前，Plaid 的产品线早已不仅仅局限于核心的银行账户绑定（bank linking），而是围绕这一核心场景向上下游及临近领域进行了深度延展： - **信贷与核验产品线**：拓展了资产核验（Asset Verification）与收入核验（Income Verification）。用户在申请房贷时需要验证资产与收入，在申请汽车贷款时需要证明在职就业状态，Plaid 将这一原本繁琐冗长的流程全面数字化。 - **身份验证收购与整合**：收购了一家 ID 验证公司。由于用户在绑定银行账户的前后往往需要核实身份，Plaid 将两者无缝串联，实现“一次验证 ID，后续每次绑定银行卡即可自动携带已核验身份”的极简体验。 - **银行原生支付基础设施**：突破了“仅抓取账户路由号码（Routing Number）”的局限，让企业可以直接发起 ACH 扣款或账户转账，用于账单支付、账户资金充值等场景。 - **底层分析层（Analytics Layer）**：在此之上构建出庞大的数据智能与风控分析网络。 到了 2022 年，Plaid 凭借丰富的端点和超过 100 多个 SKU，覆盖了市场上绝大多数头部 Fintech 公司，并加速向传统大银行与企业级客户渗透。业务增长势头凶猛的同时，Plaid 也将自己完全暴露在了宏观金融周期的剧烈波动之中。 --- ### [27:48](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1668s) 宏观加息周期下的冲击与韧性 金融服务业具有明显的强周期属性，其起伏很大程度上由基准利率所决定。在零利率政策（ZIRP）时期，海量资金在整个经济体中加速流转；而到了 2022 年下半年，美联储激进加息，直接导致投资市场骤降、借贷市场瞬间冻结——放贷机构不再放款，股票与加密货币价格暴跌。扎克分享了一个有趣的消费者行为洞察：数据表明，散户往往在股价和加密货币处于历史高位时蜂拥开户购买，而在价格暴跌时活跃度骤降，完全与“别人贪婪时我恐惧”的理性投资准则背道而驰，真实的人性恰恰是“在他人贪婪时更加贪婪”。 面对急转直下的市场环境，Plaid 的多元商业模式展现出了独特的抗风险韧性。Plaid 拥有上百个产品代码（SKU），整体计费模式主要由三部分构成： | 计费维度 | 具体模式 | 加息周期表现 | 核心影响与原因 | |:--- |:--- |:--- |:--- | | **单次操作 / 获客费** | 每新增一个注册用户收取一次性费用 | 悬崖式下跌 | 借贷机构与券商因资金成本骤增而大幅缩减拉新规模。 | | **单次支付费** | 每次发起转账/付款按笔收取手续费 | 保持相对稳定 | 存量用户仍需持续还款、缴纳基础生活账单。 | | **月度订阅费 (PUPM)** | 按每个活跃用户每月收取基础维护费 | 表现极其坚挺 | 合作企业依托底层账户维系基础客户关系与数据查询。 | 这种结构意味着，虽然借贷机构收缩了新增放贷，但绝不会停止向存量借款人收取分期还款。因此，Plaid 在绝对收入维度上并未出现下滑，但由于新增用户接入量骤降，整体营收的相对增速显著放缓。整个 2022 年下半年至 2023 年上半年是行业极其煎熬的调整期；但自 2023 年下半年起，经过 2024 年直至 2025 年，Plaid 的营收增速已经连续多个季度重新加速。 --- ### [30:24](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1824s) 2021 年大额融资与“战略底座重塑” 在被问及所谓“战略底座重塑”（replatformed our strategy）这种带着浓厚战略咨询味的说辞时，扎克笑着承认自己确实曾在贝恩咨询（Bain & Company）工作过整整“12 个月零 1 天”。他回忆道，当时在贝恩拿到了人生中第一笔年终奖金——4,000 美元，“那是我这辈子见过最大的一笔钱”，这笔钱支撑他毅然辞职去纽约创业，并在付了两个月房租后迅速重新陷入身无分文的境地。 谈到业务从单纯的“银行账户绑定”走向多元化分析业务，扎克强调这绝非因为 Fintech 泡沫破裂才仓促做出的被动转型，而是公司创立之初就确立的长远蓝图：第一阶段必须先拿下银行连接基础设施，建立无处不在的数据管道；第二阶段则是依托跨越整个系统的沉淀数据与用户行为模式，构建高价值的分析业务，从根本上重塑金融风控与信用评估体系。 实现这一战略跳跃的真正转折点，是 Plaid 在 2021 年完成的大额融资。公司利用这笔充足的资本大举扩张工程研发团队，自底向上全面重构了数据基础设施底座，使其能够承载高并发、深层次的复杂分析产品。从财务模型来看，这完全符合标准的高成长曲线：初期因大力投入技术研发使得营业利润率（operating margin）阶段性下滑，随后新业务带来规模化商业回报，驱动利润率与营收强势反弹。重构后的核心武器，正是基于数亿级绑定银行账户、覆盖海量金融 App 行为轨迹，并打通了政府签发法定 ID 与丰富行为数据残留（data exhaust）的综合数据智能网络。 ### [32:38](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1958s) 从核心连接到三大分析型新支柱：反欺诈、实时信用与银行支付 在过去的两年里，Plaid 的产品战略经历了一次关键进化——几乎所有新发布的产品都属于“分析型产品”（Analytics Products），其底层逻辑是充分复用并盘活流经 Plaid 平台的庞大聚合数据集。Zach Perret 将其划分为三大核心新领域： 第一大领域是**反欺诈（Anti-Fraud）**。Plaid 即将全面推出反欺诈产品系列的升级版。以往各家金融 App 只能闭门防守，而 Plaid 能够基于跨应用的宏观视角识别欺诈模式。如果一个恶意用户在某一家 App 内实施了欺诈并被标记，Plaid 就能将这一欺诈信号联邦化并同步给全生态；或者，当系统监测到某个身份在过去 12 小时内接连申请了 4 笔贷款，这会立即触发风险警报。这种基于全网行为模式构建的算法级反欺诈工具，是市面上其他缺乏跨平台数据的玩家根本无法打造的。 第二大领域是**实时信用评估体系（Real-Time Credit System）**。现行的 FICO 信用分体系诞生于 20 世纪 80 年代末至 90 年代初，几十年来一直是消费信贷的核心基准。Zach 认为 FICO 本身很有价值，但如果能辅以实时数据进行扩充，就能更精准地刻画借款人的还款能力。他向主持人 Ben 打趣道：“如果你明天换了一份薪水暴涨 5 倍的新工作——虽然我非常希望你继续做现在的播客——但从财务角度看，你的自由现金流大幅提升、偿债能力变强，理应获得更高的信用评分，然而传统的 FICO 很难实时捕捉这些变化。”同样地，如果用户换租到了更便宜的公寓、整体开支下降，其违约风险也会降低。Plaid 能实时分析用户的月度、季度与年度收支结构，从而在其上构建动态的信用评分层。 第三大领域则是**银行支付分析（Payments Analytics）**。Plaid 正在持续深耕银行直连账户支付（Account-to-Account），利用数据分析预测结算风险与交易成功率，使银行间直接转账变得像刷信用卡一样稳定、可预测且安全。 ```mermaid flowchart TD accTitle: Plaid从底层连接到数据分析产品的架构演变 accDescr: 展示Plaid如何基于广泛覆盖的银行连接底座，衍生出反欺诈、实时信用评估和银行支付分析三大核心产品。 subgraph Base["底层全景数据网络"] A["覆盖全美超 50% 银行账户用户"] --> D["多维聚合数据集<br/>跨App关联/实时收支/身份验证"] end subgraph Products["三大分析型衍生支柱"] D --> P1["反欺诈系统 (Anti-Fraud)<br/>跨应用异常识别与欺诈信号联邦化"] D --> P2["实时信用体系 (Real-Time Credit)<br/>突破静态 FICO，动态评估收支与偿债力"] D --> P3["支付分析 (Payments Analytics)<br/>让银行直连转账更可靠可预测"] end ``` ### [34:25](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2065s) 触达半数美国银行账户：Plaid 的独特全景数据优势 主持人在探讨中提到了一组极具震撼力的数据：**全美拥有银行账户的人群中，超过一半（50% 以上）的用户曾通过 Plaid 完成过账户连接绑定**。这种跨越各类金融工具的广度，构建起了独特的竞争壁垒。 Ben 指出，即便是像摩根大通（JPMorgan Chase）这样体量庞大的顶尖金融机构，其视线也仅限于自身围墙花园（Walled Garden）内部的客户数据；而类似 Visa 这样的全球信用卡巨头，看到的也只是用户在 Visa 刷卡网络内的消费轨迹。相比之下，Plaid 几乎俯瞰着所有人在整个消费金融生态中的完整全景图。 Zach 坦诚地厘清了其中的差异：Plaid 在单点客户信息的数据深度上绝无法比肩摩根大通，摩根大通的客户会在官方应用中沉淀大量资产、频繁交互，银行因此能为自家客户打造极致的服务体验。Plaid 的真正王牌在于**数据的“横向聚合性与独特性”**： - 能够掌握用户究竟跨平台绑定了哪些金融应用； - 能够洞察用户在不同场景应用间的资金流转与关联动作； - 能够将多方银行账户数据与政府颁发的官方身份证明（Government-issued ID）相互印证并形成交叉验证。 Plaid 产品研发的核心逻辑，并不单单建立在“数据量大”的基础上，而是聚焦于“将未曾被组合过的数据维度聚合起来”，用全新的视角解决反欺诈和信贷评估中以往无法被量化的难题。 ### [36:10](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2170s) 第二增长曲线的爆发：新业务贡献超 20% ARR 且年增 93% 对话切入了一个商业史上的经典命题：许多千亿甚至万亿美元市值的超级巨头，往往都是“第一个点子极其惊艳，后续的跨界尝试大多成了陪衬”。主持人举了台积电（TSMC）当年尝试涉足太阳能电池等领域最终又回归集成电路制造的例子，借此发问：Plaid 的核心“银行账户连接”（Bank Linking）业务已经如此性感，这些所谓的数据分析新业务到底是顺应趋势的真实突破，还是仅仅属于锦上添花？ Zach 强调，账户连接底座依然有巨大的增长空间，Plaid 正在持续追加投资，但新业务绝非闭门造车，而是**完全由客户需求倒逼催生出来的产物**。许多客户曾试图基于导入的流水交易数据自行研发信用评分和风控算法，但受限于自身单点业务的样本量不足，始终无法训练出高精度模型。甚至有客户直接找上门询问：“你们能不能帮我们把这部分分析做了？或者我们直接派一名工程师坐到你们的数据库前，帮我们写几条查询语句？”（Zach 笑着澄清，Plaid 坚决拒绝了派人直接查库的离谱请求）。正因为客户有着强烈的计算需求却缺乏跨机构的数据规模，Plaid 才顺理成章地将这些实时数据分析标准化为产品。 主持人随后直接援引了 Plaid 年度股东信中令人惊艳的硬核数据打破了 Zach 的谦虚：**反欺诈、实时信用和支付分析这三项新业务，在 2024 年已经贡献了公司超过 20% 的 ARR（年度经常性收入），并且正在以每年 93% 的年复合增长率（CAGR）疯狂狂飙！** 这正是成熟核心业务孕育第二增长曲线时最梦寐以求的增长曲线。 面对这一成绩，Zach 表现得十分克制，并分享了自己的管理哲学。他提到自己身后书架上的一本书——《欣喜但不满足》（*Pleased but Not Satisfied*）： - 对当前取得的业务增速和进展感到欣喜（Pleased）； - 但面对信贷评分、支付与反欺诈等千亿级别的深水区市场，Plaid 目前所占的份额依然只是冰山一角，因此绝不满足（Not Satisfied），必须保持持续开拓的紧迫感。 ### [38:45](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2325s) 估值过山车与基本面管理：从 50 亿到 135 亿再到 60 亿 话题随后转向了处于剧烈周期波动中的企业管理与估值波动。金融科技行业天然兼具双重周期属性：一方面，下游客户的应用活跃度会随宏观经济冷热而剧烈起伏；另一方面，一级市场本身的资金供给和估值倍数（Valuation Multiples）也在大起大落。 Plaid 经历了极其戏剧化的估值路径： 1. **2020 年**：正式签署并购协议，拟以约 **50 亿美元** 的价格出售给 Visa； 2. **2021 年**：收购因监管受阻告吹后，正逢零利率政策（ZIRP）与疫情线上金融爆发，Plaid 逆势以高达约 **135 亿美元** 的估值完成了天价融资； 3. **2024/2025 年**：在随后的融资中，公司估值回调至约 **60 亿美元**。 Zach 坦然复盘了这轮过山车。2021 年整个市场充斥着历史罕见的估值泡沫，加之美国司法部（DOJ）当时为了阻止 Visa 并购，在诉状公文中不遗余力地阐述“Plaid 的潜力巨大到足以颠覆并统治全球支付金融体系”——Zach 戏称这堪称“公司有史以来效果最好的一波免费商业广告”，构成了最不可思议的融资大环境。 而当估值回落到 60 亿美元时，宏观市场的估值乘数已经彻底变天。但 Zach 指出，此时的 Plaid 在底层基本面上比 2021 年要强健得多： - 实现了业务的高速扩张并达成收支平衡（Break-even）； - 核心客户群体更加巩固，且新业务正以 93% 的增速复合演进； - 2021 年向投资人描绘的商业蓝图，如今在 60 亿估值时已全部兑现为实打实的财务数字。 在内部管理上，面对估值的剧烈震荡与员工流动性诉求，Zach 始终坚持向团队灌输清晰的认知：“估值会随着外部资本市场的倍数周期来回波动。我们固然关心估值，但我们唯一能掌控的，是预设目标的完成度、业务指标的达成以及为客户创造的实际价值。只要把基本面持续做好，长期来看，市场终究会给出一个客观准确的称重。” ### [41:00](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2460s) 估值重置与股权激励：告别期权倒挂，在下行周期中重构透明度 风险投资的理想剧本往往建立在估值不断攀升的单向预期之上：只要估值一路向上，投资人账面回报喜人，员工握有的股权激励水涨船高，整套体系便运转如常。然而，当市场周期剧烈转向、估值发生重置时，许多初创公司往往会陷入期权深度倒挂（underwater options）、团队士气受挫的泥潭。谈及 Plaid 如何跨越这一难关，Zach Perret 解释称，Plaid 在制度设计上早早就避开了传统期权，而是全面采用受限股票单位（RSUs）。由于不存在行权价（strike price）的高门槛，估值下调并未引发复杂的税务或期权失效难题，这在机制上为员工团队筑起了一道防线。 面对估值的下调，投资人难免会有失落情绪，但得益于历次融资中极其干净的标准条款（clean terms），投资人依然坚定地留在牌桌上。事实上，在最近一轮融资中，许多现有老股东选择继续追加投资，认定公司未来仍有广阔的增长空间。Zach 直言，风投机构本身同样身处剧烈波动的宏观环境之中，对此早有心理预期。更重要的是，Plaid 始终坚持高度透明的沟通机制，在 2021 年高估值融资与当下之间，管理层已经在内部多次主动下调了内部估值（409A valuation）。虽然经历这一估值重置绝非一段轻松愉快的历程，但随着尘埃落定，管理团队终于能够卸下包袱，全神贯注于业务增长本身。主持人感叹，当外界死死锚定在那个下滑的名义估值数字上，而创始人深知当下的基本面与内在价值实际上处于历史最强阶段时，这种认知脱节难免令人感到抓狂。 - **RSU 激励架构优势**：规避了传统员工期权因估值下跌导致“行权价高于公允价”的倒挂窘境，稳定了核心团队的长期激励。 - **老股东持续加码**：在条款干净的底层架构支撑下，多数原有投资人选择在重置轮中继续跟投。 - **内部主动去泡沫**：在 2021 年高点之后，公司并未掩盖市场变化，而是数次主动下调内部估值，为组织心理预期提前着陆。 --- ### [42:40](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2560s) 周期性的辩证法：既要逆周期产品缓冲，也要坦然拥抱市场起伏 金融科技企业不可避免地与宏观金融市场深度绑定。面对行业的强周期性，创业者究竟该奋力反抗还是顺水推舟？Zach 坦言，Plaid 的策略是“双管齐下”。一方面，公司积极研发并推出非周期性（non-cyclical）或逆周期（countercyclical）的产品矩阵，随着公司体量不断扩大，力求将外部周期的冲击降至最低；但另一方面，他们也坦然拥抱这种周期性，将其视为金融基础设施业务与生俱来的现实。 Zach 形象地解释道，他发自内心地希望自己的信贷客户（lending customers）能够呈爆发式增长，即便信贷业务的狂飙会反过来放大整体营收的周期波动性，他也甘之如饴——因为这恰恰证明客户取得了巨大的商业成功，在 Plaid 的业务版图中占据了极其庞大的份额。对于这类无法彻底抹平的宏观因子，Plaid 选择将其写入每一封股东信中，让所有股东建立起清晰且理性的预期。这种对增长驱动因子的绝对透明，也是在为未来走向公开资本市场奠定基石。正如 Zach 所言：“我们能够牢牢掌控业务的大量基本面，但我们无法预测宏观经济将如何冲击我们；我们唯一能做的是充分准备并敏捷应对，而非妄谈预测。” --- ### [43:47](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2627s) 银行态度的三次破冰：从抗拒上云到追随摩根大通与 Robinhood 在创业早期，Plaid 几乎瞬间就在开发者和终端用户端找到了强烈的 Product-Market Fit（PMF）——普通用户极其渴望授权账户以体验炫酷的新型金融 App。然而在另一端，银行的态度却曾是一座难以逾越的高山，双方早期的互动宛如“推石上山”般沉重。银行究竟是在哪个节点突然转变了态度，开始认可开放金融的未来，甚至主动伸手合作？Zach 回顾了推动传统银行业心态转变的三大关键历史拐点： ```mermaid flowchart TD accTitle: 银行业拥抱金融科技的三大历史拐点 accDescr: 展示推动传统银行从抗拒技术转向主动合作的三个关键事件驱动节点 subgraph Catalyst1["催化剂一：云基础设施认可"] A["2014-2015 AWS GovCloud 推出"] --> B["政府机构上云破冰"] B --> C["银行破除顾虑：'政府能上云，我们也能上云'"] end subgraph Catalyst2["催化剂二：领头羊军备竞赛"] D["摩根大通 Tech 预算连年激增"] --> E["从 20 亿、40 亿飙升至 170 亿美元"] E --> F["同行恐慌跟进：倒逼全行业重金砸向数字化"] end subgraph Catalyst3["催化剂三：Fintech 巨头颠覆威胁"] G["Robinhood 等新型平台极速崛起"] --> H["老牌机构被迫应战：嘉信理财取消交易佣金"] H --> I["银行全面转向：采购 Plaid 防欺诈/承销/获客工具"] end C --> G F --> G ``` 首先是大约在 2014 至 2015 年间推出的 AWS GovCloud（政府云）。当对安全极其严苛的联邦政府宣布将数据搬迁至云端时，保守的银行业如梦初醒：如果连政府都能放行云端架构，银行自身的数据上云便不再是不可逾越的红线。其次，行业领头羊摩根大通（JPMorgan Chase）掀起了前所未有的科技军备竞赛，其首席执行官公开宣布的技术预算从 20 亿美元、40 亿美元一路狂飙至惊人的 170 亿美元。面对这一天量数字，全美每一家传统银行都在自问“我们该投入多少、能换回什么”，数字化转型由此从选配变成了生存必配。 第三个决定性推力来自以 Robinhood 为代表的新一代金融科技公司的野蛮生长。这些新兴平台凭借极佳的用户体验蚕食了大量传统银行的年轻客群，并收获了令人瞩目的高估值。传统巨头不得不紧急反击：最典型的标志便是嘉信理财（Charles Schwab）被迫全面取消股票交易佣金——Robinhood 用“零佣金”重塑了规则，逼迫全行业彻底告别固定交易手续费时代。老牌银行在震惊之余开始经历“先被迫应对、再全面抄袭（reacting and then copying）”的阶段，最终选择大举拥抱金融科技工具，将其深度集成到自身获客、反欺诈拦截、信贷承销风控以及内部运营提效等各项基础设施中。 - **AWS GovCloud 效应**：彻底打消了银行业对数据托管在第三方云设施中的安全合规顾虑。 - **摩根大通的 170 亿美元技术支出**：树立了金融行业科技资本支出的绝对标杆，引发了全行业的追随效应。 - **佣金革命的连锁反应**：Robinhood 迫使嘉信理财等老牌经纪商抹除交易佣金，促使银行意识到不进行技术升级就会面临失去下一代客户的生死危机。 --- ### [46:06](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2766s) 成分品牌与三边信任：在 15 秒的极简体验中传递安全感 针对主持人将 Plaid 比作“1990 年代末至 2000 年代初的微软”——即必须同时在 C 端消费者与庞大的 B 端金融企业之间建立顶级品牌声誉，Zach 欣然笑称“任何将 Plaid 与微软相提并论的评价都是能得到的最高赞誉”。在他看来，Plaid 实际上在精密构建一个复杂的“三边品牌”： 1. **开发者（Developers）**：公司的立足之本，无论工程师身在何处，Plaid 必须保持开发者首选的技术极客底色； 2. **金融机构（Financial Institutions）**：它们不仅是关键合作伙伴和数据源头，本身也是核心采购客户； 3. **终端消费者（Consumers）**：这也是最难攻克的一环。 消费者端的难点在于 Plaid 所占有的界面极其微小（very thin user experience）。Plaid 的弹窗通常只在第三方 App 调起绑卡时出现，用户仅停留 15 到 30 秒，而产品优化的终极目标竟然是“尽可能把这个交互时间压缩得越短越好”。Visa 或万事达卡尚且能在实体卡片背面印上实体 Logo，而 Plaid 却必须在转瞬即逝的极短时间内让用户感受到两件事：“这玩意刚刚顺畅搞定了（oh that just worked）”以及“这套工具足够安全值得信赖”。为此，Plaid 打造了多套让用户随时查看个人数据流向与授权管理的异步安全工具。Zach 明确将 Plaid 定位为一种“成分品牌（ingredient brand）”：正如芯片之于电脑，Plaid 旨在为 Venmo 等客户应用提供信任与极简背书，但永远绝不越俎代庖去掠夺客户自身的品牌光芒。 --- ### [48:31](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2911s) 解构 7 Powers：为什么银行开放 API 无法撼动 Plaid 的网络效应？ 既然如今各大主流银行已经普遍开放了规范且现代化的 API 接口，那么创业者是否能在明天直接抄袭出一个“高仿版 Plaid”？结合知名商业战略理论《7 Powers》（七大力量），Zach 从底层防御壁垒的角度给出了剖析。他指出，首要且最直观的护城河正是**双边与多边网络效应（Network Effects）**。 在商业本质上，Plaid 提供的是“入驻即服务（Onboarding as a Service）”。对于像 Robinhood 这样的应用而言，最攸关生死的指标就是注册转化率（conversion rate）：一旦意向用户在绑卡或身份校验环节遇阻流失，客户平台将承受极大的获客损失。而当一名用户已经在其他 App 中使用过 Plaid 时，Plaid 便能凭借先验数据提供无感且极速的自动填入与身份复验能力。见识过的用户越多，Plaid 在下一次开户流程中的转化率溢价就越高，这种以海量用户沉淀为基础的入驻成功率是单一银行 API 绝无法比拟的。紧随其后的则是**后端数据网络效应（Data network effects）**，跨越无数机构、场景与数以亿计用户的大规模数据网络，赋予了 Plaid 在风控预测和金融产品创新上无可替代的差异化壁垒。 ### [49:32](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2972s) 品牌护城河与跨越代际的信任 Zach Perret 谈到，Plaid 业务的核心根基在于利用规模与网络效应打造别人无法复制的分析能力与信贷产品，比如独一无二的反欺诈分析引擎（fraud analytics）和信贷风控方案。谈及“品牌”是否能成为长期壁垒时，Zach 自嘲地引用了 Hamilton Helmer 在《7 Powers（商业七力法则）》中的观点——品牌往往是七种力量中最脆弱的一种。然而，Acquired 主持人对此提出了反驳，认为品牌在金融科技领域尤为关键。主持人指出，Plaid 实际上完成了一项极为艰难的品牌演进：最初它赋能的是一批让老一辈觉得“有点危险”的年轻人尝鲜应用（例如千禧一代当年毫无防备地将银行账户凭证直接输入 Venmo），而如今，Plaid 反而成了安全和可靠的代名词。 Zach 分享了一个生动的家庭趣事来印证这种跨代际的转变。几年前，他快 70 岁的母亲给他打电话，咨询朋友推荐的一款账单管理软件 Rocket Money（原 Truebill）是否安全。Zach 随口回答：“妈，页面中间写着‘Powered by Plaid’呢，你没认出来吗？”母亲恍然大悟，随后不仅自己放心使用，还向身边所有的同龄好友疯狂安利。紧接着，Plaid 观测到了来自此前从未触达过的老龄客群的爆发式增长。这表明 Plaid 的信任背书不仅深入开发者社区，更渗透到了传统非数字原生用户的意识中。 ### [51:05](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=3065s) “快速迭代”算不算竞争壁垒？ 谈及竞争优势，Zach 提出了一个在经典商业理论之外但对创业公司至关重要的要素：极其迅捷的行动速度与产品迭代能力。他承认在 Hamilton Helmer 的框架下，单纯的“快”也许算不上持久的竞争优势，但他深信速度本身至关重要。因此，“快速行动”（Move Fast）是 Plaid 的核心企业价值观，团队始终全力以赴争做市场第一、最高质量和最敏捷的产品交付者。 主持人开玩笑说，戴上“假装有 MBA 学位”的眼镜来看，这充其量属于“卓越运营”（Operational Excellence），而不是战略，因为人人都想快、人人都在强调快速交付的文化，只是 Plaid 自信执行得更好而已。Zach 则笑着搬出《7 Powers》里的概念回应：在最顶尖的企业里，这会被称作“过程权力”（Process Power），尽管 Hamilton 通常认为科技公司不该轻言自己拥有这种壁垒，但 Plaid 依然全力以此为目标。 ```mermaid flowchart LR accTitle: Plaid 核心竞争力演进 accDescr: 展示从早期的速度与执行力到规模形成后的网络及过程权力的演进路径。 subgraph Execution["早期执行层"] A["快速行动 (Move Fast)"] --> B["高频迭代与暴力交付"] end subgraph Scaling["规模化与产品化"] B --> C["连接 12,000+ 家银行数据"] C --> D["独家反欺诈与信贷分析能力"] end subgraph Advantage["护城河壁垒"] D --> E["品牌信任 (Powered by Plaid)"] D --> F["过程权力 (Process Power)"] end ``` ### [52:00](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=3120s) 从 2013 年的 REST API 到 AI 时代的 MCP 与 Vibe Coding 回顾 Plaid 的创业起点，Zach 强调他们始终将开发者视为企业内部的核心“变革者”（changemaker）。在今天，“面向开发者销售”早已是人尽皆知的套路，但在 2013 年，除了 Twilio 喊出著名的“Ask your developer”口号外，金融领域几乎没人这么做。当时传统的金融基础设施巨头将 API 文档层层加密、极难获取，架构还停留在古老的 SOAP 协议；而 Plaid 只是做了在现代软件工程看来“理所当然”的决策——把清晰易懂的文档完全公开在线上，并提供简洁现代的 REST 架构。正是这些在当时看来处于前沿的基本操作，引爆了开发者的热情。 转眼到了生成式 AI 大爆发的今天，软件开发的底层范式已经发生了翻天覆地的巨变。写代码的门槛被彻底打破，如今涌现出了大量所谓的“氛围编码者”（Vibe Coders）和泛开发者人群。为了适应这一浪潮，Plaid 迅速推出了 Plaid MCP（Model Context Protocol）服务器，开发者可以直接在 Cursor、Windsurf 等新一代 AI 编程环境中无缝集成 Plaid。Zach 指出，软件的分发与发现逻辑正被重写：未来工程师或 AI Agent 在 Cursor 聊天框中提问“我想做个银行转账或者账户连接应该用什么”时，AI 不仅应该直接推荐 Plaid，还能顺带引导用户“提供两三个参数，我来帮你一键配置完毕”。 - **2013 年的降维打击**：竞争对手采用 SOAP 且隐藏文档；Plaid 采用公开文档 + REST 架构。 - **2025 年的开发范式**：通过 Model Context Protocol（MCP）服务器直连 Cursor / Windsurf，迎接由 AI 驱动的代码编写与工具发现。 ### [54:03](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=3243s) 汗水红利的终结：从“堆人力”转向“高决策杠杆” 在谈到“曾经坚信不疑但如今彻底改变的观念”时，Zach 坦言，当今工作形态最剧烈的一个变化在于：纯粹靠勤奋硬拼（hard work）所带来的回报已经大不如前。在 Plaid 早期，团队的核心竞争优势正是解决那些被他称作“苦力型工程”（grinder problems）的能力——凭借更凶狠的执行力和蛮力（brute force）去堆工作量。Plaid 团队当年就是靠建立对应的激励、团队与文化，硬生生接入了多达 12,000 家银行。这项工程在技术上并无玄机，核心就是靠极高强度的工作量去硬啃。 但在现代软件开发环境下，单纯的劳动量已不再是护城河。AI Agent 正在接管甚至独立完成绝大部分常规苦活累活，产品经理、设计师和工程师的核心价值急剧转移到了“决策质量”上。战略眼光与审慎决策的回报率呈指数级上升，而单纯的体力劳动回报率正在断崖式下跌。 - **苦力型工程（Grinder Problems）**：早期凭蛮力接入 12,000 家银行是战略优势；如今同等工作量可被 AI Agent 极速摊薄。 - **决策杠杆**：未来竞争比拼的是对架构、产品定位与核心战略的深度判断力，而非单纯的代码行数或工作时长。 ### [55:51](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=3351s) 浪费的二十岁？劳动转化为资本的时代变局 主持人与 Zach 深入探讨了工具杠杆对整个世代的冲击。即便假设所有人未来都能获得同等的 100 倍 AI 生产力放大效应，在目前技术普及尚未达到终极有效前沿（efficient frontier）的过渡期，谁能率先高效运用这些新工具，谁就能建立巨大的短期不对称优势。主持人感慨道，自己从 18 岁起连续十年每周工作 80 多个小时，那代人正是靠极致的拼命才在世界上立足，如果放在今天，那种缺乏高阶战略的死磕可能会被无情降维淘汰。 Zach 爆料称，他一两周前和曾经的前联合创始人（如今已重新创业）及一位朋友聚餐，席间大家半开玩笑地感慨：“我们基本上全把自己的 20 多岁浪费掉了。”当年他们投入海量青春汗水所做出的繁琐工作，今天一款软件工具转瞬之间就能搞定。主持人随即调侃：“但不可否认的是，你们成功地把当年的体力和汗水转化成了 Plaid 这家伟大公司的股权与资本！”Zach 笑称自己对结果当然非常满意，只是惊叹于当年的工作方式是多么低效。如今新创立的 AI 原生公司往往在短短数月内就能冲上 1 亿美元的 ARR（年化经常性收入），过去那种依靠漫长年岁与巨大劳动量才能换取的规模壁垒，在当前时代正在被彻底粉碎。 在访谈结尾，Zach 分享了关注与联系他的途径： - 官方网站：`plaid.com`，社交平台 X：`@plaid` - Zach Perret 个人 X 账号：`@zackperret`，欢迎开发者和用户直接私信反馈产品建议。 - 节目最后，主持人提醒听众，Acquired 即将于 7 月 15 日在纽约无线电城音乐厅（Radio City Music Hall）举办现场专场活动，门票已近售罄，购票链接为 `acquired.fm/nyc`。 ### [58:39](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=3519s) 结语与听众互动渠道（邮件列表与 Slack 社区） 节目尾声，主持人向听众介绍了保持连接与参与后续互动的官方渠道。如果你希望在每期新节目上线时第一时间收到通知，可以订阅节目的专属邮件列表（acquired.fm/email）。邮件通讯不仅用于推送新内容，主持人还会在这里悄悄剧透下一期节目的线索提示，并分享对往期节目的勘误、最新进展更新以及团队在制作过程中新学到的有趣细节和幕后花絮。 此外，如果听众想要围绕本次与 Plaid 首席执行官 Zach Perret 的深度对谈展开讨论，或是交流其他商业与金融科技话题，可以前往 acquired.fm/slack 加入官方 Slack 社区，与广大听众一起互动交流。最后，两位主持人向大家亲切道别：“Listeners, we'll see you next time. We'll see you next time.”（听众朋友们，我们下期再见！） - **邮件列表（acquired.fm/email）**：提供新单集上线提醒、下期选题预告线索、往期勘误与动态追踪，以及未在正片中展开的幕后花絮。 - **Slack 讨论社群（acquired.fm/slack）**：供听众深入探讨 Zach Perret 访谈中的核心观点，并与整个商业爱好者社群实时交流。 ## Notable quotes > [01:24](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=84s) 但事实上这趟过山车从来没有终点，它一直在往前开，而且波动的振幅还越来越大 — Zach Perret > [02:30](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=150s) 你看着台下的人群……能同时看到愤怒、沮丧、震惊，还有喜悦。有些人甚至在哭泣，我一时都分不清他们脸上的泪水到底是因为高兴还是因为难过。 — Zach Perret > [04:57](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=297s) 我们从来没有认为彼此的产品存在直接竞争。如果这场交易的底层逻辑本身是反竞争的，我们当初根本就不会选择推进这项交易。 — Zach Perret > [07:31](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=451s) 别担心，我们合同里有‘大流行条款’（Pandemic Clause）。 — 外部律师 > [12:15](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=735s) 我知道很多人已经在脑海里盘算着怎么花这笔收购款了，甚至打算去买房——但我必须很抱歉地告诉大家，房子买不成了，这笔钱不会按时到账了。 — Zach Perret > [15:10](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=910s) 但是 Zach，我只想告诉你，如果这笔收购最终黄了，你们不仅不会完蛋，而且很可能会变得极其伟大。 — Scott Cook > [15:10](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=910s) 没错，我这辈子经历过最棒的事情，就是当年微软试图收购 Intuit 时，那笔交易被反垄断机构彻底叫停了。 — Scott Cook > [17:25](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1045s) 根本没人想用一个整天教育自己‘少花钱’的应用。 — Zach Perret > [20:18](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=1218s) 听着伙计们，你们做的那些 App 确实挺蠢的，但我对你们搭建的后端底层架构非常感兴趣，我们愿意出钱授权使用它。 — Venmo 工程负责人 > [38:45](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2325s) 估值会随着外部资本市场的倍数周期来回波动。我们固然关心估值，但我们唯一能掌控的，是预设目标的完成度、业务指标的达成以及为客户创造的实际价值。只要把基本面持续做好，长期来看，市场终究会给出一个客观准确的称重。 — Zach Perret > [42:40](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2560s) 我们能够牢牢掌控业务的大量基本面，但我们无法预测宏观经济将如何冲击我们；我们唯一能做的是充分准备并敏捷应对，而非妄谈预测。 — Zach Perret > [49:32](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=2972s) 妈，页面中间写着‘Powered by Plaid’呢，你没认出来吗？ — Zach Perret > [55:51](https://www.youtube.com/watch?v=IIsP8xFGqKk&t=3351s) 我们基本上全把自己的 20 多岁浪费掉了。 — Zach Perret ## Mentioned Plaid, Zach Perret, Visa, 美国司法部 (DOJ), 摩根大通 (JPMorgan Chase), J.P. Morgan Payments, Acquired, Ben Gilbert, Altimeter, Scott Cook, Intuit, 微软 (Microsoft), Venmo, Braintree, PayPal, eBay, Robinhood, Coinbase, Mint, Stripe, Rocket Money, 富国银行 (Wells Fargo), 消费者金融保护局 (CFPB), 多德-弗兰克法案 (Dodd-Frank Act), 贝恩咨询 (Bain & Company), FICO, AWS GovCloud, 嘉信理财 (Charles Schwab), 万事达卡 (Mastercard), Hamilton Helmer, 《7 Powers》, 《欣喜但不满足》(Pleased but Not Satisfied), Twilio, Cursor, Windsurf, Model Context Protocol (MCP), Atlassian ## Video description We sit down with Zach Perret, CEO of Plaid, to discuss the remarkable journey of Plaid and the broader fintech landscape over the past several years. Zach takes us blow-by-blow through journey of almost getting acquired by Visa, the challenges faced during the COVID-19 pandemic… which quickly reversed with ZIRP tailwinds, and how Plaid navigated the volatile market conditions to build a diversified business. We explore the company’s strategic pivots, including their expansion into analytics for fraud detection, alternative credit systems, and bank payments. If you’ve ever wondered “how do you turn from one simple product into a more durable business?” this episode is for you. *More Acquired:* - Get email updates https://www.acquired.fm/email and vote on future episodes! - Join the Slack http://acquired.fm/slack - Check out the latest swag in the ACQ Merch Store https://www.acquired.fm/store! _Note: Acquired hosts and guests may hold assets discussed in this episode. This podcast is not investment advice, and is intended for informational and entertainment purposes only. You should do your own research and make your own independent decisions when considering any financial transactions._ [Source](https://www.youtube.com/watch?v=IIsP8xFGqKk)

[1:24] Zach Perret: 但事实上这趟过山车从来没有终点，它一直在往前开，而且波动的振幅还越来越大 [2:30] Zach Perret: 你看着台下的人群……能同时看到愤怒、沮丧、震惊，还有喜悦。有些人甚至在哭泣，我一时都分不清他们脸上的泪水到底是因为高兴还是因为难过。 [4:57] Zach Perret: 我们从来没有认为彼此的产品存在直接竞争。如果这场交易的底层逻辑本身是反竞争的，我们当初根本就不会选择推进这项交易。 [7:31] 外部律师: 别担心，我们合同里有‘大流行条款’（Pandemic Clause）。 [12:15] Zach Perret: 我知道很多人已经在脑海里盘算着怎么花这笔收购款了，甚至打算去买房——但我必须很抱歉地告诉大家，房子买不成了，这笔钱不会按时到账了。 [15:10] Scott Cook: 但是 Zach，我只想告诉你，如果这笔收购最终黄了，你们不仅不会完蛋，而且很可能会变得极其伟大。 [15:10] Scott Cook: 没错，我这辈子经历过最棒的事情，就是当年微软试图收购 Intuit 时，那笔交易被反垄断机构彻底叫停了。 [17:25] Zach Perret: 根本没人想用一个整天教育自己‘少花钱’的应用。 [20:18] Venmo 工程负责人: 听着伙计们，你们做的那些 App 确实挺蠢的，但我对你们搭建的后端底层架构非常感兴趣，我们愿意出钱授权使用它。 [38:45] Zach Perret: 估值会随着外部资本市场的倍数周期来回波动。我们固然关心估值，但我们唯一能掌控的，是预设目标的完成度、业务指标的达成以及为客户创造的实际价值。只要把基本面持续做好，长期来看，市场终究会给出一个客观准确的称重。 [42:40] Zach Perret: 我们能够牢牢掌控业务的大量基本面，但我们无法预测宏观经济将如何冲击我们；我们唯一能做的是充分准备并敏捷应对，而非妄谈预测。 [49:32] Zach Perret: 妈，页面中间写着‘Powered by Plaid’呢，你没认出来吗？ [55:51] Zach Perret: 我们基本上全把自己的 20 多岁浪费掉了。

[17:05] Zach Perret: free equals good growth, but free is very expensive. [20:14] Venmo head of engineering: pretty dumb [15:10] Scott Cook: if the acquisition fails, you're going to be fine and actually you might be great. [10:06] Zach Perret: Hey, all of you people that thought you were going to get a bunch of money from this acquisition, we're now not going to do it. [27:10] Zach Perret: get greedy when others are greedy. [29:26] Zach Perret: replatformed our strategy [29:26] Zach Perret: 12 months and one day [38:26] Department of Justice: take over the world [41:03] Zach Perret: laser focused on growth. [43:47] Traditional banks: if the government can put data in the cloud then maybe I can put data in the cloud. [46:05] Zach Perret: the greatest compliment that you could ever say about our business. [46:05] Consumers: oh that just worked [46:05] Consumers: oh that was simple [52:00] Twilio: ask your developer [57:07] Zach Perret: all wasted our 20s

[0:00] Hello acquired listeners. Today's [0:02] episode is with Plaid CEO Zack Pereé. [0:06] You are probably familiar with Plaid as [0:08] the bank linking service that lets you [0:09] link your bank account to other bank [0:11] accounts and fintech apps. They have had [0:14] a pretty insane journey where in 2020 [0:17] they agreed to be purchased by Visa for [0:19] $5.3 billion. Plaid then terminated this [0:23] deal after the US Department of Justice [0:24] filed an antitrust lawsuit to block it. [0:27] And the few years since then have been [0:29] pretty wild, too. They were valued as [0:31] high as $14 billion, then a recent [0:33] fundraising round of $6 billion. So, [0:36] more than the Visa acquisition, but of [0:37] course lower than the sky-high valuation [0:39] in between. Today's conversation with [0:41] Zach is about weathering all of that and [0:44] about Plaid's actual business they were [0:46] building along the way, growing, [0:48] diversifying, and finding ways to [0:50] leverage all the data that they have to [0:52] build a better and more modern finance [0:54] ecosystem. Well, one quick announcement [0:57] before we start. As many of you know, we [0:59] are doing a massive live show at the [1:01] 6,000 seat Radio City Music Hall in New [1:03] York City on July 15th with our friends [1:06] at JP Morgan Payments. There are just a [1:08] few seats left, so make sure to get [1:10] yours before they are gone at [1:14] acquired.fm/nyc. The uh evening is going [1:16] to be awesome. We can't wait to see you [1:18] there July 15th. That's [1:21] acquired.fm/nyc. So, please enjoy our [1:24] conversation with Zach Pereé, the [1:26] co-founder and CEO of Plaid. Zach, [1:29] welcome to the show. Thank you so much [1:31] for having me. I'm honored to get to do [1:33] this. I'm I'm a big fan of everything [1:34] acquired. So, thank you. Thank you for [1:37] listening. We've enjoyed following your [1:39] journey through the wild turns of events [1:42] that have happened over the last 5 years [1:44] or so. you were uh high-flying unicorn [1:46] and then I saw some news that said you [1:48] were acquired by Visa and then I saw [1:50] some news that said you weren't and then [1:52] you raised a lot of money at a big [1:54] valuation and then you raised more money [1:57] later at a lower valuation all while [2:00] growing the business and then starting [2:02] new business lines that I bet most [2:04] people are totally not familiar with and [2:06] they just think of Plaid as isn't that [2:08] the way that I off my fintech app to [2:09] connect to a bank account. Does that [2:11] seem like a reasonable summary of your [2:13] life over the last five years? Yes, very [2:15] reasonable summary. They say that [2:16] startups are like a roller coaster. [2:18] Indeed, the ride it doesn't end. It just [2:20] keeps [2:21] going. The magnitude gets even larger. [2:24] It's been an awesome journey and a lot [2:26] of learnings and a lot of fun and a lot [2:27] of frustration along the way too. You [2:29] got your fair share of everything. Maybe [2:30] to start, take us back to January 2020 [2:36] and your partnership with Visa. catch us [2:39] up from there. For a podcast that's [2:41] called Acquired, you don't talk a lot [2:42] about acquisitions. No, this is like a [2:45] throwback episode for us. Yeah, but this [2:47] acquisition didn't happen. So [2:48] ultimately, [2:51] that is true. That is very true. All [2:52] right, so in January of 2020, I think it [2:57] was January 18th or 19th of 2020, we [2:59] announced that Visa was acquiring Plaid [3:02] for a big valuation. It was just above [3:04] $5 billion. [3:05] And this was the culmination of a lot of [3:09] back and forth over the past 3 to four [3:11] months. Visa had been an investor. They [3:14] showed interest in in in acquiring us. [3:15] There had been some other acquisition [3:17] interest as well. We ended up making the [3:19] the very difficult but at the time very [3:21] logical decision to sell to Visa. We [3:24] announced it to our team on January of [3:26] 2020. And I remember because this is the [3:28] last big all hands that we did in the [3:30] office. Our office, it's the X Atlassian [3:33] office and it has these gigantic stairs [3:35] that people can sit on. And so, you [3:37] know, it's a two-floor staircase. It's [3:39] super wide, probably like 30, 40 feet [3:41] wide, just filled with people. We were [3:43] planning this all hands. And I remember [3:45] right before the all hands, the news [3:47] leaked. And so all of our news addicted [3:50] team saw it and the others that were not [3:52] news as news addicted were like, "What's [3:54] going on? Why is there so much buzz? [3:55] What's happening?" Do you know who [3:57] leaked it? You never know. I mean, the [3:59] investment bankers, maybe like one of [4:01] the lawyers, and you know, it wasn't [4:03] leaked far in advance. It was leaked [4:05] like 20 minutes in advance. So, it was [4:07] not the worst leak in the world. And [4:08] you're probably already like really [4:09] nervous about messaging this to the [4:11] company, right? This is not helpful. [4:13] It's the second hardest all hands that [4:15] I've ever done. The first hardest I'll [4:16] tell you about in a minute. Yeah, we'll [4:17] get to that coming up. The second [4:19] hardest all hands that I've ever done. I [4:20] go there and I'm in front of everyone [4:21] and I like really practiced on on what I [4:24] was going to say and how it was going to [4:25] go and it went well and it was amazing. [4:27] You just you look at the crowd. We [4:28] probably had 350 people in this all [4:32] hands almost all of them physically [4:34] there and there was anger, frustration, [4:37] surprise, like joy. Some people were [4:40] crying. I couldn't tell if they were [4:41] happy or sad when they were crying. It's [4:42] like this totally weird thing. And then [4:43] the Visa team came in and and Facebook [4:45] and they talked about how they were all [4:46] excited about the acquisition. We ended [4:48] up closing and had tons of team [4:50] messaging so on and so forth. And it's [4:51] one of the hardest things is explaining [4:53] this this level of change management [4:54] within the organization. But we got [4:55] through all that stuff. So it worked. [4:57] Just to be like super clear for everyone [4:59] listening up until this point in the [5:01] company other than being kind of an [5:02] investor and a partner, Visa is this [5:05] company they represented the existing [5:08] financial system and you were trying to [5:10] create this new faster more tech [5:12] friendly fabric you know data fabric [5:15] around the world for finance. Is that [5:17] right? So this is one of the things that [5:18] I think is actually most misunderstood [5:21] about platinum visa. We are not direct [5:23] competitors. We were not direct [5:25] competitors. We weren't trying to issue [5:27] cards or like we build a card payment [5:29] network or things like that. In some [5:31] meta sense, I guess you could say we [5:32] produce substitutes, meaning you could [5:34] pay for something with a bank account or [5:35] you could pay for something with a [5:36] credit card. But even those in almost no [5:38] single transaction would you consider [5:40] both of those payment methods to be [5:41] viable. And so despite the fact that the [5:43] DOJ later investigated this transaction [5:46] for antitrust and alleged that Visa was [5:49] a monopolist and buying Plaid was a path [5:51] to expand their monopoly, I have never [5:53] thought that our products directly [5:54] compete. We probably wouldn't have done [5:56] the transaction if the premise of the [5:57] transaction was anti-competitive. So [5:59] certainly there were many developers who [6:01] were using Plaid as a core part of [6:03] building their apps which were intended [6:06] to be payment networks and you know [6:09] transaction rails outside the Visa [6:10] system. But yeah, I totally get what [6:12] you're saying. Yeah, that's very fair. [6:13] So like in the meta sense, we were [6:15] trying to make financial services more [6:17] efficient, better. We were trying to [6:18] create new payment methods. We were [6:19] trying to create a lot of kind of [6:20] innovation in the system. And so in that [6:22] sense, if you say Visa is the historical [6:24] system and you say plaid is the new [6:26] system or plat and all of our customers [6:27] and this this fintech ecosystem or the [6:29] the new system, then yes, we were [6:31] competitive. But specifically on any one [6:32] product, I can't say that we were [6:34] directly competitive. Makes total sense. [6:36] Why did it feel so emotional? And why to [6:40] the extent that there were people crying [6:41] and it was you know tears of sadness. [6:43] Why did they feel the sadness? It's a [6:46] huge release of pent up emotion or a lot [6:50] of people when they join a startup [6:52] certainly when you found a startup like [6:53] a lot of your identity is wrapped up in [6:54] that startup itself. A lot of your [6:56] identity is wrapped up in work and [6:58] people had worked for us for seven eight [7:00] nine years. They had a massive amount of [7:02] equity that had been issued when Plaid [7:03] was worth, I don't know, $50 million [7:05] that was now being sold for $5 billion. [7:07] You know, there's a lot of a lot of [7:09] excitement there. There's also people [7:10] that had joined us a year before and [7:12] thought that Plaid was going to go from [7:13] our current valuation to 100 times that. [7:15] Some people were excited. Some people [7:16] were disappointed. Some people were [7:18] saying, "This is going to be a great [7:19] outcome both financially and [7:20] product-wise." Some people were saying, [7:22] "I'd hoped for more." So, there was a [7:24] mix. But yeah, definitely reading the [7:25] face is one of those unique experiences [7:26] where you see so many different emotions [7:28] all at once. I bet. Okay. So then what [7:31] is life like after that? Life after that [7:33] was kind of straightforward. So you go [7:35] February, first half of March or great, [7:37] we're doing messaging. We're moving [7:39] forward with the acquisition process. [7:40] There's a lot of paperwork we have to [7:41] file. You have to submit it to all the [7:43] relevant governments and so forth. And [7:45] we had no reason to suspect antitrust [7:47] was going to be an issue because again [7:48] we didn't think that we competed with [7:50] Visa. And we had no reason to think that [7:52] anything else was going to change in the [7:53] world that was gigantic. And then co co [7:55] happens in March. We're very focused on [7:57] the team through March and into April. [7:59] Come mid April, we pulled up and thought [8:00] about what does this mean for us? Is [8:02] this transaction even going to happen [8:04] still? I called the lawyers that we'd [8:06] used and said, "Hey, what's going on [8:08] with the transaction?" And their [8:09] response was, "Don't worry, we have a [8:11] pandemic clause." And I said, "What in [8:13] the world could a pandemic clause [8:15] possibly mean?" It turns out there's [8:17] this thing called a material adverse [8:18] effect. Like a pandemic was something [8:20] that we had contemplated and I think [8:21] it's just in the form letter that like [8:22] if a global pandemic happens, you still [8:24] have to close this transaction visa. [8:25] Okay, great. So, we feel like we're [8:27] geniuses because Visa's stock price has [8:29] crashed. We have a pandemic protection [8:32] clause that says Visa still has to close [8:33] the transaction. And we did the math on [8:35] it and we said, "Well, we're going to [8:37] own a large percentage of Visa if this [8:40] transaction closes tomorrow." Whoa. It [8:43] was a stock deal, partially stock and [8:45] partially cash. It was majority cash, [8:46] but from the employee side, the [8:48] employees ended up getting a ton of [8:49] equity in Visa, which was dollar [8:52] denominated equity. And so, it was based [8:54] on Visa's most recent share price. The [8:56] math that we did was, "Oh my gosh, we're [8:57] going to own a big piece of Visa." Fast [8:59] forward a little bit more and the world [9:00] turned around. A few things happened to [9:02] our business. So, one, consumers were [9:03] stuck at home. They realized they still [9:04] needed financial services. And so, usage [9:07] of plaid powered products just caught on [9:09] fire. Also, then money was free. So, [9:12] startups being founded happened [9:14] everywhere. All of our customers got [9:15] funded. All of this growth started [9:17] happening. And that looked great. Crypto [9:19] is taking off. There's all sorts of [9:20] on-ramps and off-ramps and so many neo [9:23] banks. I mean, this is the fintech boom [9:25] time right here. Yes, exactly. Man, that [9:28] was fun. Then we fast forward going a [9:30] little further and we get, you know, our [9:32] first request from the government, which [9:34] is the first request of, hey, we want to [9:36] review this for antitrust. We'd expected [9:37] that one. Then we get a second request [9:39] from the government and that one we [9:40] didn't expect because, you know, any big [9:42] transaction, a first request, great, [9:43] they want to take a look, makes sense. [9:44] Second request, we thought that we would [9:46] get through pretty quickly because we're [9:47] really not competitive with Visa. [9:49] Certainly, that's the narrative that we [9:50] have. But it turns out there were some [9:52] bones that the DOJ wanted to pick with [9:54] Visa about things and I'm not sure that [9:56] all of them were really about plaid [9:57] even. So fast forward with that, our [9:59] business continues to boom. We get into [10:01] October, November, everything's [10:04] continuing to go really well. My exact [10:06] team, we start looking at each other [10:07] saying, "What if we hit one year?" [10:09] Because at one year, the exclusivity [10:11] lapses and we can walk away from the [10:12] deal. It's kind of incredible, too. I [10:14] mean, most companies that would have [10:16] found themselves in a situation like [10:18] yours where there's a pending deal for a [10:20] large acquisition by a big player, the [10:23] business kind of just sort of operates [10:25] on autopilot. Like it's uncommon that [10:27] the business, you know, grows like a [10:29] rocket ship during this period, right? [10:32] That is true. I think we made a couple [10:34] of intelligent, you could also call them [10:36] lucky decisions at the time where the [10:38] deal we negotiated with Visa was that we [10:39] would run fully independently and I [10:41] would be the CEO of Plaid within Visa. [10:43] And so my view was great, we're [10:45] operating the same as a startup. We're [10:46] going to push as hard as we can, as fast [10:48] as we can on everything. Some of the [10:50] things were harder while the acquisition [10:51] was pending. So hiring execs was harder [10:53] because people were like, oh, like how's [10:55] this going to work post close, but we [10:57] did still hire a lot of people. Other [10:59] than that, we were just full speed [11:01] ahead. And you know, we had this great [11:02] tailwind of what was going on with Zer [11:04] and and growth rates and such like that. [11:06] So we get to October, November thinking [11:08] that it may lapse. So we get into [11:09] December, January, like knowing that it [11:11] is probably going to lapse and at some [11:13] point in their the DOJ signaled that [11:15] they were going to sue to block. While I [11:17] actually think we would have won that [11:18] lawsuit, I think that lawsuit would have [11:19] taken 2 years for us to close the deal. [11:21] And at that point, I believed the plaid [11:23] was worth more. And frankly, a lot of [11:25] the concerns that we had about our [11:27] business, a lot of the impetus for [11:28] selling was gone. So we then hit January [11:32] of 2021 and we made the decision to walk [11:35] away. This is an interesting mental [11:36] model I hadn't really contemplated [11:37] before that if it's going to be that one [11:41] year plus another two, you're basically [11:42] locking in a price and almost giving the [11:45] free option to the acquirer and saying, [11:47] "Yeah, we'll sell it to you for the same [11:49] price two to three years from now [11:52] despite the fact that we're going to [11:53] keep growing. Our business is going to [11:54] keep improving." Why would you ever give [11:56] someone that option? Exactly. In some [11:58] cases, it does make sense because the [12:00] purchase price is so significant. [12:02] Sometimes the market craters afterwards [12:04] and and you really want to close a deal [12:05] and so forth. In our case, the market [12:07] had conspired and the business had [12:09] conspired to make it such that our [12:10] business had grown quite a lot and we [12:12] didn't feel like the price held anymore. [12:15] January of 2021, this is the actual [12:17] hardest all hands that I ever had to do. [12:19] Um, and this one was over Zoom. I was in [12:21] the office I think by myself and the [12:24] entire company was on Zoom and I [12:26] basically said, "Hey, all of you people [12:28] that thought you were going to get a [12:29] bunch of money from this acquisition, [12:30] we're now not going to do it." But on [12:31] the plus side, look at all the amazing [12:33] things that we did over the last year. [12:34] Think about all we're going to go build [12:35] together and it's going to be an awesome [12:36] future. And we've had some people [12:38] approaching us with investment offers [12:40] around that time. So I had good reason [12:41] to believe that the valuation would have [12:43] gone up. But I can't say that to the [12:45] company. You know, I'm not going to [12:46] promise something that I'm not yet sure [12:48] we're going to be able to deliver. Then [12:49] we had this whole next wave of employee [12:51] comes and this one was even harder [12:52] because people despite the fact that you [12:54] tell them not to spend the money, they [12:55] spend the money in their heads. So I had [12:57] to go to everyone and say, "Hey, look, [12:58] you can't buy that house anymore. I'm [13:00] sorry." or like basically communicating, [13:02] hey, you're not going to get that cash [13:03] value at this timeline that you expect, [13:05] but don't worry, we believe the equity [13:06] is going to be worth a lot more in the [13:07] future. We believe that this is an up [13:09] into the right story. And then fast [13:11] forward a little bit more, we did this [13:12] this fundraising round is a big step up [13:13] in valuation. So whereas we sold to Visa [13:16] for about $5 billion, the funding round [13:18] was at $13 billion. And we were able to [13:20] do some employee secondary for most [13:22] employees, which is great. So it was a [13:24] smaller proportion of their holdings at [13:25] a higher price. They couldn't maybe buy [13:27] a house, but they could certainly take [13:29] something off the table, maybe buy a car [13:30] or something like that, relieve some of [13:31] the pressure. We felt good about what we [13:34] were able to offer employees and [13:35] certainly felt great about the [13:36] trajectory and the long-term prospects [13:37] of the company independently. That was a [13:39] journey. I've learned more about [13:41] antitrust and company governance than I [13:44] ever expected to in that way. And yeah, [13:46] I hope I never have to go through that [13:47] again. How close then were you to the [13:50] altimter financing? Because that was a [13:52] big step up. What was it? 5 billion to [13:54] 14 billion at the valuation of that [13:57] round. That was pretty soon after if I [13:59] recall from the deal falling apart with [14:02] Visa. Pretty soon. Yeah. So we raised [14:04] the altimeter round was about I think it [14:06] was about 13 and a half in valuation and [14:09] that round was in April and the Visa [14:12] deal fell apart in January. Okay. So at [14:14] least this promise or this implication [14:17] that you've thrown out there to [14:18] employees of the intrinsic value of our [14:20] business is higher. someone's going to [14:22] recognize that you're going to be able [14:23] to recognize some paper value of that. [14:25] That came true reasonably quickly. It [14:27] did. We were able to show some good [14:28] growth proof points as well. So, we [14:30] constructed a good narrative and [14:31] frankly, I mean, the best narratives [14:32] come from very hard data on the on the [14:34] back end. So, we were able to talk [14:36] employees through it, but still it's [14:37] it's a jarring prospect. You expect [14:39] you're going to work for Visa, you don't [14:41] work for Visa, and and you know, we have [14:42] a certain expectation of the culture [14:43] you're going to work in, and now it's a [14:44] very different startup oriented culture. [14:46] We'd hired people that expected that [14:48] they were going to work for the startup [14:50] version of Visa but as kind of Plaid as [14:53] a subsidiary and we found that they [14:55] weren't as much of a cultural fit or [14:56] they themselves opted out when we were [14:58] long-term independent. So there there [14:59] was definitely some whiplash. We had [15:00] hired quite a lot of people actually [15:02] during the the one-year transition [15:03] because we were growing so fast and so [15:06] very hard set of employee coms and then [15:07] a lot of actions in the back end. So [15:10] funny. Usually, when a big acquisition [15:12] like this falls apart, I don't want to [15:14] say company killing, but it's usually so [15:16] demoralizing that it's hard to ever [15:18] really recover and hit new all-time [15:21] highs on any metric you want to assign [15:23] to it, valuation or revenues or anything [15:25] like that. It's a very rare scenario [15:28] where a $5 billion acquisition from a [15:32] big public company falls apart for a [15:33] startup and then the next two years [15:36] looks amazing and gang busters and like [15:38] nothing but rainbows and sunshine for [15:40] the company. I agree it's rare. I don't [15:42] think we're alone in this though. One of [15:44] the most interesting moments of the [15:46] acquisition for me was actually this [15:48] call that I got the day after the [15:49] acquisition went public. So I'd known [15:51] Scott Cook for a long time. Scott Cook [15:53] is the founder of Into It. I know the [15:55] story you're gonna tell. Yeah. He sent [15:57] me an email more or less the day of the [15:59] acquisition or the day after the [16:00] acquisition. He said, "Hey, call me. We [16:02] should chat." I called him and he says, [16:03] "Hey, Zach, congratulations on the [16:05] acquisition. I'm really impressed. I'm [16:06] really happy for you. It seems like [16:07] it'll be a great landing spot and and [16:09] it'll be great for the company, but I [16:10] just want you to know that if the [16:12] acquisition fails, you're going to be [16:13] fine and actually you might be great." [16:16] And I said, "Okay, interesting. Does [16:17] this have something to do with your [16:18] acquisition?" And he said, "Yeah, [16:20] absolutely. The best thing that ever [16:21] happened to me is that the acquisition [16:23] of Intuit by Microsoft was blocked for [16:25] antitrust. After that, Intuitit went on [16:28] to be multiundred billion dollar company [16:30] like just amazing growth trajectory and [16:32] so forth. And so his point was, look, [16:35] congratulations, but if it doesn't work, [16:36] you might be better off. In some sense, [16:38] I guess that was foroding. That's [16:40] amazing that he planted that seed in [16:41] your mind like the next day. Exactly. [16:44] And I think the day that I announced [16:46] that we were not going to do the [16:47] acquisition, I sent an email and said, [16:48] "Hey, can we get dinner? I'd like to [16:50] talk to you. [16:52] Thank you for manifesting this. And [16:54] yeah. What is your advice to make your [16:57] idea come true? Exactly. Yeah. It was [17:00] also I wanted to ask some advice on how [17:01] to do employee comps and keep people [17:03] from uh from revoling and such. Okay. [17:05] So, you entered this period of rainbows [17:07] and sunshine that ended. What did the [17:10] end of the fintech boom look like for [17:13] you guys and how did that manifest in [17:14] your business? And maybe also to give [17:16] context to that, what was plaid during [17:20] this era and that'll help ground the end [17:24] of this and then into the next era. [17:25] Perfect. So I'm going to go backwards a [17:27] little bit and explain kind of a little [17:28] bit of where we came from. So we started [17:30] working on Plaid in 2012. At that time [17:33] the core thesis was that banking and [17:35] financial services broadly was built for [17:36] a world that hadn't envisioned the [17:38] internet. And despite the fact that in [17:40] 2012 we all carried smartphones in our [17:42] pockets. We did a lot of things online. [17:44] 2012 was the year that Instagram sold to [17:45] Facebook. Instagram had had a gazillion [17:47] users. People were used to doing things [17:49] on the web and on mobile devices. Yet, [17:51] financial services required you to walk [17:53] into a bank branch and talk to a banker [17:54] for almost everything in your financial [17:56] life. So, you wanted a new checking [17:57] account, you want a loan. Oftentimes, if [17:59] you wanted a replacement debit card, you [18:00] had to go into the bank branch. And [18:02] Venmo had launched but was only a couple [18:04] years old right at this point. And on [18:06] and off ramps from Venmo probably still [18:08] very difficult. Yeah. So, in 2012, Venmo [18:12] was live but very few users and paying [18:16] someone on Venmo involved linking a [18:18] debit card or a credit card and paying a [18:20] 3% fee. You know, it's not that [18:22] attractive of a proposition to pay the [18:25] fee. Venmo, I think, for a little while [18:26] also gave away free fees, but then they [18:28] were losing a lot of money. And so, they [18:30] were in this conundrum of free equals [18:31] good growth, but free is very expensive. [18:34] I imagine for you guys back when you [18:36] were starting it's kind of this this [18:37] proof point of like hey this is what's [18:38] possible and there's so much consumer [18:40] demand for these experiences that people [18:42] are willing to eat 3% to make it happen. [18:46] That was a very good proof point. At the [18:47] time, there were very few fintech [18:50] companies. I mean, you look at in it, [18:52] they had Mint at the time. Mint was one [18:54] of the examples. You look at Venmo, [18:57] there's PayPal that's out there. And I [18:59] guess it's right around that time was [19:00] when PayPal ended up acquiring Venmo. [19:03] Brainree for for online payments. This [19:05] was pre-stripe or pre-stripe growing [19:07] really. I think it's Stripe was early at [19:09] that point. Brainree was early even even [19:11] themselves. It was early fintech. We [19:13] didn't even have the name fintech. I [19:14] think fintech came in 2016. Anyway, so [19:16] we actually started prior to Plaid, we [19:18] were building consumer apps. So we tried [19:20] to create mobile budgeting tools, [19:23] recommendations like where to spend, so [19:24] on and so forth. Turns out all of these [19:26] apps, think of a personal financial [19:27] management tool. We launched a personal [19:29] financial management tool. We [19:30] recommended to people that they spend [19:31] less money. No one wants to use an app [19:33] that tells you to spend less money. And [19:35] so there are still like 20 or 30 of them [19:38] out there today and none of them are [19:40] plaid type businesses because ultimately [19:42] they're telling people to spend less [19:43] money. Well, some of them are actually [19:45] really interesting. As a brief aside, [19:46] Rocket Money is a really cool one. It's [19:48] like a budgeting tool that ties into [19:50] your ability to get a mortgage and that [19:51] sounds super well and like Rocket has [19:53] been able to monetize this thing [19:54] massively. So I actually believe that [19:55] personal financial management is a [19:56] viable market. I'm just not a good [19:58] enough consumer product builder to come [20:00] up with the idea of how you do that. And [20:02] this was before the infrastructure even [20:05] existed in the way that it does today. [20:06] So we were spending 90% of our time [20:09] figuring out how to get the data into [20:10] the app. And then you know at that point [20:12] we didn't have the time even to build a [20:14] good app. So we ended up doing this [20:15] pivot where we dropped the consumer [20:17] side. We just focused on the [20:18] infrastructure side. It was actually [20:19] this conversation with the head of [20:21] engineering at Venmo where he kind of [20:23] said, "Hey, hey look guys, your apps are [20:25] pretty dumb and I would like to license [20:27] the back end to what you do and we'll [20:29] pay you for that." You were like, "Oh, [20:31] that sounds better. Yeah, let's do [20:33] that." Were they your first customer? [20:34] They were the first meaningfully sized [20:37] customer to show significant intent. [20:39] What then happened was we went and got [20:40] into this like 9-month procurement [20:43] process because they'd been bought by [20:45] Brainree, which got bought by PayPal, [20:47] which got bought by eBay. And so I was [20:49] dealing with the eBay procurement [20:50] officer for like nine months. But along [20:53] the way, we sold to a bunch of other [20:54] other small startups, including Robin [20:55] Hood and Coinbase and, you know, a [20:56] couple of these other quite meaningfully [20:58] sized companies now. But yeah, it took a [20:59] while to get there with Venmo. But [21:01] anyway, so the theory was we wanted to [21:02] build an API for your bank account. We [21:04] believe that if we create an API for [21:05] your bank account, that is the missing [21:06] link in creating great digital financial [21:08] products. And we got the insight for [21:10] that from building these consumer [21:11] financial products ourselves, which were [21:13] all terrible, but we pivoted to finding [21:15] this really important niche that allows [21:17] a consumer to link their bank account to [21:18] a digital application. Originally, that [21:21] started with consumers linking, let's [21:23] say, their Wells Fargo account to the [21:24] Venmo app so that you can actually pay [21:26] your friends with Venmo or Wells Fargo [21:28] account with the Robin Hood app. So, you [21:30] could fund your Robin Hood account or [21:31] fund your Coinbase account or build [21:33] budgets or do expense management or [21:34] bookkeeping or whatever it is. So, we [21:36] work with basically every fintech [21:37] company. Over time, now it's become, you [21:39] know, the way that you open a city [21:40] checking account, you need to fund that [21:42] with an existing checking account. So, [21:44] how do you actually do that connection? [21:46] And increasingly, a lot of the large [21:47] enterprises are using us. So, you know, [21:49] the way that you pay your bills or the [21:51] way that you apply for an apartment and [21:52] pay your apartment rent or something [21:53] like that. Can I ask a funny question [21:55] about this era? So, these days you [21:58] formally partner with banks in a bunch [22:00] of ways. You talk engineering team to [22:02] engineering team. That was not the case [22:05] back in 2012 and 13. What code was your [22:08] team writing to build such an API to a [22:11] bank account that doesn't necessarily [22:13] contemplate having an API? It's a good [22:15] question. So in the early days, we knew [22:18] that there was this principle outlined [22:20] under the DoddFrank regulations that [22:22] said that consumers own their financial [22:24] data and they should be able to [22:25] permission their financial data in the [22:27] way that they want to. If you look at [22:29] the European regulations, there were [22:30] pretty clear open banking laws in both [22:33] the UK and Europe that said that [22:34] consumers are able to get their [22:36] financial data and they actually [22:37] mandated how banks have to build APIs. [22:39] In the US, there is this clear statute [22:42] written into DoddFrank, but the rules on [22:44] how the data would be shared weren't [22:46] written yet. It was assigned to the [22:48] CFPB. The CFPB hadn't started working on [22:50] it. And so, in the early days, we went [22:52] and talked to a bunch of lawyers and [22:53] basically figured out this thesis that [22:55] consumers own their financial data and [22:57] banks have to find a way to provide the [22:58] consumers with that financial data. And [23:00] if the banks don't have APIs, then we [23:02] could build kind of outside in [23:04] integrations to the banks. But we we [23:05] built screen scrapers to actually allow [23:07] a consumer to collect their own [23:08] financial data. And doing this at scale [23:11] is very complex. Yeah. To hundreds of [23:14] banks with all these different types of [23:16] terrifying screen scrapers. 12. Oh my [23:18] god. Of banks and financial [23:20] institutions. I mean, you knew you had [23:22] the legal air cover to do so, but like [23:26] sure their IT teams are having alarm [23:27] bells going off. Well, we we tried our [23:30] best to do it in partnership clearly [23:33] with the banks. I don't think every bank [23:35] if you ask them would necessarily agree [23:37] that they had as much data as they might [23:38] have wanted. Broadly our goal was always [23:41] to partner with the banks. Like we [23:42] wanted the banks to be customers. We [23:44] wanted them to be a participant [23:45] ecosystem. And many of the banks [23:47] actually came to us and said great we'd [23:49] like to move to an API. We haven't built [23:51] it yet. Please just screen scrape us [23:52] because our consumers really want to use [23:54] apps like Venmo and Robin Hood and [23:55] Coinbase but we haven't built these APIs [23:58] yet. In a lot of senses very [23:59] collaborative. In some senses it was a [24:00] little bit more antagonistic. So in some [24:02] senses the banks were frustrated with [24:04] the fact that Robin Hood existed because [24:06] Robin Hood was stealing their trading [24:07] volume and so they didn't love fintech. [24:09] So they made it a little harder for us. [24:11] Eventually it got to the point where the [24:12] CFB started to write the rule and take [24:14] the actions and it became very clear [24:16] that you can't really tell a consumer [24:18] they can't have access to their bank [24:19] transaction data or they're they're not [24:21] allowed to see their account routing [24:22] number. If a consumer wants to open a [24:24] new investment account, you can't stop [24:26] the consumer from doing that. You know, [24:28] eventually we got to the sense of of [24:30] really strong partnership and [24:30] collaboration. At this point, we have [24:32] the vast majority of our data comes from [24:34] APIdriven integrations because the banks [24:36] have had the technical sophistication [24:37] and time to figure out how to build APIs [24:39] in a good way. We've also partnered with [24:41] them. We've built API platforms for [24:42] them. We help them launch APIs and so [24:44] forth. And so, we now are in a much more [24:46] long-term sustainable technical [24:48] infrastructure state. But in the early [24:50] days, there was a lot of leg work. I was [24:52] just imagining your early VC pitches [24:54] where you're like, "Yeah, yeah, we're [24:55] screen scraping the banks now, but don't [24:57] worry. In the future, we'll get to APIs. [25:00] We're going to get people's username, [25:01] usernames, and passwords. We're going to [25:02] store those and then we're going to have [25:04] a little bot that logs into their [25:05] account for them and executes a transfer [25:08] acting on their behalf as if we're them [25:11] logged into the bank. It's got like [25:13] scary written all over it." But here we [25:15] are today. This is an amazing tech [25:17] company story where you envisioned it's [25:20] going to be ugly getting there, but then [25:22] the future when we do get there is [25:24] actually the thing that everyone wants [25:25] and is much better for consumers. I [25:28] think a lot of elements are correct [25:30] individually. When we first thought [25:31] about building this company, I wasn't [25:32] sure if anyone was going to use it. When [25:34] we talked about, all right, great. A [25:35] user wants to link their bank account [25:37] with Venmo. One, is anyone going to do [25:39] that? Two, is the infrastructure going [25:40] to work? Three, is any consumer going to [25:42] choose to enter the necessary data, [25:43] which was at the time a username and [25:44] password. It turns out people did it in [25:46] droves. Despite the fact that we felt [25:48] very very good about the infrastructure [25:50] that we built on the back end, the [25:51] security team that we'd built, the level [25:53] of oversight, the privacy protections we [25:54] put in place, so on and so forth, we [25:56] couldn't guarantee that consumers would [25:58] feel good about it when we launched it. [26:00] And so we were skeptical. It turns out [26:02] the value to consumers was very, very [26:03] high. The customers that we worked with [26:05] ended up building amazing products and [26:07] this ecosystem just took off. And I [26:09] think despite the fact that we had this [26:11] theory early on that all the banks [26:13] should launch APIs, I think if the [26:15] ecosystem hadn't taken off, we probably [26:17] wouldn't have seen the banks actually [26:18] build this. In some sense, it was a [26:20] chicken and the egg thing where we had [26:21] to build the less scalable integrations [26:25] before we could get to the more scalable [26:26] ones. We interrupted you earlier when [26:28] you were taking us all the way to the [26:30] fintech boom and then bust. Call it [26:32] 2022. Yeah. So, you asked about what [26:34] products we were building. So we built [26:36] bank linking broadly and as we kind of [26:38] moved forward in the arc of plaid we [26:40] started to build things that rely on [26:41] linking or things that are adjacent to [26:43] bank linking. So we built out a set of [26:45] credit oriented products. So asset [26:47] verification, income verification. [26:49] Whenever you apply for a mortgage, you [26:50] need to verify your assets. Uh you need [26:52] to verify your income. If you're [26:53] applying for an auto loan, you need to [26:54] to prove that you have verified [26:56] employment. We can do all that [26:57] digitally. We acquired an ID [26:59] verification company. to every time you [27:00] link your bank account, you often times [27:02] then have to verify your ID that the [27:04] step before the step after. If we can [27:05] tie that together, we can make it so [27:06] that you can verify your ID once and [27:08] then every time you link your bank [27:10] account, we can pull along a verified [27:11] identity with it becomes very very [27:13] powerful. We built bank payments [27:15] infrastructure. So, not only do we let [27:17] you collect your account routing number, [27:18] but we actually let you set up an a [27:20] payment. Um, you can do something like, [27:21] you know, paying a bill or funding an [27:23] account or or whatever that is. And then [27:26] on top of that, we've built this big [27:27] analytics layer, which I'll come back to [27:28] later. But in 2022, we had this huge [27:31] variety of products and endpoints and [27:33] and we're touching a bunch of different [27:34] markets. And what that meant is in the [27:36] one side, we were working with most of [27:38] the fintech, the majority of the fintech [27:40] companies to do many things. We're [27:42] working with more and more of the banks. [27:43] We were working with more and more of [27:44] the enterprises. So growth was good, but [27:46] we're also very exposed to the macro [27:48] cycles of financial services. And as all [27:52] of you know and all of your listeners [27:53] probably know, financial services has [27:54] these the these big cycles that they go [27:56] through basically driven by interest [27:58] rates. A lot of money was moving when [28:00] interest rates were zero. Exactly. A lot [28:02] of money was moving. In the second half [28:04] of 2022, we saw this kind of rapid [28:06] increase in the interest rate. What that [28:08] meant is that investment markets crashed [28:11] and lending markets basically froze. [28:14] Lenders didn't want to lend anymore. [28:16] Stock prices dropped. Crypto prices [28:17] dropped. The funny thing is in our data, [28:19] we see that consumers rapidly sign up [28:22] for and purchase stocks when stock [28:23] prices are high. Consumers rapidly sign [28:25] up for and purchase crypto when crypto [28:27] prices are high. And when the prices [28:29] drop, then the usage drops. The exact [28:31] inverse of what perhaps people should [28:33] do, but it is the reality of what people [28:35] do do. Get greedy when others are [28:36] greedy, I think, is probably the phrase, [28:38] right? Yeah, something like that. And [28:40] what was your business model at this [28:41] point in time? Were you getting paid on [28:43] a per transaction fee, a per token fee? [28:46] Behind the scenes, we have a ton of [28:47] different endpoints. We have more than [28:49] 100 SKs. Each of those is priced a [28:51] little bit differently, but you can [28:52] think of our business model as making [28:53] money in in three senses. The first is a [28:56] per action sense, meaning every action a [28:58] user takes. The second is in a when I [29:01] say actions, that actually breaks into [29:02] two sides. One is a per user signup fee, [29:04] and then one is a kind of per payment [29:07] fee, and then we have a per user per [29:08] month fee for the different products [29:09] that we have. So, per user per month [29:11] actually stayed incredibly stable. per [29:12] payment stayed fairly stable. Then new [29:14] user signups kind of fell off a cliff. [29:16] Yeah. So the portion of your business [29:17] that is tied to essentially volume and [29:19] usage within the system was going great [29:22] during Zer and then all of a sudden not. [29:25] Yeah. Exactly. So that that that [29:26] definitely slowed down. For the most [29:27] part though, the business continued to [29:29] stay fairly stable. So we didn't see any [29:30] revenue declines. We saw slower revenue [29:33] growth because again the per user per [29:34] month and the usage based fees continued [29:36] to grow. But we saw a decline in in [29:38] relative growth. The best way to think [29:40] about it is, you know, if a lender slows [29:42] down their lending, they may sign up [29:44] less new users, but they're not going to [29:46] stop collecting repayments from the [29:47] existing users that they have. You know, [29:49] our growth may slow, but the the [29:50] absolute amount of revenue that we have [29:52] didn't. The second half of 2022, the [29:54] first half of 2023 were much slower for [29:56] financial services startups generally, [29:58] but then started to reacelerate in the [30:00] second half of 23 and and into 24 and [30:03] now into 25. We've seen kind of many [30:05] quarters of accelerating growth. While I [30:07] would like to not go through a interest [30:09] rate cycle like that again, I'm aware [30:10] that we will as a society and and [30:12] probably as a company we'll go through [30:14] another interest rate cycle. And in a [30:15] lot of senses, we came out of that much [30:17] stronger. We launched major new product [30:19] and business areas. We we fully [30:20] replatformed our strategy and the [30:22] business has been doing really great, [30:24] especially this year. I'm not going to [30:25] let you get up with a phrase like [30:27] replatformed our strategy. What does [30:29] that mean? That's MBA speak right there. [30:31] That is fair. I'm not an MBA. Uh, but [30:35] you did work at Bane, though. So, I [30:36] worked at Bane. I worked at Bane for for [30:38] 12 months. I think it was 12 months in [30:39] one day. I earned one bonus check, which [30:41] was $4,000. It was the most money I'd [30:43] ever seen in my life, and that gave me [30:45] all of the money that I needed to quit [30:46] and try to start a startup, and lasted [30:48] me a full two months of rent in New York [30:52] City until I was out of money again. You [30:54] have a totally different strategy now. [30:56] You have diversified into three new [30:58] businesses. I just read your annual [30:59] letter. Would you have diversified the [31:02] business away from just bank account [31:04] linking if it weren't for this bursting [31:08] of the fintech bubble? We definitely [31:10] would have. So the strategy that we've [31:12] always had is we need to build the API [31:15] for your bank account, the bank linking [31:17] functionality. But in the long term, we [31:19] believe that based on all the data [31:21] that's coming through the system and the [31:22] user patterns that we see, we can build [31:24] a much more valuable analytics business [31:26] that can meaningfully change the way [31:29] that we do important things in financial [31:30] services over time. So, we'd always had [31:32] the multi-step process of first build [31:34] the bank linking business, then use the [31:35] data to to go solve fraud or to go solve [31:38] credit scoring, but we had to get to [31:39] this certain amount of scale. What I [31:41] will say really enabled it was actually [31:43] raising the large round in 2021. We took [31:46] that capital, we rapidly kind of ramped [31:49] our headcount predominantly in [31:51] engineering to go replplatform basically [31:53] all of the the the data backend so that [31:56] we could now shift to being an analytics [31:58] business so that we can build great [32:00] analytics products and then we allowed [32:02] revenue to catch up and and and surpass [32:04] and that's what you're supposed to do [32:06] when you raise a fundraising round and [32:07] actually looking retroactively on our [32:08] financials it kind of follows the exact [32:10] right trajectory. So you see the the the [32:12] decline in operating margin and then the [32:13] reaceleration in operating margin. That [32:15] round really enabled us to go make this [32:17] big investment. The products that we've [32:19] been focused on are using the aggregate [32:21] data set that we see. So using the users [32:24] that have linked bank accounts through [32:25] plaid seeing all the applications they [32:27] linked to again understanding the fact [32:29] that we own an identity verification [32:31] platform and we can link that to a [32:32] governmentisssued ID ingesting a bunch [32:35] of data exhaust from the actions users [32:38] are taking ingesting thirdparty fraud [32:40] signals we've created kind of three [32:42] major new product area the first one is [32:43] around anti-fraud so we have some [32:46] products that we're going to launch very [32:47] shortly kind of expansion of the [32:48] anti-fraud product set where we can look [32:50] at a user's patterns across all of the [32:52] apps that they're using and if they [32:54] commit fraud in one of the apps that can [32:55] then be reported to us. We can federate [32:57] out that fraud signal or we can see the [33:00] fact that they've just signed up for [33:01] four loans in the last 12 hours. All [33:04] right, maybe that's a flag. So we can [33:05] build this kind of algorithmic type of [33:07] fraud tool, serve that up to all of our [33:09] customers and they can better protect [33:10] fraud in the ecosystem. And it's an [33:12] angle on fraud detection that no one's [33:14] ever had the data to build. That's one [33:16] example. The second big example is [33:18] building a real-time credit system. So [33:20] FICO came out with the way that credit [33:23] scores are built in, you know, the late [33:25] 80s, early 90s. And up until now, most [33:28] loans have been made based on FICO data. [33:31] We believe that FICO data is great, but [33:33] if you augmented that or expanded that [33:35] by using real-time data, you have a much [33:37] better picture of a consumer. So Ben, if [33:38] you got a a new job tomorrow, don't go [33:40] get a new job. I really am glad that [33:41] you're doing the job you're doing. But [33:43] if you got a new job tomorrow that paid [33:44] you five times as much, you would be a [33:46] much better credit risk because you [33:47] would have more free cash flow [33:48] available. you would be able to pay down [33:50] greater debt service. Yet, FICO doesn't [33:52] include that kind of data very well. And [33:54] so, we have a lot of real-time data that [33:56] can look at changes in jobs, changes in [33:58] spending levels. If you move into a [33:59] cheaper apartment, but you didn't get a [34:01] new job, you're actually a better loan [34:02] risk. So, we can understand like what is [34:04] the underlying spend on a given month or [34:06] or quarter or year, and then build a [34:08] credit score that that sits on top of [34:09] that. And then we've done a bunch of [34:10] work on payments analytics, helping [34:12] enable bank payments to be something [34:13] that can be more reliable, more [34:14] predictable, so on and so forth. [34:16] Basically, everything that we've [34:17] launched over the last two years has [34:19] been a version of an analytics product [34:20] that leverages the aggregate data set [34:22] that that comes through the rest of the [34:23] platform. And this is really cool [34:25] because like I imagine there's nobody [34:27] else who sees as much data across [34:32] consumer finance as you. What's the 50% [34:35] number you guys have? Something like is [34:36] it is it 50% of Americans use Plaid? [34:39] Yeah. More more than one in two people [34:41] in the US that has a bank account that [34:43] has linked account through Plaid. That's [34:45] incredible. You know, you could be JP [34:46] Morgan and be the largest financial [34:49] institution out there, but the only data [34:51] you're seeing is JP Morgan Wald Garden [34:54] data or likewise, you could be Visa and [34:57] be the largest credit card network out [34:59] there, but all you're seeing is credit [35:02] data across the actions that your [35:04] customers are taking with payment data [35:07] with Visa. You're seeing the whole [35:10] picture of everything everyone is doing. [35:12] The way that we think about our data set [35:13] is that we have a lot of unique elements [35:15] of our data set. So we don't have [35:17] anywhere near the depth of customer [35:20] information that JP Morgan has on JP [35:22] Morgan customers. The customers upload a [35:24] lot more data to JP Morgan. The [35:26] customers use apps a lot more than they [35:28] would interact with plaid. So there are [35:29] a lot of things that JP Morgan would see [35:31] that they can build really wonderful [35:32] products for JP Morgan customers. We [35:35] have a unique data set in as much as [35:37] it's an aggregation of of a bunch of [35:39] things that hasn't been aggregated in [35:41] this way before. Meaning we see all the [35:43] applications that that consumers link [35:44] to. We can see the activity within some [35:46] of those applications. We can link that [35:48] to government issued IDs or bank account [35:50] data so on and so forth. And so our [35:52] thesis on product development is not [35:55] just that we have a large data set, but [35:56] it is that we have a unique set of [35:58] things that come together in a way that [36:00] no one has ever analyzed for fraud [36:02] before or no one has ever analyzed in [36:04] order to build credit scores before. [36:05] That's the vector that we push on, which [36:07] is what are the unique things that we [36:08] can do that haven't been built before. [36:10] David and I are so interested in these [36:13] multiundred billion dollar trillion [36:15] dollar businesses where their first idea [36:17] was actually the good idea and then [36:19] everything else after that has just been [36:20] like cute. You know, you look at TSMC [36:22] tried to get into solar cells and all [36:26] sorts of things and the right answer was [36:27] just keep making integrated circuits. [36:29] Are these new business lines interesting [36:33] for you guys and working or was it just [36:35] like actually banklinking is an amazing [36:36] idea? Well, banking is certainly an [36:39] amazing idea and don't get me wrong, we [36:40] are still investing in that. That is [36:42] still growing and I think there's a lot [36:43] of room left for it. The way that we [36:45] came up with all these, it's based on [36:47] the things that our customers are trying [36:49] to do with the data, but they don't have [36:51] the scale to do themselves. So, we had a [36:54] lot of customers that were trying to [36:55] build credit scores on top of [36:57] transaction data that was coming in or [36:58] credit analytics that were based on [37:00] transaction data that's coming in, but [37:01] they just didn't have the data scale to [37:03] be able to do the level of analysis that [37:04] they needed to do. And so they would [37:06] come to us and they would say things [37:07] like hey can you just do this analysis [37:08] for me or could I like send someone to [37:11] like work in your database and just [37:13] build some queries for for us. We didn't [37:15] do the second one to be clear. We were [37:17] taking these real-time things that exist [37:18] out there and then just backing into [37:20] okay great because we have the data set [37:22] structured in this way then great we can [37:23] build a lot of these analytics. I would [37:25] say on the fraud side it's it's fairly [37:26] similar. We're looking a little bit more [37:28] into the future on the fraud side [37:30] because not a lot of people have had [37:32] this kind of data queryable in the way [37:33] that we have. But I would say for the [37:36] most part, yes, these products are [37:37] working. They're still very early. [37:38] Credit scoring is a gigantic market. We [37:40] are a tiny tiny sliver of it today. And [37:42] we hope that that that is the start of [37:44] something really large. Same for fraud [37:46] and frankly same for bank payments. It [37:48] is very fun to go in and be able to sit [37:50] down with customers and and and hear [37:52] them say, "Hey, I I really wish I could [37:53] do this thing with my data, but I think [37:54] you can do it with your data in a much [37:56] bigger, better way." Okay. I was [37:58] expecting you to answer with a stat from [38:00] your annual letter, and I just have to [38:01] say it because I think it's astonishing. [38:02] You you were very humble there. new [38:04] products, the three that you just talked [38:06] about, represented over 20% of ARR in [38:09] 2024, compounding at 93% [38:12] annually. That's what it's supposed to [38:14] look like. That is when you have a core [38:16] business and you're launching new growth [38:17] businesses, that's the exact profile [38:20] that you are hoping and dreaming for. I [38:22] agree. And we've been pleased with the [38:24] progress so far. I think there's a lot [38:26] more room to go. Wow. You are ready to [38:27] be a public company CEO. There's a great [38:29] book and I can't remember who wrote it, [38:31] but the title of it is pleased but not [38:33] satisfied. And just the title of that [38:35] book, I have it on my shelf behind me [38:36] and normally I have a bunch of books [38:38] behind me when when when I do these [38:40] pleased but not satisfied. I think that [38:41] that represents a lot of my management [38:42] philosophy. Okay. I want to shift back [38:45] to the managing a growing business in a [38:50] cyclical environment. both cyclical on [38:52] your customer side because people are [38:54] using fintex apps more and less as the [38:57] or or that there's an eb and flow there [38:59] but also on your own funding your own [39:02] valuation you raised a large successful [39:05] financing round this year at a much [39:07] lower valuation than your big 2021 round [39:10] can you talk us through how you thought [39:12] about that the mechanics of it and it [39:14] seems like you were pretty intentional [39:16] about managing things like employee [39:18] liquidity in that absolutely so We [39:21] signed paperwork to sell the company to [39:22] Visa for about $5 billion in 2020. In [39:26] 2021, we raised a round of financing at [39:30] 13.5 give or take billion dollars of [39:32] valuation. We then raised around this [39:35] year at around $6 billion valuation. So [39:37] big spike in valuation and then a [39:39] decline in valuation on the other side. [39:41] What's happened in that period is [39:42] largely market multiples have changed. I [39:44] mean 2021 was all-time high market [39:46] multiples and even better we had just [39:48] had the best advertising ever for the [39:51] company happen meaning you know the DOJ [39:53] wrote this big letter of why plaid was [39:54] going to take over the world so awesome [39:56] I shouldn't be able to have it and and [39:58] so on and so forth and so that is the [40:00] best possible environment to raise money [40:02] in and then this year we raised money [40:03] very different multiple environment the [40:05] business was fundamentally a much [40:06] stronger business so great growth break [40:09] even new products growing really quickly [40:11] compounding at a great rate really solid [40:13] customer base [40:14] All of the metrics looked looked much [40:16] better in in in this year than they did [40:18] in 2021. We were kind of telling the [40:20] story in 2021 of getting to where we are [40:22] now. The reality is market multiples [40:24] change. From an internal standpoint, I [40:26] think we try really hard to to [40:28] consistently have this employee message [40:29] saying valuate our success based on the [40:32] metrics that we set out, the impact, the [40:33] progress that we're making along the [40:34] stretch of of goals that we have. And [40:36] the valuation will fluctuate as [40:38] multiples fluctuate. Now, we can say [40:40] that until we're blue in the face and [40:42] employees will still care quite a lot [40:44] about the valuation. It's not to say [40:45] that we don't care about the valuation. [40:46] We do. But ultimately, we focus on the [40:48] things that we can control and try to [40:50] improve the fundamentals over time. And [40:52] then we believe in the long term the [40:54] market will will weigh that accurately. [40:57] So [40:58] mechanically the way that venture [41:00] capital financing works is someone [41:03] invests at a valuation and then as long [41:05] as that valuation keeps going up, every [41:08] party is happy. The investors are happy, [41:10] the employees are happy, given this [41:12] whole crazy strike price options thing, [41:14] the way that employee compensation works [41:16] in this system, that's just the way it's [41:19] supposed to work. What things break when [41:22] your valuation goes down and how did you [41:24] account for that? We don't issue [41:26] options. We issue RSUs. And so because [41:29] you're issuing RSUs, you don't have the [41:31] kind of strike price issue. Without [41:34] getting into too much of the tax detail, [41:36] it means that that is not as much of a [41:38] problem for us. We were fortunate that [41:40] all of our rounds have had very clean [41:42] terms, meaning that investors may well [41:45] be disappointed that the price has come [41:46] down. Actually, I shouldn't say may well [41:48] be disappointed. I'm sure they were [41:49] disappointed that the price has come [41:50] down, but ultimately they are still [41:52] along for the ride. We actually had a [41:54] lot of existing investors reinvest in in [41:56] the most recent round believing that [41:57] there's a great deal of growth ahead of [41:58] us. I mean, they obviously operate in [42:00] the same fluctuating market conditions [42:02] as as founders do, too. So, and I would [42:04] say it's not a surprise to any of our [42:06] investors. We've been pretty transparent [42:07] with our investors and our team over [42:08] time. We've cut the internal valuation [42:10] multiple times between the 2021 round [42:12] and now. As much as this is not a fun [42:14] process to go to to get here, I am, you [42:16] know, excited now to have this behind us [42:18] and and and be focused on, you know, [42:20] postvaluation reset just just really [42:22] laser focused on growth. It must just be [42:24] maddening when a number that everyone is [42:27] anchored on is going down, but you are [42:30] well aware that the business is at the [42:33] strongest point in history intrinsically [42:35] worth more. [42:36] Yeah. Yes. Yes. [42:40] I wonder if it's perhaps because your [42:43] actual operating business is also [42:46] indexed to the financial markets. You [42:49] can either fight the cyclicality of your [42:52] industry or embrace it. And it feels [42:54] like, you know, if you're going to [42:55] operate in financial services, you just [42:57] kind of need to embrace it. It feels [42:59] like you've adopted that mindset. We do [43:02] both. We fight the cyclicality in as [43:04] much as we launch products that are [43:05] either not cyclical or counteryclical. [43:08] And over time as we get bigger, you [43:10] know, cyclicality should impact us less. [43:12] But we also embrace it. It is just a [43:14] reality of our business. I hope that my [43:16] lending customers grow a ton. And even [43:18] if I launch products that are [43:19] countercyclical, I hope that they [43:20] continue to create cyclicality forever [43:22] because they've grown so much and they [43:23] represent such a large portion of our [43:25] business. But on the flip side, it is [43:27] something that we have to be aware of. [43:28] It's something that we write in our [43:29] shareholder letters so that our [43:30] investors understand it. Thinking about [43:32] one day becoming a public company. We [43:34] want to continue to be very transparent [43:35] on uh you know, the drivers of growth [43:36] for our business. We control a a lot of [43:39] the fundamentals of the business, but we [43:41] don't necessarily control the way that [43:42] macro is going to impact us. We can [43:43] react to it. We can prepare for it, but [43:45] we can't predict it. Could I ask about [43:47] your different moments in time of [43:49] product market fit with banks? Because [43:52] it seems like you always had product [43:53] market fit with users being willing to [43:56] off so apps can do cool things. When did [44:00] it start to change with banks where it [44:02] went from feeling like you were pushing [44:03] a rock up the hill to banks looking and [44:05] saying, "Oh, your vision's right and we [44:09] agree with you about this future. How do [44:10] we help you?" It's a good question. [44:12] There there were a handful of moments [44:16] where banks started to understand and [44:18] really think about technology using [44:20] technology to serve their customers in [44:22] in a more modern way. The first one was [44:24] actually GovCloud. I think that was in [44:27] was it like [44:28] 2014 maybe 2015 somewhere around there. [44:31] I'm not sure actually when it launched [44:32] but this is certainly when the bank [44:34] started talking about it when the [44:36] government said oh my gosh we can put [44:37] data in the cloud. Amazon launched gov [44:39] cloud behind it. That was a big step [44:41] forward where the bank said if the [44:42] government can put data in the cloud [44:43] then maybe I can put data in the cloud. [44:45] The next one was a series of [44:47] announcement from JP Morgan saying, you [44:50] know, this year we spent, you know, $2 [44:51] billion on technology. Next year we [44:54] spent $4 billion on technology. Recently [44:56] they're saying like $17 billion in [44:57] technology. They're the leader. They're [44:59] the biggest. But every every other bank [45:01] looked at that and said, "You're [45:02] spending $17 billion in technology. I [45:04] should be spending on technology." And [45:06] you know what will I get for it? The [45:08] third big thing that happened is that [45:10] the fintex did really well. these [45:12] companies, Robin Hood, grew a lot and in [45:15] growing that much, they started to steal [45:17] customers. They they started to have [45:18] high valuations. A lot of the the [45:21] financial services players either had to [45:23] react to them. So, you saw Schwab [45:25] cutting their fees, cutting their [45:27] trading fees. You basically don't find [45:28] flat trading fees anymore in the [45:30] industry because Robin Hood said, "No, [45:31] no, no, no, it's free." And everybody [45:33] had to react to it. Or you find them [45:35] copying. You oftenimes find both. So, [45:37] reacting and then copying. And as the [45:39] banks started to lean in and build great [45:41] financial technology products, they were [45:42] using it to better acquire customers, to [45:44] serve customers more more cheaply. [45:45] Things that are less visible, they [45:47] started to use tools to avoid fraud, to [45:48] to make better underwriting decisions, [45:50] to improve efficiency of their teams, so [45:52] on and so forth. So the reality is just [45:54] the value is there, but it took a [45:56] handful of these proof points, the [45:57] expansion of fintech, the the GovCloud [45:59] launch, JP Morgan just announcing these [46:01] gigantic numbers in terms of how much [46:02] they spend for the industry to follow [46:04] along. In a lot of ways, you guys are [46:06] kind of like Microsoft or maybe [46:08] Microsoft in the, you know, late 90s, [46:11] early 2000s in that it's important that [46:15] you have a great brand both with [46:18] consumers and with large enterprises and [46:21] financial institutions and usually those [46:24] things are not the same playbooks. How [46:26] do you think about Plaid's brand and [46:29] building trust with consumers and trust [46:31] with like you know JP Morgan? Well, [46:34] first I have to acknowledge any sentence [46:36] that compares Plaid to Microsoft is the [46:38] greatest compliment that you could ever [46:39] say about our business. So, um, thank [46:41] you. But you're welcome. We think quite [46:45] a lot about how to build a brand on [46:48] three sides. We want to build a brand [46:50] for developers regardless of where they [46:52] work, but for developers. So, where we [46:55] started, we we started with developer [46:56] roots. We want to build a brand that [46:58] financial institutions know and [47:00] understand because they are they are key [47:02] partners to us. Um, yes, they're often [47:03] times customers. They're also data [47:04] sources and and and partners in a lot of [47:06] ways. And we want to build a brand that [47:08] consumers understand and trust. And [47:10] actually, I think the consumer one is [47:11] the hardest because we build a very very [47:14] thin user experience. Our user [47:16] experience is embedded in every Plaid [47:17] powered product and it says, "Hey, [47:19] you're linking an account via Plaid in [47:20] order to do this thing and this is what [47:22] Plaid does." But we don't have a lot of [47:24] real estate there. In a lot of senses, [47:26] we think about a brand that's like the [47:28] card networks, but the card networks [47:29] have the cards with the logo on the back [47:31] of it. We have just this this 15 30 [47:34] second user experience that a consumer [47:36] sees and and and goes through and your [47:38] goal is to actually minimize that to [47:40] make that user experience as short as [47:41] possible. Yes. And so it's hard to take [47:43] credit there. But the feeling that we [47:45] want to create is a sense of oh that [47:47] just worked or oh that was simple and a [47:50] sense that oh I trust this like I feel [47:52] like I have resources and tools. So [47:53] we've built a handful of tools that you [47:56] could interact with either via that [47:58] linkage or asynchronously to see where's [48:00] my data going, how is it working. We're [48:02] working on a handful of things that are [48:04] related to kind of increasing security [48:05] and privacy protections for consumers. [48:07] So ultimately we think we want to create [48:09] the brand that is steeped in trust, [48:11] steeped in clarity and understanding for [48:13] consumers, but ultimately it's an [48:15] ingredient brand. It's a thing that you [48:16] interact with for only a few seconds [48:18] that supports the brand of the customer [48:20] that you're going to go use. So, you [48:21] know, Plaid's brand should support Venmo [48:23] in some sense. It should create a little [48:24] bit more trust or a little bit more [48:25] simplicity for Venmo, but we never want [48:27] to use SER, the brand that Venmo has [48:29] created with their own customers. All [48:31] right. I know you're close with Hamilton [48:33] Helmer also. Give me your own seven [48:35] powers analysis. Why can I not just [48:38] wander out there tomorrow now all these [48:40] banks have these nice APIs and I'll just [48:42] build something that looks a lot like [48:44] Plaid? I think a lot of it comes back to [48:47] network effects. I think it's probably [48:48] the easiest place to start. So the thing [48:50] that matters most to our customers is [48:52] how many users that try to sign up end [48:55] up signing up. You know, in some sense [48:56] you could say we're onboarding as a [48:58] service. If a user wants to sign up for [49:01] Robin Hood and that user fails to sign [49:03] up for Robin Hood, Robin Hood is [49:04] frustrated. If the user signs up for [49:06] Robin Hood successfully, they're very [49:07] happy and we provide a lot of lift [49:09] there. So for users that we've seen [49:10] before, we can do a lot of things to [49:11] make the user experience faster and [49:12] easier. If you've verified your identity [49:14] once, we can automatically pull that [49:16] identity through in a in a better way. [49:18] So like first and foremost it's we get [49:20] onboarding right and we do that a lot of [49:21] that through network effects. Second is [49:23] that there are data network effects on [49:25] the back end as well. So because we see [49:27] all this data coming from all of these [49:28] users in all these places we can now [49:30] build products that are differentially [49:32] useful to our customers. So we can build [49:34] a fraud analytic that no one else can [49:35] build. We can build a credit product [49:37] that no one else can build. And so [49:38] that's probably the foundation of it. I [49:41] would like to think that in the long [49:42] term we'll have some brand [49:43] differentiation but I don't want to kid [49:45] myself because even Hamilton says that [49:47] brands are the weakest form perhaps of [49:50] power. Ah it's funny we disagree with [49:54] him on that but no I think brand is [49:56] actually super important. That's why I [49:57] asked the question a minute ago like [49:59] you've kind of to my view done this [50:01] amazing job of you started by powering [50:05] this new wave of fintech apps that felt [50:08] almost sort of like dangerous to use. It [50:11] was millennials who were young at the [50:12] time who were like using Venmo and being [50:14] like yeah why not you know sure I'm [50:17] going to give my bank credentials to [50:19] Venmo taking things into my own hands [50:21] even if it's a little risky. And over [50:24] time, you've now become, oh no, this is [50:26] actually the safer way to do this. [50:28] That's a very difficult thing to do. The [50:31] big shift for us was my mom a couple [50:33] years ago, and she was in her late 60s [50:35] at the time. She called me and she said, [50:37] "Hey, you know, my friend recommended [50:40] this Rocket Money product and it and it [50:41] helps me manage my bills. Is that safe?" [50:44] And I said, "Yeah, mom. It it says [50:46] powered by Plaid in the middle of the [50:47] side. Don't don't you recognize that?" [50:49] And she said, "Oh, I I I guess so." [50:51] Okay. And she's like, "I'm going to tell [50:52] all my friends." that she did. She told [50:53] all her friends and then we we saw this [50:54] like, you know, mass number of signups [50:56] from people that we never thought would [50:57] be interacting with finish technology [50:59] and um not just from your mom spreading [51:02] the word. Not just from my mom. No, no, [51:04] in a lot of different areas. The one [51:05] other thing while we're talking about [51:06] power, the thing that I I know might not [51:09] fall into the seven powers framework, [51:10] but I think is really important is also [51:12] just a willingness to move fast and [51:15] build products faster than others in the [51:17] market. And I know that it is not, you [51:20] know, what Hamilton might call a durable [51:21] competitive advantage, but I think it [51:23] matters quite a lot. And so that for us [51:25] is one of the things that we also push [51:27] on is we should be like move fast as a [51:29] company value. We push quite a lot on [51:31] trying to be the first, the best, the [51:33] highest quality, the fastest moving and [51:34] so forth. My pretend MBA, which I don't [51:37] have putting that hat on, I would say [51:39] that's just operational excellence. [51:40] That's not strategy. Everybody wants to [51:43] be the fastest and everybody wants to [51:44] have the culture that ships the most [51:46] product and you just believe you're very [51:48] good at it. Exactly. And Hamilton would [51:50] call the best companies process power [51:52] and he would also say that tech [51:53] companies are not allowed to consider [51:54] themselves as having process power. But [51:56] I should say we should also aspire to [51:57] have that one. Okay. So you talked about [52:00] how you started as a developer focused [52:02] company and how important it is to build [52:03] trust with developers and make their [52:04] lives easier. The whole notion of being [52:07] a software developer is extremely [52:09] different today than it was a year ago [52:12] with the advent of AI. How are you [52:14] thinking about being a company with [52:15] developers as sort of your primary [52:18] stakeholder? We started out building [52:21] products for developers. We thought that [52:22] developers would be the changemaker [52:24] within the organization. When I say it [52:25] today, it's really boring and people [52:28] probably roll their eyes, but when you [52:29] started saying this in 2013, it was [52:32] unique. You know, Twilio was the one [52:34] that Twilio. Yes. Exactly. Ask your [52:35] developer. Exactly. You know, when we [52:37] went to financial services, the [52:38] developers were like, "Oh my gosh, [52:39] you're building things for us. How do I [52:41] even do this?" At the time, our [52:42] competitors made it really hard to [52:43] access the docs. We just put the docs [52:44] online. That seemed very easy and [52:46] simple, but it generated a lot of [52:48] excitement. And there were a bunch of [52:49] these examples on on things that we did. [52:51] Uh, you know, we built an arrest [52:52] architecture. Our competitors built in [52:53] soap. You know, we tried to stay what I [52:56] would call just making normal decisions [52:59] along the way, but it turns out that was [53:00] that was more on the cutting edge. [53:02] Building modern software. Exactly. [53:04] modern software these days the nature of [53:06] development has changed massively. So so [53:08] many more people can consider themselves [53:10] to be developers or vibe coders or [53:12] whatever they want to call themselves. [53:14] The way that we think about writing [53:16] software has changed and I think will [53:18] change much more over the coming years. [53:20] For example, we've we've shifted to [53:22] launching a plaid MTP server so that [53:24] developers can very easily build plaid [53:27] integrations directly within cursor or [53:28] windsurf or or whatever it is that they [53:30] want to build in. We're continuing to [53:32] push down this path. I think that the [53:34] way that people not only build products [53:36] but potentially even discover products [53:38] is going to change hugely as they in the [53:41] cursor chat they just type in like hey [53:43] what should I use to set up an a payment [53:45] or what should I use to set up a bank [53:47] linkage we need to make sure that cursor [53:49] both both says hey you should use plaid [53:50] but also then when it says hey you [53:52] should use plaid they might say and do [53:54] these two or three things or tell me [53:55] these two or three things and I'll set [53:56] it up for you we're continuing to push [53:58] quickly down that path and I think we [54:00] have a lot more to do interesting [54:03] All right. What is something that you [54:05] used to believe that you have completely [54:07] changed your mind on? You know, I think [54:10] we're in an interesting period [54:15] of [54:17] work wherein the returns on hard work [54:23] are not as obvious as they used to be. [54:26] So in the early days of Plaid, one of [54:30] the things that we were able to [54:33] differentiate on was what I called [54:36] grinding projects or grinder problems [54:38] where if we just went and did the work [54:42] more efficiently, oftentimes through [54:44] brute force than any of our competitors, [54:47] we would end up with a better product. [54:49] So you know, we built 12,000 bank [54:50] integrations. How did we build 12,000 B [54:52] integrations? There's nothing hard about [54:54] building 12,000 bank integrations. we [54:55] just did it. We set up incentives. We [54:57] set up teams. We set up structures. We [54:58] set up a culture where we just said, [55:00] "We're just going to go brute force, do [55:01] all the stuff that we need to do." In [55:03] today's software development [55:05] environment, that is no longer a [55:06] differentiator. It's not about the pure [55:08] volume of work that can be done [55:10] necessarily because agents will do much [55:12] if not most of that work for you. There [55:14] are increasingly the decisions that you [55:17] make as a developer or as a product [55:19] manager or as a designer or whatever it [55:20] is, the quality of those decisions [55:22] matters quite a lot more. And so I think [55:24] there are increasing returns to [55:27] strategy. There are increasing returns [55:29] to thoughtful decisions, but there's a [55:32] decreasing return on just pure volume of [55:36] work. And that may change in the future [55:38] again. So it may get to the point where [55:40] we've all evolved to use the new tools [55:42] and then volume of work actually matters [55:43] again. But right now, I don't think pure [55:45] volume of work matters as much. That's [55:47] good news for our friend Hamilton. [55:51] Yeah, that's interesting because in [55:53] theory once everyone learns to use the [55:56] new tools and this is making an [55:58] assumption that the amplification of new [56:00] AI tools are static which is is wrong [56:03] but let's just go with it for a minute [56:04] then theoretically everyone just has the [56:06] same 100x multiplier on their you know [56:10] their work. Exactly. And I think we're [56:12] in a unique period of time where [56:15] we're not there yet. It hasn't happened [56:17] yet. Yeah. No, no one is using the tools [56:18] at maximum efficiency yet. Like we're [56:20] not at the next kind of efficient [56:22] frontier. And until we get there, use of [56:25] tools effectively will matter perhaps [56:28] more than, you know, volume of work. I [56:31] think Hamilton would probably say, okay, [56:32] you have to assume in the future that [56:34] people will learn how to use tools [56:35] effectively. That'll be evenly diffused [56:38] throughout society. Yeah, your point [56:40] that like strategy is going to matter a [56:42] lot more in the current slashcoming [56:45] world is a really interesting one. I'm [56:47] so glad that I came up in a different [56:50] era. I feel like I just worked extremely [56:53] hard for 80 plus hours a week for a [56:55] decade starting at age 18. And that's [56:58] the only way that I achieved anything in [57:00] the world was by working harder. I was [57:04] not a good strategist in my 20s. [57:07] My co-founder and I were having dinner [57:10] with a friend a week or two ago, my [57:12] former co-founder. He's now off to [57:14] starting a new company. And we're having [57:15] dinner with a friend a week or two ago. [57:17] And the conversation was basically we [57:19] all wasted our 20s. You know, we all did [57:22] work in our 20s that at the time was [57:23] really valuable, but today, you know, a [57:25] software tool could do in an instant, [57:28] except that it turned into you becoming [57:29] a large owner of a company called Plaid. [57:31] Like, you converted that labor into [57:34] capital for sure. Sorry, I'm not saying [57:36] I'm unhappy with the results of the work [57:37] that we did at the time. It's just the [57:39] inefficient way that we did it. We're [57:40] seeing it with new AI companies now. [57:42] Like companies are getting started and [57:44] in a matter of months hitting 100 [57:45] million in ARR. You know, that's because [57:48] like the barrier of to get to that scale [57:50] you needed to grind that existed [57:52] before just kind of doesn't exist [57:54] anymore. It's gone. Well, Zach, that's a [57:57] great place to end it. Where would you [58:00] like to point listeners if they want to [58:01] learn more about you or Plaid or follow [58:03] you anywhere in the world? So to find [58:06] Plaid is just plaid.com or plaid onx. Uh [58:10] if you want to find me, I'm zack par z a [58:12] c h p ru t onx. And if you have thoughts [58:16] or feedback on plaid products, I hope [58:18] that you'll reach out to me. I would [58:19] love to hear it as always. Awesome. All [58:22] right, listeners, that is a wrap on our [58:24] conversation with Zach. If you want to [58:27] come to Radio City Music Hall on July [58:30] 15th, we would love love love to see you [58:32] there. That is [58:34] acquired.fmnyc to get one of the last [58:36] few tickets before it is sold out. If [58:39] you want to know every time a new [58:40] episode drops, check out our email list. [58:42] It's also the place where we show little [58:44] hints at what our next episode will be, [58:46] share corrections, updates, and little [58:48] tidbits that we learned from previous [58:50] episodes. That's [58:52] acquired.fm/email. If you want to chat [58:54] about this conversation that we had with [58:56] Zach or anything else, go to [58:59] acquired.fmslack to hang out with the [59:00] entire Slack community. So with that, [59:03] listeners, we'll see you next time. [59:05] We'll see you next time.