Nubank Enters US to Attack Incumbents' $600 Billion Deposit Spread

David Vélez：打造Nubank与全球银行业的未来——[Invest Like the Best，第495期] David Vélez 详述了 Nubank 如何仅凭约 1 万名员工就将客户规模扩展至 1.4 亿，并对传统网点型银行保持了 97% 的结构性成本优势。他解释了全球银行业为何仍处于技术颠覆的成熟期，展示了实时支付轨道、开放金融和自研 AI 模型如何为历来缺乏银行服务的人群推进财富管理与信贷创造的普惠化。Vélez 回顾了 Nubank 内部由 AI 驱动的“红色警戒”重组——该举措扁平化了管理层级并重构了 120 项业务工作流，同时概述了进军美国市场和稳定币生态系统的国际扩张战略。在整个访谈中，他强调了创始人信念、以客户为中心以及选择痛感足够强烈、足以支撑数十年执行力的问题具有无可替代的价值。 Nubank 仅拥有约 1 万名员工且无任何实体网点，却服务着 1.4 亿客户，其服务成本仅为传统银行的约 3%。 (2:52) 全球零售银行业仍处于“百视达时代”，约 90% 的业务依然通过传统分支网络进行，其中包括美国每年将约 6000 亿美元的存款利差从储户转移以补贴实体网点开销。 (6:31) 在 AI 代码模型的进步引发 2025 年末的“红色警戒”后，Nubank 重构了 120 项内部工作流，并将管理层级从 10 层缩减至 5 层。 (45:15) Nubank 正在进军美国市场，并推出覆盖超过 35 个国家的稳定币支持账户，旨在凭借软件驱动的承保能力与跨境支付轨道，直击收费高昂的传统零售银行。 (34:19)

Nubank founder and CEO David Vélez argues that global consumer banking remains in its "Blockbuster era," with branch-reliant incumbents controlling 90% of a $10 trillion industry. He contends that branchless, software-driven cost structures combined with artificial intelligence and digital payment rails will dismantle incumbent margins and extend modern credit to billions of underserved consumers. Structural Cost Advantages Over Legacy Branches (4:15) According to Vélez, traditional banking models are constrained by physical branch networks and massive corporate headcount. Nubank serves roughly 140 million customers with approximately 10,000 employees, whereas traditional competitors deploy 100,000 employees and 6,000 branches to serve 100 million customers. This operational gap reduces Nubank's cost to serve to approximately 3% of an incumbent's, enabling the company to deliver returns on equity two to three times higher than legacy peers despite charging zero account fees. Vélez characterizes incumbent deposit spreads in developed markets like the United States as an obscured, regressive tax on ordinary depositors. Across the US system, retail customers hold $15 trillion in deposits yielding an average of roughly 40 basis points while base interest rates hover between 4.5% and 5.0%. This spread transfers an estimated $600 billion annually from depositors to banks, primarily sustaining branch footprints and overhead. As financial rails transition to lightweight software architectures, Vélez forecasts that remittance fees and checking costs will compress toward zero, leaving legacy profit pools vulnerable. Underwriting Informal Economies and Expanding Access (27:08) Vélez identifies access as a primary bottleneck in consumer finance, noting that roughly 3 billion people globally remain excluded from institutional credit, with credit card penetration lingering at approximately 13% in major economies like Mexico and Colombia. Historically, extensive cash reliance prevented lenders from verifying borrower solvency. Sovereign instant-payment networks, such as Brazil's PIX and India's UPI, are removing this hurdle by converting cash flows into auditable digital ledgers, creating underwriteable transaction histories for small businesses and individuals that drive the majority of emerging-market employment. Rather than waiting for conventional credit histories, Nubank initiated lending with credit limits as low as $20, gradually graduating compliant borrowers to $100 and eventually $1,000. Through this tiered framework, the company became the first credit card issuer for 30 million people across Latin America. Vélez views this laddered underwriting approach, combined with algorithmic personal financial coaching, as a repeatable catalyst for financial inclusion and broader GDP growth. Defensibility: Scale, Data Moats, and AI Agents (12:21) Evaluating competitive dynamics, Vélez argues that neither startup agility nor incumbent size guarantees an enduring advantage. While early-stage AI startups can write software stacks rapidly, they cannot easily bypass the two- to three-year regulatory process required to secure banking charters. Conversely, legacy banks suffer from what Vélez terms the "Blockbuster fallacy": possessing decades of analog records across fragmented IT systems and legacy mainframes, which cannot easily be converted into predictive intelligence without modern cloud infrastructure and an experimental engineering culture. Nubank leverages a proprietary data asset covering 140 million users and 30,000 data points per customer over 13 years, bolstered by foundation models acquired through Hyperplane. Looking forward, Vélez projects that AI will democratize elite private banking—such as automated yield optimization, tax efficiency, and cross-lender rate comparisons—for the 99% of consumers who lack human advisors. Furthermore, Nubank treats autonomous software agents as a distinct demographic segment, developing machine-readable interfaces and agentic search optimization while deploying internal agents that can recommend both internal and third-party financial products. Internal Restructuring and Flattened Organizational Design (47:35) After discovering that frontier models could competently write and refactor its core Clojure codebase, Nubank initiated an internal "Code Red" to accelerate operational automation. The shift led to a major restructuring that halved Nubank's management hierarchy from ten organizational layers down to five, while expanding Vélez's direct reports from 1 to 20. Senior managers are also required to build and deploy functioning software quarterly to maintain practical understanding of the shifting technological frontier. To eliminate the filtering of information through middle management, Nubank implemented an internal real-time data engine known as the "company brain" to surface project telemetry and operational performance directly to leadership. Vélez also asserts that AI requires replacing semi-annual, backward-looking performance reviews with continuous telemetry-driven coaching. Counterintuitively, he argues that the automation of routine, asynchronous cognitive tasks strengthens the necessity for in-person office work, which fosters high-bandwidth debate, trust-building, and real-time collaboration that algorithms cannot replicate. Nubank's long-term trajectory depends on whether it can replicate its Latin American playbook across 100 countries to reach one billion customers, or if it will face growth plateaus due to domestic saturation in Brazil—where it already serves roughly 65% of the adult population—and the operational challenges of competing offline in cash-heavy and heavily defended international markets.

## Background 本期内容由投资商业类知名播客栏目 Colossus 于 2026 年 10 月 13 日在 RSS 及各大播客平台发布，属于《Invest Like the Best》第 495 期节目。节目简介指出，本期访谈嘉宾为服务拉美 1.4 亿用户的数字银行 Nubank 创始人兼首席执行官 David Vélez；他探讨了全球银行业仍处于“百视达时代”的现状、Nubank 如何围绕 AI 全面重塑组织，以及将业务拓展至美国等全球市场的长期战略野心。 ## TL;DR Nubank 创始人兼 CEO David Vélez 指出，全球个人金融仍有约 90% 的份额被沉重且昂贵的传统线下网点把持，而数字原生架构与 AI 将对银行业带来几何级数的效率颠覆。节目深入剖析了 Nubank 如何凭借 1 万名员工服务 1.4 亿客户、将单客服务成本压缩至传统巨头的 3%，并通过自研金融垂直大模型打通授信与风控。对话还探讨了全球无现金化进程、开放银行打破数据惯性、稳定币重构跨境金融，以及 Nubank 进军美国市场、服务 Agent 新客群并彻底重构内部 120 个工作流的激进变革。最后，David 分享了平衡“创始人”与“CEO”内在张力的心路历程，呼吁创业者在 AI 门槛降低的时代更应勇敢挑战那些看似“不可能”的硬核使命。 ## Atmosphere & style 整场对话兼具宏大商业推演的战略纵深与极为坦诚的创始人内心独白，语调从容理性却又暗涌着战时统帅般的危机感。主持人 Patrick O'Shaughnessy 的精准提问不断激发深度复盘，促使 David 从冷酷精准的商业数据（如 3% 的服务成本与 6000 亿美元隐形税）无缝切换到童年颠沛流离后重返哥伦比亚的高光泪水。听众能真切感受到一位经历过“代码红色警报”、曾深陷冒名顶替综合征的实干家，在面对技术范式转移时所展现出的纯粹兴奋、对极致效率的偏执，以及那份愿意为改变世界而承受漫长痛苦的深沉使命感。 ## Bulletin - `00:00` 全球约 10 万亿美元的个人消费金融市场中仍有近 90% 被传统银行把持，数字银行整体渗透率不足 10%，行业依然处于类似百视达时代的初期演进阶段。 - `02:52` Nubank 仅凭 10,000 名员工和零物理网点服务 1.4 亿用户，其单客服务成本仅为传统巨头银行的 3% 左右，实现了极悬殊的结构性成本优势。 - `04:53` 全美零售存款规模约 15 万亿美元，但平均存款收益率仅为 0.4%，在高息环境下相当于储户每年向传统在位银行补贴近 6,000 亿美元以维系昂贵实体网点。 - `46:00` 在意识到落后硅谷 AI 演进 60 天后，Nubank 拉响全员“红色警报”，将全公司拆解为 120 个核心工作流进行 AI 原生重写，并将组织汇报层级由 10 层削减至 5 层。 - `34:19` Nubank 正式启动进军美国市场，并面向多达 35 个国家推出基于稳定币支持的离岸金融账户，目标从拉美本土走向全球服务 10 亿用户。 ## Key takeaways - `00:00` 全球银行业仍处于“百视达时代”，90% 的业务依附于传统实体网点，而拉美等新兴市场中仍有数十亿人缺乏正规信贷服务。 - `02:52` 摆脱线下物理基础设施赋予了数字银行近乎不对称的成本优势，Nubank 单客服务成本仅为传统银行的 3%，并可将利润空间直接反哺给消费者。 - `10:00` 传统银行依靠信息差与惰性对普通人征收沉重的“隐形成本”，而大众缺乏的不是金融教育，而是能自动化接管并优化全盘财务的个人 AI 私人银行家。 - `12:21` 纯 AI 原生初创公司虽然在产品代码生成上具备极高迭代速度，但核心银行业务受到牌照、合规周期与海量用户历史数据的刚性制约。 - `14:56` Nubank 通过战略性收购初创公司 Hyperplane，自研了专用于信贷风控与反欺诈的金融垂直基座大模型，大幅降低了组织内部的试错与特征工程成本。 - `23:05` 无现金化与开放金融（Open Banking）正在瓦解传统寡头银行的数据垄断壁垒，使消费者能够真正掌控自身财务数据并获得议价权。 - `25:20` 全球流通的约 190 种法定货币在未来或将缩减至十余种，美元稳定币不仅不会削弱美元霸权，反而会成为全球数十亿人接入离岸资产的核心避风港。 - `28:45` 通过将小微商户交易流水全面数字化，数字银行不仅能够预测信用风险，更能从零创造底层信用，让数千万边缘群体首次获得正规金融支持。 - `32:35` 金融科技的竞争护城河正在重构，拥有庞大客群、金融牌照、丰富数据且积极拥抱 AI 重塑的新兴数字银行处于行业最有利的战略生态位。 - `40:00` 将亚马逊的“客户痴迷”与迪士尼的情感魔法融入冰冷的银行业，是抵抗短期暴利诱惑并构建深厚用户信任的底层支柱。 - `43:15` AI Agent 未来将演化为一个与人类并列的独立客户群体，金融机构需要为其专门设计接口、定制化产品及专属搜索分发机制。 - `44:45` 远程单打独斗的工作形态极易被 AI 快速替代，重返办公室强化高频面对面协同与复杂共识决策构成了人类员工独有的护城河。 - `50:00` 打破逐级上报的指挥链，通过扁平化架构直接穿透一线并构建全公司数据大脑，是防范大企业病与“独裁者困境”的关键举措。 - `50:00` AI 时代的人力资源管理应从传统低效、充斥近因效应的半年评审，转变为类似奥运教练级别的日常、实时、高颗粒度 Copilot 指导。 - `01:00:00` 由创始人主导的企业在面对重大技术范式转移时往往表现更优，因为创始人的危机感、破局意愿与颠覆本能远比职业 CEO 的常规体系化思维更为关键。 - `01:10:00` 在 AI 使得初创技术门槛极度扁平化的时代，顶尖创始人反而应当避开琐碎的浅层痛点，坚定选择挑战深科技与看似“不可能”的宏大难题。 ## Detailed walkthrough ### `00:00` 银行业的“百视达时代”与拉美数字银行的奇迹 五年前接受访谈时，Nubank 拥有 3500 万用户，而如今这一数字已飙升至 1.4 亿。面对这种爆发式增长，外界很容易产生一种错觉：金融科技（Fintech）是否已经见顶，移动优先的数字化转型是否已经进入尾声？然而，在 Nubank 创始人兼首席执行官 David Vélez 看来，事实恰恰相反，全球金融业目前仍处于类似于流媒体普及前的“百视达（Blockbuster）时代”。 全球个人消费金融服务是一个规模高达 10 万亿美元的庞大市场，但目前大约 90% 的业务依然由传统银行把持，并通过线下物理网点完成。全球所有新兴数字银行（Neobanks）与金融科技公司加在一起，在整体市场中的占比还不足 10%。这就好比当年全球 90% 的人还在去百视达租借 VHS 录像带和 DVD，而 Netflix 这类流媒体服务只占据了 10% 的份额。媒体行业经历了从百视达向 Netflix 的跃迁，零售业经历了亚马逊对传统书店的颠覆，出行领域见证了 Uber 的重塑，而在金融服务领域，由传统机构向数字化原生企业的代际迁移才刚刚拉开序幕。 Nubank 在巴西乃至整个拉丁美洲创造了独一无二的先例：它是全球唯一一家在规模上彻底超越本土所有传统巨头的新兴数字银行。在巴西，Nubank 的用户量已接近 1.2 亿，覆盖了全国 65% 的成年人口。更关键的是，约 60% 的用户将 Nubank 作为自己的主要银行账户（Primary Bank Account）。它不再是那种可有可无的零钱钱包，而是彻底取代了传统银行，承载着人们日常的薪资与支付流转。目前，Nubank 正在墨西哥和哥伦比亚复刻这一路径。这不仅代表着存量市场份额的重新分配，更意味着增量市场的巨大扩容——全球仍有近 30 亿人口完全处于现代银行体系之外，无法获得任何正规信贷服务。 - 全球个人消费金融服务市场规模约为 10 万亿美元，传统网点型银行仍控制其中约 90% 的份额。 - Nubank 在巴西服务近 1.2 亿用户，渗透了 65% 的成年人群，且 60% 的客户将其作为首选主账户。 - 全球仍有约 30 亿人未被正规银行系统覆盖，缺乏获得信贷的途径。 ### `02:52` 极度不对称的成本结构：10,000 名员工 vs 100,000 名员工 从百视达向 Netflix 的跨越，第一阶段的驱动力来自于智能手机彻底消除了物理基础设施的开销。脱离实体网点后，数字银行获得了极其悬殊的成本结构优势，并能通过零费率、高收益和更优质的产品体验将这笔效率红利直接反哺给消费者。 这组对比数字展现了极其悬殊的人效差异：Nubank 仅凭 10,000 名员工就支撑起 1.4 亿活跃用户；而作为竞争对手的传统银行，服务 1 亿用户却需要雇用 100,000 名员工，并维持 6,000 家线下营业网点。实体网点带来的租金、人员和运营成本极其沉重，直接导致 Nubank 单个用户的服务成本（Cost to serve）仅为传统大型银行的 3% 左右。即便在全面拥抱 AI 之前，仅仅依靠移动互联网的原生架构，数字银行就拥有了传统机构无法企及的利润垫，从而能够彻底取消账户管理费，提供更具竞争力的贷款利率。 ```mermaid flowchart TD accTitle: Nubank与传统银行服务成本结构对比 accDescr: 对比传统银行因庞大网点与员工产生的高昂成本，与Nubank轻资产模式产生的效率反哺闭环。 subgraph Incumbent["传统巨头银行模式"] A["100,000 名员工 + 6,000 家网点"] --> B["高额不动产与人工运营开销"] B --> C["高服务成本 (100%)"] C --> D["收取高额账户费、低存款收益"] end subgraph NubankModel["Nubank 数字原生模式"] E["10,000 名员工 + 0 线下网点"] --> F["云端与移动原生自动化架构"] F --> G["极低服务成本 (仅占传统银行 3%)"] G --> H["反哺用户：零月费、更优利率、更好体验"] end ``` 而当下正在爆发的 AI 浪潮，将对这一效率模型产生几何级数的倍增效应。在未来，一家银行凭借区区 5,000 名员工服务 5 亿甚至 10 亿用户将不再是天方夜谭。对于雄心勃勃的企业而言，这种效率不仅意味着内部利润的挤压，更意味着向全球扩张的底气——例如凭借数字化与 AI 杠杆直接切入全球 100 个国家的市场，全面加速全球银行业的重构。 - **员工规模与网点对比**：Nubank（1.4 亿用户 / 1 万员工 / 0 网点）vs 传统银行（1 亿用户 / 10 万员工 / 6000 家网点）。 - **单客服务成本**：Nubank 仅相当于传统在位银行的约 3%。 ### `04:53` 全球金融的三大瓶颈：准入、隐形成本与智慧普惠 立足未来三到五年的技术变革，David Vélez 将金融服务领域悬而未决的核心痛点归纳为三个维度：“准入（Access）”、“成本（Cost）”与“智慧（Intelligence）”。 **第一是准入问题。** 全球仍有近一半的人口被排斥在信贷体系之外。即便利益丰厚如墨西哥或哥伦比亚，信用卡渗透率也仅有可怜的 13%。之所以长期停滞，是因为缺乏数字信用数据沉淀、风控审核手段原始、金融系统底层过于落后。但随着现金交易的消亡，以及类似巴西 Pix、印度 UPI 等国家级实时支付网络的普及，新型金融轨道正在迅速铺就。如果未来十年内全球所有民众和小微企业都能顺畅获得信贷支持，对全球 GDP 的拉动将是难以估量的。 **第二是发达国家中被掩盖的隐形成本。** 在美国等发达市场，高收入群体看似享受着完善的服务，但整个金融体系对普通人征收着隐秘而庞大的税负： - **6000 亿美元的“存款财富税”**：全美零售存款总额约为 15 万亿美元，但传统银行给出的年化平均存款收益率仅为 40 个基点（0.4%）左右。在基准利率处于 4.5% 至 5% 的高息环境下，这相当于每年从储户手中转移了约 6000 亿美元给银行，用来补贴那些昂贵的线下网点。普通收入人群的存款收益几乎被剥削殆尽。 - **高昂的跨境汇款与账户月费**：每年从美国流出的约 1000 亿美元个人汇款中，手续费率高达 5% 至 10%。在数字化本质只是“移动比特”的时代，随着充分竞争，这一费用必然趋近于零。此外，欧美日等国消费者每月还需为普通账户支付 5 到 15 美元的账户维护费，这些无谓的摩擦成本都将被新型数字银行彻底抹平。 **第三是金融智慧的民主化。** 过去，只有收入处于金字塔顶端 1% 的超级富豪才能享受私人银行家（Private Banker）的服务。私人银行家会统筹全盘账户、优化现金存款收益、规划最省税的资产架构，并在客户需要借贷时让各家银行竞价以锁定最划算的利率。而剩下 99% 的普通大众不仅得不到任何专业财务指导，反而承受着最苛刻的金融条款。随着 AI 技术的成熟，每一个普通人未来都能在手机里拥有专属的虚拟私人银行家，金融领域的认知特权将被彻底终结。 - 全美约 15 万亿美元零售存款，平均收益仅约 40 个基点，在 5% 利率环境下相当于储户补贴给传统银行近 6000 亿美元。 - 全美年跨境汇款额超 1000 亿美元，传统通道收取 5% 到 10% 的高额费用。 - 墨西哥和哥伦比亚等拉美主要经济体的信用卡渗透率目前仍低至 13%。 ### `10:00` 用 AI 代替“金融教育”：重构 6000 亿美元的存款税与银行利润池 长期以来，业内在讨论大众财务困境时，总喜欢把“缺乏金融教育”当成万能借口，坚信只要多办几场讲座、多给用户上课就能改善情况。但 David Vélez 直截了当地指出，这种思路从未奏效过——普通人绝不会真的坐下来听金融课，教育根本解决不了认知与智力不对称的问题。真正能够改变现状的不是灌输知识，而是 AI。AI 正在全球范围内充当每个人的“超级大脑”，代表用户去自动优化全盘财务生活。 当智能化真正普及后，银行业中原本依靠信息差与惰性获取的高额利润池将被彻底分流消解。传统银行依靠零息或极低利息的存款沉淀来支付庞大线下网点的开销，这笔高达 6000 亿美元的“隐形税收”终将被转交还给普通储户。以墨西哥和巴西的传统大银行为例，资产负债表上约 70% 的存款付息极低甚至完全为零，只有 30% 享受收益（通常仅面向高净值客群或大企业）。在理想的充分竞争市场中，银行本质上是一种同质化商品（Commodity），没有手续费、体验更好的产品理应拿下 100% 的市场份额；但现实中由于巨大的系统惯性，这种流动一直很慢。而当人工智能叠加以 Open Banking（如 Plaid 模式）的技术杠杆，能打通上亿账户时，用户便获得了全景视角的决策能力。 ```mermaid flowchart TD accTitle: AI Driver vs Traditional Bank Inertia accDescr: 展现传统银行沉淀资金与 AI 驱动的开放银行优化机制对比 subgraph Legacy["传统银行模式"] A["大众零息/低息存款<br/>(占比约 70%)"] --> B["沉淀高额利差<br/>(全球约 6000 亿美元)"] B --> C["补贴线下实体网点与高管成本"] end subgraph NubankAI["AI + Open Banking 模式"] D["接入跨行账户与全量数据"] --> E["AI 实时监测账单与利率"] E --> F["自动规避高利息罚款"] E --> G["自动低息再融资借贷"] F & G --> H["利润直接回馈普通消费者"] end ``` 这种降维打击在巴西已经落地。Nubank 每天会向约 1500 万名客户发出预警：系统检测到他们在其他银行的信用卡明天到期，但卡内余额不足，如果不理睬就会被对方银行收取高达每天 15% 的罚息，Nubank 会主动提供资金协助垫付或进行低息再融资；或者发现用户在他行有一笔 10% 利率的贷款，直接提议换成 Nubank 5% 的贷款。虽然小微企业、百万富翁在理财时，或者办理大额房贷等复杂业务时，可能仍有 5% 的场景需要与网点里的真人面对面沟通，但对于大众市场的普通存贷卡需求，传统实体网点已经彻底失去了存在的必要。 - 传统大银行的存款付息结构呈现严重倾斜：约 70% 的大众资金承担零息或极低利率，仅有约 30% 面向大企业和顶级富人的资金能获得合理收益。 - 银行业长期依靠用户的惯性（Inertia）维持高毛利，而智能化结合开放银行（如连接 1.2 亿账户）直接打破了这种壁垒。 - Nubank 在巴西每天对约 1500 万用户进行主动干预，利用降息再融资与垫付策略直接切入竞品利润池。 --- ### `12:21` 初创公司攻打 Nubank 的假想策略与牌照/监管护城河 面对“如果今天把你的一切资源剥离，作为一家 AI 原生初创公司的创始人，你该如何攻击并干掉 Nubank”这个假设，David 坦言，这恰好是 13 年前 Nubank 颠覆巴西传统老牌银行时做过的事。在今天，一家 AI 原生初创企业唯一的绝对优势就是**极致的迭代速度**——能够以极高的效率打造并测试各种产品。在未受监管约束的阶段，团队甚至可以用 ChatGPT 或 Codex 极速重构软件架构。 然而，金融软件与真正的金融机构有着不可逾越的鸿沟：你可以用 AI 快速写出产品代码，但无论如何也不可能让 ChatGPT 帮你“生成”一张合规的银行牌照。金融监管机构绝对不会因为团队使用了大模型，就将两三年的牌照审核流程缩短至六个月。一旦业务从纯软件走向核心信贷与资金结算，任何初创企业都必然会撞上牌照与合规的坚硬墙壁。 此外，Nubank 经历了 13 年的沉淀，积累了 1.4 亿客户，拥有每位客户跨越十余年、多达 30,000 个数据点的庞大资产。初创团队虽然灵巧敏捷，但在数据规模、用户信任度、品牌和监管资质上完全处于劣势。相比纯粹的早期创业公司，Nubank 这类本身具备先进科技底座、同时又掌握牌照、用户与海量数据的公司，反而处于吸收 AI 红利最舒服的“甜蜜点”（Sweet Spot）。 - AI 原生初创公司的核心武器是速度，但在信贷与银行业务中，牌照壁垒无法依靠大模型代码生成跳过。 - 监管审批周期的客观规律（往往需 2-3 年）天然限制了无牌照 AI 初创公司在核心银行业务上的扩张上限。 - Nubank 坐拥 1.4 亿客户与单客 30,000 个维度的历史行为数据，形成了坚固的数据护城河。 --- ### `14:56` Nubank 的 AI 进化史与自建金融垂直大模型 Nubank 并非在 ChatGPT 爆发后才临时抱佛脚。早在 2013 年成立之初，面对巴西消费者承担每年高达 500% 到 600% 的信用卡惩罚性年利率，公司便立志以机器学习和数据分析为核心来定价信贷风险。与许多避开重资产、只做轻量技术撮合的纯软件 Fintech 不同，Nubank 从第一天起就选择将信贷风险纳入自己的资产负债表，做高资本密集的真正银行，因此整个组织的基因天然就是数据驱动的。 三四年前 ChatGPT 掀起大模型浪潮时，Nubank 迅速行动，收购了位于帕洛阿尔托（Palo Alto）的初创公司 Hyperplane。Hyperplane 当时正致力于构建金融垂直领域的基座大模型（Foundation Model），但他们向传统大银行推销时屡屡碰壁，一个销售周期就要拖延整整两年。Nubank 果断将其收入囊中并完成团队整合，借助自身独有的庞大金融数据，构建出全球银行业首个专用于信贷决策与反欺诈的自研基座模型，并公开发布了相关研究成果。 如今，这套自研基座模型不仅支撑着巴西本地的信贷与催收决策，还被无缝扩展至墨西哥和哥伦比亚等市场，反欺诈能力也全面接入。在 Nubank 进军美国等新市场时，这套体系可以在“第一天”直接落地启用。然而 David 也冷静地指出，单纯的“规模”并不能直接转化为胜势：当年百事达（Blockbuster）拥有亿万进店租碟用户的行为数据，但那些数据从未能帮助他们训练出超越 Netflix 的推荐算法；只有将海量数据与真正的技术架构、企业文化结合在一起，规模才能转化为真正的竞争壁垒。 - 2013 年创立伊始，Nubank 便依靠机器学习攻克巴西 500%-600% 极高年利率下的信贷风控难题。 - 战略性收购 Palo Alto 的 Hyperplane，直接化解了其面对传统银行长达两年销售周期的落地痛点，迅速完成了技术内化。 - 自研金融基础模型已成功推广至哥伦比亚、墨西哥及反欺诈等多元场景，成为国际化扩张的标准化基础设施。 - “百事达谬误”警示：没有适配的数字化基础设施与算法文化，单纯的历史用户与数据规模毫无价值。 ### `20:00` 数据文化与底层架构：为何传统银行陷入“大型机泥潭” 如果仅仅拥有数据规模就能筑起护城河，百视达（Blockbuster）当年就不会被 Netflix 击溃。传统巨头倒下的根本原因在于：他们虽然接触海量数据，却缺乏采集并沉淀数据的正确系统，更没有建立起依托数据和算法模型进行决策的企业文化。相比之下，Netflix 最初毫无用户数据积累，却能凭借更优越的算法后来居上。规模和数据固然重要，但底层的系统架构与组织文化具有同等、甚至更决定性的权重。 这种底层基因构成了数字原生银行（如 Nubank）与传统现存银行（Incumbents）之间的鸿沟。传统银行经过数十年频繁的并购整合，内部往往同时跑着 15 套互不连通的异构系统。当今许多传统机构探索 AI 时，最迫切的应用场景竟然是“如何用 AI 把 COBOL 语言重写以摆脱老旧大型机（Mainframe）”。对它们而言，仅仅将系统迁移至云端就是一场长达 10 到 15 年的漫长跋涉。Nubank 自创立之初就将自身定位于“一家恰好身处金融服务领域的科技公司”，从第一天起就彻底拥抱云端与数据决策。 --- ### `21:29` 竞争维度的转变与自研基础模型的实验飞轮 传统银行在物理时代的竞争基石是线下网点密度。过去数十年间，消费者选择银行的核心逻辑极其简单——“哪家银行在我家或公司附近有网点，我就选哪家”。因此，老牌银行将绝大部分资源和精力消耗在不动产布局、线下基础设施维护以及通过品牌广告维系用户信任上。而 Nubank 抛弃了所有线下包袱，将技术与数据作为构建比较优势的唯一支点。 随着公司部署自研的金融基础模型（Foundation Model），内部决策与产品迭代的试错成本呈现断崖式下跌： ```mermaid flowchart TD accTitle: AI模型降低试验成本的工作流对比 accDescr: 对比传统机器学习特征工程与自研基础模型驱动下的低成本自动化实验流程 subgraph Legacy["传统机器学习时代 (耗时数周至数月)"] L1["提出业务假设"] --> L2["耗费数周手动提取并测试特征"] L2 --> L3["花费长达6个月将新特征整合入模型"] end subgraph Modern["Nubank 自研基础模型时代 (低边际成本与自动化)"] M1["生成实验方案<br/>(例如：邮件文案用黄色还是紫色)"] --> M2["自研基础模型并行极速验证"] M2 --> M3["模型替代人工进行端到端自动化决策"] end ``` 在引入基础模型之前，团队如果想在授信或风控模型中测试并上线一个新特征（Feature），往往需要耗费数周乃至长达六个月的时间。而如今，模型能够以极高速度测试海量不同方案。甚至在微观运营层面，测试一封推广邮件究竟采用“紫色字体还是黄色字体”的边际成本都几乎降为零。组织内部的每一项决策，正逐步由过去依赖人工判断转变为由模型端到端自动化执行，使原本就存在的数据驱动文化实现了指数级放大。 --- ### `23:05` 全球金融系统的外生巨变：无现金化与开放银行 抛开 Nubank 自身的微观运营，全球金融体系未来五年最值得关注的外生性变革主要由两大趋势主导： 首先是**现金的终结（The End of Cash）**。在墨西哥、印度尼西亚以及许多非洲国家，现金交易长期占据主导地位（过去的巴西也是如此）。随着各国央行与金融科技推动的实时支付系统普及，现金正迅速向数字化支付迁移。无现金化不仅意味着物理网点需求的急剧萎缩和银行业运营效率的跃升，更重要的是加速了全社会的货币周转速度，并沉淀出海量的数字化交易轨迹，为普惠授信提供了底层基石。 其次是**全球范围内的开放银行与开放金融（Open Banking / Open Finance）**。这一机制的核心逻辑在于确立“用户对自身跨机构金融数据拥有完整所有权”。长久以来，金融、电信（Telecom）以及公用事业（Utilities）等垄断行业最大的护城河便是“数据惯性（Inertia）”——用户与某家机构建立的长年信贷往来无法打包转移，新进入者无法获知其信用历史，老牌机构借此锁死客户。 - **打破数据惯性与自主变现**：开放金融切断了传统巨头的垄断杠杆，用户可以主动将历史数据授权给愿意出资或提供更优利率的新平台，实现数据主权与价值变现。 - **瓦解全球银行业寡头格局**：目前全球银行业普遍呈现高度寡头垄断（Oligopoly）形态，无论在美国、欧洲、亚洲还是拉美，前五六大银行基本控制了全行业 60% 至 70% 的资产。开放银行将瓦解这一高集中度结构，催生更加分散、竞争更激烈的市场环境，最终大幅削减消费者的借贷成本与服务费。 --- ### `25:20` 稳定币与全球离岸体系：从 190 种法定货币到仅存十余种 在长达数十年的演进维度中，加密资产与稳定币（Stablecoins）将深刻重构全球货币格局。随着全球人口流动性与数字化生存程度的加深，未来大量普通人将同时在“本土法币系统”与“全球离岸金融系统”之间无缝穿梭。 许多国家的居民天然缺乏对本土银行体系及本国法币的信任，他们迫切需要接入离岸世界。全球目前流通着约 190 种法定货币，但在未来 10 到 20 年内，由于大量国家糟糕透顶的财政赤字与主权信用恶化，绝大多数弱势法币的实质价值与凭空发行的空气代币无异，“这 190 种货币最终可能会缩减到仅剩大约 10 种”。全球数十亿人口将把资产配置在以美元稳定币和比特币为代表的离岸架构中。 关于美元未来作为全球储备货币的地位，市场存在激烈争论，但美元稳定币很可能成为美元霸权最大的救星： - **美元访问权的民主化**：虽然美国的财政状况并不健康且正在恶化，但比起拉美及其他新兴市场央行资产负债表的脆弱性，美元的信誉仍具有压倒性优势。市场上对美元稳定币有着近乎无限的需求，而几乎没有人想要“巴西雷亚尔稳定币”。 - **多元储备格局成型**：美元稳定币的全球普及将帮助美元在极长周期内巩固其主要储备货币地位。不过它不会是唯一的避风港，未来更可能形成以美元、其他五至六种核心法币、以及比特币为核心的多元化全球资产储备体系。 --- ### `28:45` 墨西哥卷饼摊的启示：数据如何无中生有创造“信用” 无现金化进程中最常被外界低估、却极具变革力量的环节，是它对**底层信贷市场（Credit Markets）**的重构。在广大新兴市场，绝大部分经济活动依旧发生在线下。 > **墨西哥卷饼摊的授信实验** > 在墨西哥城的某个街角，一个流动卷饼摊（Taco stand）常年靠现金做生意。摊主每天卖出大量卷饼，产生着极为稳健真实的现金流。但因为现金交易毫无数字化记录，这些交易流水在发生瞬间便消散于无形。传统银行的风控系统看不到这些数据，摊主在金融体系中是“隐形人”，无法获得任何信贷。 > > 一旦这些交易接入数字账本，哪怕仅是用手机完成扫码收款，摊主的现金流就会被完整沉淀。他可以直接拿着这串数字记录向银行申请贷款：“看我的日常现金流，请给我授信。”获得信贷支持后，这位摊主便有能力将业务扩张至两个摊位，乃至五个摊位。 这种从小微商户数字化交易中衍生出的信贷乘数效应，正在释放数万亿美元级的流动性： - **新兴市场的中小微支柱**：在巴西，全国约 70% 的就业人口由中小微企业（Small Businesses）吸纳，印度和东南亚同样呈现类似格局。这些长期依赖现金结算的微型商业实体，正在随着数字化浪潮首次接入正规金融体系。 - **创造信用（Creating Creditworthiness）而非被动预测**：传统银行长期为自身的懒惰寻找借口，声称底层民众“不需要信贷产品、排斥信用卡、害怕被国税局查税，因而永远不会使用”。Nubank 击碎了这种偏见——在拉美地区，Nubank 已经为超过 3000 万人发放了他们人生中的第一张信用卡。通过数字账本记录真实交易，再结合先进算法与定制化金融产品，平台不仅能预测违约率，更是在从零为底层民众构建信用资产，驱动广泛的阶层向上流动。 ### `30:00` 从 20 美元信用额度起步与跨国模式的可复制性 传统银行在面对没有信用记录的人群时往往直接关上大门，但 Nubank 意识到这些人并非不可信，而是市场上缺乏一款真正匹配他们需求的产品。Nubank 最开始为许多用户提供的第一笔信用卡额度仅仅只有 20 美元。当时传统同行看后觉得极其荒谬，但用户却非常感激，因为 20 美元远比“零”有价值得多。随着时间推移，这 20 美元的信用额度逐渐提升到 100 美元，再一步步增长至 1000 美元，用户便借此建立起了完整的信用历史。在此基础上，Nubank 能够接入 AI 智能副驾驶（Copilot），充当用户的财务教练，指导其如何规划消费、储蓄与投资来积累积分、提升信用评级，从根本上改善个人的财务状况与信贷获取能力。 在谈及跨国拓展时，不同国家之间的业务经验比想象中更具可复制性，但也会遇到意料之外的阻力。最大的差异在于巴西是一个“已银行化”（banked）程度相对较高的市场，而当 Nubank 刚进入墨西哥时，听到的普遍反馈是墨西哥人深爱现金，根本不信任数字支付、信用卡或借记卡。David Vélez 形象地比喻道： > “我常说我们在数字水域里是一条鲨鱼；但如果要被迫在有物理网点的线下世界竞争，我们的竞争力就会弱得多，因为我们既没有实体分行，也没有 ATM。” 在早期缺乏线下基础设施的情况下，Nubank 在墨西哥确实经历了数年艰难竞争。然而如今 Nubank 已经成为墨西哥第二大银行以及当地最大的数字银行。随着墨西哥等国数字化进程的加速，各国市场正变得高度趋同。就像全球用户使用的社交软件、Netflix、Spotify、Uber 和 Amazon 几乎毫无二致一样，金融服务未来的面貌也会远比过去更加同质化。 ```mermaid flowchart TD accTitle: 从微额信贷到主流金融服务 accDescr: 展示 Nubank 如何通过 20 美元起步的信贷产品帮助未被传统银行覆盖的人群建立信用历史，并结合 AI 教练逐步扩大授信与服务。 subgraph Inclusion["建立信用起点"] A["20 美元初始信贷额度"] --> B["按期还款与数据沉淀"] B --> C["额度提升至 100 美元及 1,000 美元"] end subgraph Growth["AI 赋能与全球扩展"] C --> D["AI 财务副驾驶 (Copilot)"] D -->|"指导消费、储蓄与投资"| E["综合信用画像健全"] E --> F["数字银行核心客群"] end ``` --- ### `32:35` Fintech 投资逻辑的演变与初创企业的突围空间 知名投资人 Mickey Malka（Ribbon Capital 创始人）曾在几年前提出“金融科技已死”（Fintech is dead），并将精力大量转向 AI、算力与数据基础设施。David Vélez 透露自己近期也与 Mickey 探讨过这一话题，询问他是否真的放弃了 Fintech。Mickey 的实际回答是并未放弃，而是选择将筹码集中押注在那些规模庞大的新兴银行（Neobanks）与新金融科技巨头身上。这类企业正处于绝佳的战略甜点区：一方面能够利用 AI 实现自我重构，另一方面已经构筑了规模效应、海量数据、品牌信任度以及稀缺的金融牌照等多重护城河。 对于新入局的 Fintech 初创公司而言，未来的生存空间发生了明显转变： - **切入路径更加隐蔽**：直接面向 C 端从零打造一个拥有数亿用户的消费金融品牌变得极其艰难，除非能拿出颠覆性的底层创新。 - **立足于 AI 垂直应用**：初创团队的突围机会更多在于构建在底层大模型之上的特定金融应用层，通过高度聚焦的垂直场景切入市场。 --- ### `34:19` 进军美国市场与推出覆盖 35 国的稳定币账户 尽管外界仍普遍将 Nubank 视为一家立足于巴西或拉美的区域性数字银行，但从其 2013 年的第一版商业计划书（Seed deck）来看，该公司的愿景从一开始就属于全球层面。其核心假设是：全球零售银行业终将被数字原生企业重塑，因为数字原生平台能够以更低的成本、更广泛的触达和更智能化的手段彻底解决金融服务的痛点。过去 13 年里，Nubank 在拉美完成了模式验证：不仅成长为整个拉美大陆最大的私营金融机构，拥有最高的净推荐值（NPS），在巴西市场更取得了全行业顶尖的净资产收益率（ROE）。 现在，Nubank 正式宣布进军美国市场，并面向多达 35 个国家推出基于稳定币（Stablecoin）支持的账户。对于美国市场，David Vélez 分析道： - **看似红海，实则陈旧**：美国虽是全球最大且竞争激烈的市场，聚集了众多优秀的 Fintech 公司，但市场绝大部分份额仍被少数几家传统巨头及社区银行牢牢掌控。 - **“百视达时代”的银行业**：绝大多数美国银行业务依然依赖线下实体网点进行，“就好像他们的影视娱乐依然发生在百视达（Blockbuster）录像带租赁店一样”。传统银行必须收取高昂费用并支付极低的存款利息，以维持庞大物理网点的运维。 - **Nubank 的破局点**：利用大众市场信贷风控能力、全球化支付网络以及痴迷于用户的服务体验展开降维打击。 在美国市场面临的最大挑战是如何以极高的资本效率建立起强大的有机品牌。美国消费者高度依赖品牌认同来挑选主银行账户，Nubank 尽管资金充裕且盈利强劲，但并不打算通过盲目铺投广告来烧钱获客，而是需要通过更聪明的策略赢得用户信任，成为其主力银行账户。 --- ### `38:00` 衡量金融模式的底层指标与对标亚马逊的“客户痴迷” 如果面向所有覆盖 Nubank 的买方分析师，David Vélez 认为衡量其商业模式优劣应关注一套环环相扣的指标体系： 1. **规模与产品契合度**：关注用户总数增长（User Count）与净推荐值（NPS），这反映出产品是否击中了未被满足的市场痛点，并提供了独一无二的体验。 2. **增长的健康度**：关注获客成本（CAC）与用户生命周期价值与获客成本比率（LTV/CAC），确保增长并非单纯靠营销费用堆砌而来。 3. **结构性成本优势的终极体现**：净资产收益率（ROE）。传统银行收取繁杂的手续费，而 Nubank 几乎完全不收手续费、单客收入更低，但其 ROE 却能达到传统银行巨头的 2 到 3 倍。这种极致的盈利能力直接证明了技术带来的底层成本结构优势，形成了极深的护城河。 令人意外的是，部分美国传统银行的 ROE 比许多新兴市场银行还要高。这种现象恰恰印证了亚马逊创始人杰夫·贝佐斯（Jeff Bezos）的名言——“你的利润率就是我的机会”。传统银行在背负沉重线下运营成本的同时仍维持着极高回报率，意味着它们留给颠覆者巨大的降价与让利空间。 谈及创业灵感与商业偶像，David Vélez 强调 Nubank 的诞生源于他在巴西办银行卡时遭遇的极端痛点。在意识到全巴西绝大多数普通民众都在忍受同样的折磨后，他将这种对消费痛点的愤怒转化为了企业使命。因此，Nubank 最为推崇那些具有“客户痴迷”（Customer Obsession）特质的企业，其中亚马逊自始至终是其最重要的学习榜样——唯有打造出极致的产品与服务，才能在原本冷漠疏离的银行业中激发出用户近乎狂热的忠诚度。 ### `40:00` 迪士尼的“魔法体验”与银行业的反直觉结合 大卫·韦莱斯（David Vélez）透露，Nubank 团队曾深入细致地研究过迪士尼。对于一家金融科技公司而言，这听起来可能有些不可思议，但 Nubank 的核心野心正是要在最枯燥无味的银行产品中，为用户创造出意想不到的“魔法体验”。人们通常认为银行业务冰冷繁琐，而迪士尼的精髓恰恰在于与消费者建立情感共鸣，并营造纯粹的奇迹感。 当游客踏入迪士尼乐园时，从入口通道的设计开始，整个人就像瞬间穿越进了一座梦幻王国；乐园里的每位员工对年幼孩童关怀备至，巡游表演与夜空焰火更是将造梦艺术推向极致。迪士尼全流程都在推敲一件事：如何让产品本身具备魔法属性？韦莱斯将这种思考方式注入了 Nubank 的团队文化——哪怕只是“转账付款”这样平淡无奇的操作，团队也必须死磕如何将其打造成惊喜时刻。例如，当 Nubank 开始向美国用户寄送实体卡时，他们对首张信用卡和借记卡的欢迎礼盒（welcome kit）进行了近乎偏执的打磨，力求让客户在开箱的瞬间感受到仪式感与冲击力。 ### `41:45` 客户至上的真正代价：抵抗短期利益的诱惑 如今“以客户为中心”（Customer Obsession）已成为科技圈人尽皆知的口号，亚马逊将其发扬光大，但韦莱斯指出，言行一致的代价极其高昂。任何企业都能把漂亮诱人的口号印在办公室墙上，真正的试金石在于：当践行这一价值观变得极其不便、甚至与切身财务利益直接冲突时，你是否依然能坚持到底？ 现实中，绝大多数公司自称深爱客户，可一旦发现巧立名目增设一项收费就能轻松赚得盆满钵满时，对客户的执念便立刻让位于短期利润。不仅传统企业如此，当前的 AI 狂潮同样陷入了这一误区。业内充斥着如何利用 AI 降本增效、优化内部流程的狂热讨论，却极少有人探讨如何用 AI 提升用户体验。韦莱斯直言，那些沉醉于技术本身的“AI 极客”（AI bro）和“科技新贵”（tech bro）往往更痴迷于技术有多酷炫，而不是普通人的实际感受。在抗拒暴利诱惑、坚守长期主义这道坎上，人类真正的同理心依然能够击败单纯追求效率的算法。 ### `43:15` AI Agent 作为全新客户群体的崛起 谈及未来的交互形态，韦莱斯提出一个颠覆性的判断：AI Agent（智能体）未来将成为一个与“人类消费者”同等重要的独立客户群体。正如 Nubank 过去针对未成年人用户（under-18）或小微企业（SMEs）设立专门业务线一样，公司目前已经组建了专属团队，专门研究 Agent 这一新型群体的诉求，并为其打造定制化产品、构建面向 Agent 的专属营销网络，甚至探索针对 Agent 的新型搜索引擎优化（SEO）。 然而，这并不意味着银行可以彻底抛弃面向人类的品牌建设。金融的底层逻辑始终是信任（Trust），用户必须确切知道自己的毕生积蓄存放在何处，这种安全感需要极强的人性维系。因此，未来的金融市场将呈现分层演进的混合生态： - **信任敏感型产品**：人类依然保留最终决策权，品牌声誉与情感连接不可或缺。 - **高频委托型产品**：深度授权给 Agent 代劳。 - **开放式平台生态**：在 Nubank 拥有庞大渗透率的市场（如巴西本土已覆盖 65% 的成年人口），Nubank 自身就能演化为面对最终用户的超级 Agent。用户可以直接在 App 中向 Nubank 助手寻求财务建议，而该助手采取完全开放的架构，哪怕最符合用户利益的是第三方竞争对手的产品，也会客观推荐。 ```mermaid flowchart LR accTitle: Nubank 双轨客户与 Agent 生态架构 accDescr: 展示 Nubank 如何同时服务人类消费者与 AI Agent 客户群体 subgraph CustomerTypes["核心客户细分"] H["人类消费者<br/>(聚焦品牌信任与资产归属)"] A["AI Agent 群体<br/>(专属产品、API 与 Agent SEO)"] end subgraph Platform["Nubank 金融服务中枢"] OpenAgent["开放式财务智能体<br/>(客观匹配自营与第三方产品)"] TrustEngine["品牌与安全基础设施"] end H --> TrustEngine A --> OpenAgent TrustEngine --- OpenAgent ``` ### `44:45` 重返办公室：AI 时代面对面协作的不可替代性 回顾过去五年的发展轨迹，韦莱斯谈到了一个反直觉的认知转变：从全面远程办公重回混合办公（Hybrid），并大幅增加面对面的工作时长。五年前包括 Nubank 在内的整个行业都在推崇远程办公，但实践表明，物理空间的面对面互动具有无法替代的战略价值。 更深层的逻辑在于 AI 对个人工作形态的颠覆。员工一个人坐在家里对着屏幕完成的独立工作，其本质和逻辑正是 AI 最擅长、最容易接管的领域。如果员工长期单打独斗，本质上是在与进化极快的 AI 同台竞技，而这场对决人类注定会输。相反，身处办公室时发生的高频协作、复杂共识推进以及高带宽的协同决策，才是人类独有的护城河。为了不让人力在居家孤岛中被 AI 侵蚀，回归物理空间、强化集体智识碰撞成为了必然选择。 ### `46:00` “代码红色警报”：Nubank 的 AI 觉醒与 120 项工作流重塑 在组织内部拥抱 AI 的过程中，Nubank 曾经历过一段惊心动魄的转折。Nubank 的核心系统广泛使用函数式编程语言 Clojure。在 2025 年的大部分时间里，团队不断用各种主流代码大模型测试 Clojure 开发，结果一团糟，连公司最资深的工程师都断言“这玩意儿根本行不通”。 转机出现在 2025 年 12 月，随着 Claude 3.5 Sonnet / Opus 等新一代模型的面世，行业的技术拐点突然降临。Nubank 内部最初并没有立刻捕捉到这一质变，整整滞后了 60 天。直到韦莱斯与几位合伙人飞赴加州，与大量硅谷早期创业者深度交流后，才猛然意识到世界已经彻底改变。他们当场调出最新的大模型对 Clojure 代码进行实测，惊骇地发现模型如今完成 Clojure 的代码审查（Code Review）竟然游刃有余。面对落后于顶尖同行 60 天的现实，韦莱斯内心直呼“Holy shit”，随即全速飞回巴西，在全公司范围内拉响了“红色警报”（Code Red）。 回到总部后，Nubank 全面启动了自我颠覆进程： - **全面盘点业务链**：将全公司拆解为 120 个核心工作流（Workflows），涵盖软件编码、产品设计、KYC 实名认证、内部审计、合规审查以及监管资本管理等所有环节。 - **AI 原生重构**：不再修修补补，而是要求每一个工作流从第一天起以 AI 原生方式重写。 - **搭建 Eval 评测基建**：建立严密的评估基础设施（Eval infrastructure），在重塑的每一个节点量化衡量效率提升与准确率。 韦莱斯强调，许多所谓的效率暴增表面上归功于 AI，但背后更大的红利在于：逼迫团队停下惯性脚步，借 AI 之名彻底铲除多年积累的组织冗余和陈旧规程。 ### `48:30` 扁平化重构：打破“独裁者困境”与打造公司大脑 这场由 AI 触发的剧变，最终演变为一场彻底的管理架构革命。在用户规模从 100 万一路狂飙至 1.4 亿的过程中，Nubank 难免患上大公司病——每当业务快速膨胀、“后院起火”时，管理层最省事的应对方式就是招募中层管理者来灭火。不知不觉间，公司的汇报层级多达 10 层，组织极度垂直化，而韦莱斯本人的直接下属竟一度被压缩到仅有 1 人，最高决策层与一线真实情况完全脱节。 韦莱斯引用了 Shopify 创始人托比·吕特克（Tobi Lütke）对“独裁者困境”（Dictator's Dilemma）的描述：层级越深、最高领导者手中的权力越大，信息经过层层过滤与美化后，最高层反而越难获悉前线的“基本事实”（Ground Truth）。为此，Nubank 借由 AI 工具大幅拓宽管理跨度，打响了组织扁平化战役： - **组织结构大瘦身**：将原先臃肿的 10 个管理层级大刀阔斧削减至 5 层。 - **扩大管理跨度（Span of Control）**：韦莱斯的直接汇报人数从 1 人直接跃升至 20 人，组织运转从纵向垂直全面转向横向高效协同。 - **构建“公司大脑”（Company Brain）**：建立起贯通全局的情报与数据系统。该系统兼具主动反馈与被动响应能力，能将 AI 转型进度、新产品上线指标、前线核心数据直接、无损地推送到韦莱斯面前。 - **重塑穿透式沟通习惯**：打破“按汇报线层层上报”的大企业病惰性，管理者被要求主动越过中层，直接与正在从事一线生产的员工对话，让战略意志和现场实情实现零时差对齐。 ### `50:00` 打破层级指挥链与 AI 对企业 HR 的重塑 在谈及组织运作时，David Vélez 强调传统机构往往受制于严格的“指挥链”（chain of command），管理层认为问询必须逐级向上汇报，而不能越级直接去找一线干活的人。但他作为 CEO 的理念截然相反：如果对某件事有疑问，或者发现业务推进受阻，他会直接去找掌握最优质信息的人共同解决，绝不拘泥于层级规矩。这种扁平穿透的模式，既依赖技术带来的信息透明，也取决于管理层是否愿意深度扎根于组织基层的文化和态度。 面对如何利用技术与数据拔高全员人才门槛、建立敏捷绩效文化的问题，Vélez 指出这正是一场由 AI 驱动但极少被行业深入讨论的人力资源变革。在传统企业中，员工往往每半年才从主管那里获得一次反馈，这种考核不仅充斥着近因效应（往往只看评定前最后 30 天的表现），而且建议空泛、缺乏指导意义。他形象地比喻道：“试想如果用这种方式去训练一名奥运选手，那简直荒谬绝伦，他们绝不可能成功。” 顶尖运动员需要的是每天即时、具体的教练指导，而 AI 正赋予每个普通员工拥有“奥运级专属教练”的机会。 Nubank 正在将其内部 HR 运营体系重构成一套全天候反馈机制：无论是工程师、设计师还是产品经理，系统都会根据其角色提供针对性建议。比如代码提交（pull request）时，AI copilot 不仅会评估工具的使用情况，还会提示如何更高效地优化代码；设计师也能获得具体的交付改进方案。对于背负极宽管理幅度（span of control）的主管而言，以往那些效率极低、信息密度薄弱的每周一对一（1-on-1）会议也将被革新——管理者通过全量仪表盘即可实时掌握全员动态，从而提供真正高价值的辅导。 ```mermaid flowchart TD accTitle: AI-Driven Performance Feedback vs Traditional Review accDescr: 传统半年一次且充满近因效应的低效反馈，转变为 AI 驱动的日常、实时、高颗粒度教练系统。 subgraph Traditional["传统组织反馈模式"] A["每 6 个月主管评审"] --> B["严重受制于最近 30 天表现<br>（近因效应）"] B --> C["抽象、缺乏可操作性的建议"] end subgraph AINative["AI 驱动的原生反馈模式"] D["实时代码 PR / 设计输出 / 业务指标"] --> E["AI Copilot 每日即时分析"] E --> F["类似奥运教练的日常精准指导<br>（高信息密度 & 行动指南）"] E --> G["赋能管理者：全局数据透明化，告别低效 1-on-1"] end ``` ### `52:45` 领导者的技术敏锐度与“亲自下场造产品” 关于如何紧跟突飞猛进的 AI 浪潮，Vélez 分享了自己的日常习惯。除了频繁浏览社交平台 X（Twitter）获取一手行业动向，他还会借助自建的 AI Agent 筛选、整理技术细节的高价值摘要。但更关键的一点在于：他在日程表中硬性划出固定时间，亲自上手体验工具并动手开发产品。 Vélez 坦言自己多年未写代码，但近期亲自用 AI 工具写了一款非托管型加密货币 App（non-custodial crypto app）——这是 Nubank 内部探讨已久的项目，他选择直接下场验证。Nubank 还要求所有高层管理者每个季度都必须亲自动手做出实际的软件或产品。Vélez 解释说，这并不是要求 CEO 把 70% 的时间用来敲代码（专业工程师显然效率高得多），而是在技术指数级演进的当下，如果管理者不真正去用、不去构建，就无法切实感知技术迭代的真实速度。 ### `55:00` 驱动力演进：战时将军与 10 亿用户的雄心 回首从初创至今的心路历程，Vélez 坦言自己的驱动力始终源于“创造真正的社会影响力，做真正有意义的事”。在 Nubank 经历爆发式增长后，他也曾陷入过许多创始人都经历过的反思期：自己是否该在 Nubank 之外做点别的事情？经过一两年的审视与权衡，他得出了明确结论——世界上没有任何事情的影响力能够比得上把 Nubank 推向新高度：从现有的 3 个国家扩张到 100 个国家，从 1.4 亿客户扩展到 10 亿客户，在全球范围内彻底颠覆低效的传统金融，让数十亿人享受到现代金融服务的红利。 这种业务规模的扩展不仅是对外的挑战，更是内在自我认知的重塑。当重大的技术范式转移（如 AI）席卷而来时，所有企业都被迫站队，选择成为颠覆者还是被颠覆者。Vélez 直言自己骨子里是个“战时将军”（wartime general）而非“平时将军”（peacetime general）。世界的剧烈变动不仅不会让他退缩，反而会瞬间激发他的战斗欲与能量，促使他带领团队在不确定性中全速前行。 - **核心动机**：追求实质影响力，让所投入的时间与精力产生意义。 - **长期视野**：将业务从拉美拓展至全球 100 个国家，跨越至服务 10 亿级用户大关。 - **心力状态**：在颠覆性技术冲击与激烈竞争中更具活力，天然适应战时领导风格。 ### `57:30` 克服冒名顶替综合征：创始人与 CEO 的内在博弈 伴随公司体量膨胀，Vélez 也曾面临深刻的自我怀疑：在创立 Nubank 之前，他从未管理过任何人，如今却突然成为拉美规模最大的银行之一的掌舵人。最初他以为公司需要从外部引入大量资深职业经理人，Nubank 也确实招揽了一批成熟管理者。但他很快意识到这并非企业最核心的解药——组织真正需要的，恰恰是创始人与生俱来的特质：极度清晰的愿景引领、鼓舞士气的能力、敢于做疯狂大梦的魄力，以及带领大家挑战“看似不可能的任务”的勇气。 他反思了长期伴随自己的“冒名顶替综合征”（imposter syndrome）以及持续自我发问的焦虑：自己是否逼得太紧、要求太高？历经 13 年的淬炼，他终于坚定了信心：“逼得紧”正是创始人的职责所在。企业必须有人去打破舒适区、探索极限，如果越界，组织自然会产生反馈与平衡。 最后，Vélez 深刻拆解了“创始人”（Founder）与“CEO”这两种角色的本质冲突： - **创始人（Founder）的底色**：充斥着强烈的急迫感（impatience）与危机感，追求极致的推进速度，永远在全速狂奔。 - **CEO 的底色**：需要耐性（patience）与全局观，注重决策的系统性与科学性，擅长通过“系统化思维”（systems thinking）搭建可扩展的制度和流程。 这两种截然相反的能量在同一个领导者身上常常会产生内在对抗，平衡这两极是带领超大规模企业持续创新的最大挑战。 ### `01:00:00` 创始人模式与 CEO 模式的融合 在谈及组织管理时，David Vélez 坦言创始人通常并不是天生的优秀管理者。创始人往往缺乏足够的耐心与意愿去坐下来进行一对一沟通（1-on-1）或对团队进行细致入微的辅导培训；他们天生渴望往前冲锋，在行为方式上往往更像独立贡献者（IC）而非传统管理者。相比之下，职业 CEO 通常是更优秀的管理者，因为 CEO 的核心思维方式是从系统视角出发去构建制度与体系。然而，David 认为破局的关键在于意识到“创始人”与“CEO”这两种角色并非水火不容，领导者必须学会在两者之间自我重塑并兼收并蓄——在必要的时候保持足够的耐心，但从根本上依然要让“创始人特质”占据主导地位。 这也是为什么由创始人领导的企业在数十年跨度内往往能持续跑赢市场的原因。尤其当面临重大的技术范式转换时，创始人身上那些打破常规的本能与核心资产会发挥出无可比拟的优势，这远比成熟 CEO 所擅长的常规管理优势更加关键。 --- ### `01:01:33` 冒名顶替综合征与最脆弱的低谷时刻 回顾 Nubank 的创业历程，David 承认规模化扩张极其艰难，尤其是领导者必须在自身知识匮乏的全新领域中不断拍板决策。在公司初期，他决心打造一家顶尖的科技公司，但他自己并不是软件工程师出身；在完全缺乏工程背景的情况下，如何识别并招募顶尖的 CTO、如何搭建世界级的技术研发团队，成了巨大的考验。在信贷与金融服务领域也是如此，由于缺乏深厚的底层信贷经验，他对团队构建所做的诸多决策一度充满了内部自我怀疑。 这种经历让他饱受“冒名顶替综合征”（Imposter Syndrome）的折磨。他曾反复叩问自己：是不是应该从传统大银行或大型跨国机构挖一个经验丰富的高管来执掌帅印？随着 Nubank 这辆跑车变得愈发精密、在赛道上疾驰得越来越快，“自己究竟是不是那个有资格继续握紧方向盘的驾驶员”，成了他信心中最动摇的至暗时刻。 --- ### `01:02:40` 商业成就与重返故土的情感高潮 在整个商业发展历程中，最具情绪冲击力的时刻莫过于在规模上逐一超越巴西本土的传统巨头银行。在创业的前十年里，几乎所有本地专家和业内人士都在不断向他们灌输：与传统大银行竞争是天方夜谭，Nubank 必将被巨头无情碾碎，因为那些老牌银行拥有本地生态中最坚不可摧的品牌信任度，初创企业绝无生还可能。然而，最终 Nubank 一路成长为巴西最大的私营银行，实现这种逆转所带来的情绪震撼是无与伦比的。 而在个人情感层面，最深刻的高光时刻是 Nubank 进军哥伦比亚市场： - **童年创伤与流亡**：David 出生于哥伦比亚，8 岁时由于当时哥伦比亚严峻的社会治安与暴力局势，举家被迫逃离祖国，迁往哥斯达黎加生活。 - **荣耀归乡**：哥伦比亚是 Nubank 拓展的第三个国家市场。在哥伦比亚正式上线当天，Nubank 与哥伦比亚总统共同举办了盛大的发布仪式。 - **民众共鸣**：发布会进行现场直播时，线上吸引了超过 10 万名哥伦比亚民众同时在线观看。能够在自己故土的同胞面前，带来一款真正能够造福该国人民的金融产品，这一刻令他无比动容与自豪。 --- ### `01:03:45` 卓越领导力的本质：在革命者与共识之间寻找平衡 对于如何定义卓越的领导力，David 提出了深刻的洞见：领导力始于**前瞻性愿景**（Vision）——领导者必须能洞察到极少数人才能看清的未来趋势，并拥有向大众兜售并推销该愿景的能力。推销愿景的本质是激励人心，因为在大多数情况下，宏大愿景听起来都如同天方夜谭，常人的第一反应必然是否定，并理直气壮地宣称这“不切实际”（not realistic）。但人类历史上每一项意义非凡的伟大演进，在萌芽之初都被所谓的专业人士贴上过“不切实际”的标签。 ```mermaid flowchart TD accTitle: 卓越领导力的双重平衡模式 accDescr: 展示领导者如何在盲目颠覆共识与屈从于现状之间寻找演化平衡 subgraph Extremes ["两个极端陷阱"] A["极端革命者模式<br/>（完全否定一切共识/试图重造所有轮子）"] B["共识顺从者模式<br/>（屈从于现状偏差/认为变革不切实际）"] end subgraph GreatLeadership ["卓越领导力核心：中间道路"] C["深刻理解现状背后的演化合理性<br/>（明确什么是不可更改的底层基石）"] D["坚决对抗陈腐观念与经验主义<br/>（强力推进那些必须被打破的变革）"] E["坚定信念 + 谦逊倾听<br/>（承认自我局限，随时准备纠错）"] end A -.->|"避免极端"| GreatLeadership B -.->|"避免极端"| GreatLeadership C --> E D --> E ``` 然而，优秀的领导者还必须具备积极倾听的能力与谦逊的姿态，清醒认识到自己的愿景并不完美。卓越的领导力往往需要在两个极端之间走钢丝： 1. **盲目颠覆一切的极端革命者**：彻底认定现有共识全是错的，试图全盘推倒重来，重新发明一切。 2. **被共识吞噬的平庸顺从者**：对现状妥协，认为突破现状是不现实的，世界理应维持原样。 David 指出，现状之所以能够存在，在生物进化论视角下是因为这种模式在历史竞争中击败了其他模式并生存了下来，它必定包含着某种优越性与存在的理由。伟大的领导者必须洞察现状中“哪些是历经考验因而无需改变的”，同时拥有铁一般的信念去强力推动“那些必须改变的事情”。 --- ### `01:05:40` Nubank 未来五年的悲观场景与万亿级乐观蓝图 在展望 Nubank 未来五年的发展格局时，David 详细拆解了业务发展光谱两端的极端推演： #### 悲观场景（Bear Case） - **区域饱和瓶颈**：业务局限在拉美现有的三个国家，巴西本土市场完全触顶。目前 Nubank 已经覆盖了巴西 65% 的成年人口，“巴西人快要不够用了”；而当地生育率又无法短期逆转，客户自然增长陷入停滞。 - **国际化折戟**：无法将商业模式成功复制到拉美以外更广阔的全球市场，国际化战略以失败告终。 - **错失 AI 红利**：未能将前沿 AI 技术深度转化为组织效率与产品壁垒，失去借力底层技术弯道超车并支撑跨国扩张的动力。 #### 乐观场景（Bull Case） - **承接传统银行份额的长期大势**：在接下来的两到三十年里，全球银行业巨头把持的市场份额将发生历史性大转移，数字原生机构将迎来爆发期；实时支付网络、开放金融（Open Finance）与 AI 的飞跃将进一步加速这一过程。 - **跨国平台的万亿级红利**：全球银行业坐拥 10 万亿美元规模的利润池，其中 90% 仍被传统金融机构垄断。Nubank 在 AI 介入前就拥有比传统对手低 50 倍的运营成本结构；若能顺应技术东风，Nubank 将把这种极致效率以“更低费率、更高收益、数字化智能咨询”的形式返还给用户。 - **全球普惠金融的实现**：搭建起跨越 100 多个国家的全球化多币种平台，解决全球 30 亿无信贷人群的金融痛点，推动现金彻底数字化，完成从区域独角兽向全球超级银行的跃迁。 --- ### `01:08:15` 创业的核心悟道：无可替代的极致关切与使命感 在谈及创业路上最宝贵的人生领悟时，David 强调了如何迅速摆脱“冒名顶替综合征”的心路历程。他认为，作为那个从零开始孕育了这家公司的创始人，是你赋予了企业最初的生命与远见；正是这种“造物者”的身份，会让你对这家公司产生一种旁人无法企及的极致关切——“甚至达到了一种令人痛苦（miserable）的程度”。 就像父母对待自己的亲生骨肉一样，世界上没有任何职业经理人能够像创始人那样深沉地在乎这家公司的生死与成败。正是这种毫无保留的心血、极致的专注与源源不断的能量，超越了所有的履历经验、专业背景或行业知识，最终让你在极为漫长的时间维度里，始终成为掌舵这家组织最合适的人选。 这也对创业者提出了终极要求：必须对所选择的使命抱有发自肺腑、源自直觉的热爱。能够找到一项让自己真正充满激情的崇高使命，是一件极其奢侈的幸事；因为规模化扩张企业的道路注定荆棘密布，唯有这种深入骨髓的使命感，才能支撑着创业者在漫长的至暗岁月中走得足够坚韧、足够长远。 ### `01:10:00` 寻找创业使命：为什么 AI 时代反而应该挑战“不可能” 创业的真实历程远比外界任何人的预期都要痛苦得多。大卫·韦莱斯（David Vélez）直言，如果创始人对所背负的使命缺乏极高维度的坚定信念，在遭遇接踵而至的绝境与折磨时，就很难支撑着走下去。当帕特里克（Patrick O'Shaughnessy）问及应当如何寻找那项真正适合自己的使命、是否有某种明确的信号可以验证时，大卫坦言，这其中确实存在一定程度的机缘巧合与运气（serendipity and luck），但更关键的是要保持主动探寻、敏锐倾听内心，去感受自己是否真正深切在乎这个项目和它所要解决的问题。 大卫分享了判断使命是否正确的核心信号：它必须是一件让你魂牵梦绕、夜不能寐的事情；你满脑子都在思考它，“甚至无法想象如果不放手一搏去解决它，自己的生活该如何继续过下去”（"You cannot imagine your life continuing if you do not decide to tackle it"）。 谈及当下的时代背景，大卫提出了一个发人深省的观察：AI 的爆发让寻找真正有价值的使命变得更难，而非更容易。由于大模型和现代化开发工具将创办一家公司的技术门槛降到了历史最低点，创业者极易陷入走捷径的诱惑，去追逐那些浮于表面、容易解决的琐碎痛点。也正因如此，当今视野最开阔的一批顶尖创始人反而开始逆向而行，集体转向那些看似彻底“不可能”的硬核边界——比如深科技硬件（hardware）与航天太空（space）。他们主动寻找那些极其艰难且对自己独一无二的挑战，直至产生一种无可替代的笃定：这项宏大难题，唯有自己才能真正攻克。 ```mermaid flowchart TD accTitle: 寻找创业使命与 AI 时代的创业分化 accDescr: 对比浅层创业与顶尖创始人追寻高难度使命的逻辑差异 subgraph Conventional ["AI 门槛降低带来的陷阱"] A["创办公司门槛极度降低"] --> B["追逐易于解决的浅层问题"] B --> C["缺乏使命信念与独特性，难以抵御痛苦"] end subgraph Frontier ["顶尖创始人的使命锚点"] D["夜不能寐的思考与执念"] --> E["直面硬件/航天等看似'不可能'的难题"] E --> F["确信唯有自己能够真正解决，穿越长期周期"] end ``` - **使命的信号**：当你因为持续思索某个痛点而无法入眠，并且认定不去做这件事人生就无法释怀时，说明你找到了值得投入生命的课题。 - **AI 时代的创业悖论**：启动成本的断崖式下跌导致大量创业者扎堆在低门槛、易解决的问题上，反而稀释了真正深沉的使命感。 - **顶尖创始人的新趋向**：硬件、航天等重度投入、看似不可逾越的领域，正成为最优秀头脑的避风港与练兵场，因为真正的壁垒只存在于那些高度困难且具备个体独特性的挑战之中。 --- ### `01:10:45` 播客结语与赞助致谢 帕特里克对大卫的这番总结表达了高度赞赏，认为以“追寻不可能的独特使命”作为全篇收尾极其有力且充满启示，并向大卫的长谈分享致以由衷感谢；大卫也对本次深入对话表达了由衷的喜悦。 节目尾声介绍了 Colossus 平台的会员专享内容，包括手工精心校对的逐字稿深度访谈与纸质季刊，并致谢了本期节目的赞助品牌： - **Ramp**：将企业支出管理系统转化为智能资本配置战略，通过更优质的数据帮助企业做出明智决策并优化长期财务表现。 - **Vanta**：提供全自动化合规审计与单点信任中心，伴随企业业务体量扩展并控制安全风险。 - **WorkOS**：为 OpenAI、Cursor、Perplexity 等顶尖 AI 与软件企业提供即开即用的企业级基础设施，免去底层繁琐开发。 - **Ridgeline**：面向资产管理机构的新一代全流程实时云端操作系统，助力投资机构实现数倍规模化扩张与精细化运营。 - **Rogo**：专为华尔街与顶级金融机构量身定制的 AI 智能体平台，深度理解投研流程并直接链接实际业务产出。 ## Notable quotes > `30:00` I often say we are a shark in digital waters, but in physical waters with branches, we are significantly less competitive because we don't have branches and we don't have ATMs. — David Vélez > `38:00` Your margin is my opportunity. — Jeff Bezos > `46:00` Code Red — David Vélez > `48:30` Dictator's Dilemma — Tobi Lütke > `25:20` The 190 currencies might go down to 10. — David Vélez > `32:35` Fintech is dead — Mickey Malka > `01:10:00` You cannot imagine your life continuing if you do not decide to tackle it. — David Vélez > `46:00` Holy shit — David Vélez ## Mentioned David Vélez, Patrick O'Shaughnessy, Nubank, Blockbuster, Netflix, Uber, Amazon, Jeff Bezos, Disney, Pix, UPI, Plaid, ChatGPT, Codex, Hyperplane, Claude 3.5 Sonnet, Claude Opus, Clojure, COBOL, Mickey Malka, Ribbon Capital, Tobi Lütke, Shopify, Ramp, Vanta, WorkOS, Ridgeline, Rogo ## Video description David Vélez is the founder and CEO of Nubank, which serves 140 million customers across Latin America.&nbsp; David believes the industry is still in its Blockbuster era, with most of the world's banking still running through branches and incumbent banks, and that AI will make digital banks dramatically more efficient than the incumbents they compete with.&nbsp; He explains how Nubank rebuilt its entire organization around AI after missing the moment the technology started working, and why he is now bringing Nubank to the US, a market he says is still dominated by big incumbent banks.&nbsp; Please enjoy my conversation with David Vélez.&nbsp; For the full show notes, transcript, and links to mentioned content, check out the episode page here. ----- Become a Colossus member to get our quarterly print magazine and private audio experience, including exclusive profiles and early access to select episodes. Subscribe at ⁠colossus.com/subscribe⁠. ----- ⁠Ramp’s⁠ mission is to help companies manage their spend in a way that reduces expenses and frees up time for teams to work on more valuable projects. Go to⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ ⁠ramp.com/invest⁠⁠ to sign up for free and get a $250 welcome bonus. ----- Trusted by thousands of businesses, ⁠Vanta⁠ continuously monitors your security posture and streamlines audits so you can win enterprise deals and build customer trust without the traditional overhead. Invest Like the Best listeners get a special offer of $1,000 off Vanta when you go to ⁠vanta.com/invest⁠.&nbsp; ----- WorkOS⁠ is the infrastructure B2B and AI-native companies use to sell to enterprise. It covers everything enterprise security requires: SSO, SCIM, RBAC, Audit Logs, AI governance, and more. Trusted by 2,000+ fast-growing companies, including OpenAI, Anthropic, Cursor, and Vercel. ----- Rogo is the AI platform for finance. They're building agents for Wall Street that are trained to understand how bankers and investors actually do work: from diligence and modeling, to turning analysis into deliverables. To learn more, visit rogo.ai/invest. ----- ⁠Ridgeline⁠ has built a complete, real-time, modern operating system for investment managers. It handles trading, portfolio management, compliance, customer reporting, and much more through an all-in-one real-time cloud platform. Visit⁠ ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ridgeline.ai⁠. ----- Editing and post-production work for this episode was provided by The Podcast Consultant. Timestamps: (00:00:00) Banking Is Still In Its Blockbuster Era (00:02:52) 140 Million Customers, 10,000 Employees (00:04:53) What Banks Could Do in Five Years (00:06:31) The $600 Billion Tax on Deposits (00:08:06) A Private Banker for Everyone (00:12:21) How a Startup Would Attack Nubank (00:14:56) When Nubank Took AI Seriously (00:17:08) Data as a Moat (00:21:29) How Global Finance Will Change (00:25:20) Stablecoins and the End of Cash (00:28:45) Creating Creditworthiness (00:32:35) Where Fintech Opportunities Remain (00:34:19) Nubank Comes to the US (00:41:11) Lessons From Disney and Amazon (00:43:44) How to Sell AI to Nubank (00:46:51) Agents as a Customer Segment (00:48:39) How AI Changed Nubank (01:01:33) What Motivates Him Now (01:06:21) Founder Versus CEO (01:14:01) What Latin America Needs (01:15:40) The Bear Case for Nubank (01:19:42) Finding the Right Mission [Source](https://traffic.megaphone.fm/CLS2134103794.mp3)

[30:00] David Vélez: I often say we are a shark in digital waters, but in physical waters with branches, we are significantly less competitive because we don't have branches and we don't have ATMs. [38:00] Jeff Bezos: Your margin is my opportunity. [46:00] David Vélez: Code Red [48:30] Tobi Lütke: Dictator's Dilemma [25:20] David Vélez: The 190 currencies might go down to 10. [32:35] Mickey Malka: Fintech is dead [70:00] David Vélez: You cannot imagine your life continuing if you do not decide to tackle it. [46:00] David Vélez: Holy shit

[12:21] David Vélez: you still cannot really get ChatGPT to get you a banking license. [30:00] David Vélez: We are a shark in digital waters, [50:00] David Vélez: Imagine that's the type of feedback you would be giving an Olympic athlete. That would be crazy. You would never be able to succeed. Olympic athletes are somebody that literally every day is coaching them and providing potential feedback. That is the opportunity that ultimately AI is going to create for every single person in an organization—you have Olympic athlete type of coaching. [55:00] David Vélez: There's nothing else that I can imagine myself doing that could have more impact than taking Nubank from where it is today to making it the next larger—and going from three countries to 100 countries, or from 100 million customers to a billion customers. [57:30] David Vélez: It's my job to push very hard and to dream very big and to shoot very high. And then the organization will react and will push back if needed. [60:00] David Vélez: Are you the right driver of the car as the car becomes more complex and goes faster and faster and faster and faster? [66:48] David Vélez: Nobody will ever care so much about this company, about this company being successful, than I will. And that will make you put so much energy and focus to make it successful to the point that that ultimately is what is going to make you be the right leader for the organization for a very long time, independently of your expertise or your knowledge. [70:00] David Vélez: you are not sleeping because you're thinking about it [70:00] David Vélez: you cannot imagine your life continuing if you do not decide to tackle it

[0:00] Ramp is the only platform built to make your finance team leaner, faster, and better, saving businesses 5% annually on average so you can stay focused on growth. Ramp just opened for business in the United Kingdom, so if you're running a business in the UK, you can now use Ramp's AI-powered finance platform to manage cards, expenses, bills, approvals, and accounting all in one place. I run my business on Ramp, and so should you. Learn more at ramp.com slash invest. Thank you. Open AI, Cursor, Anthropic, Perplexity, and Vercel all have something in common. They all use WorkOS. And here's why. To achieve enterprise adoption at scale, you have to deliver on core capabilities like SSO, Skim, RBAC, and Audit Logs. That's where WorkOS comes in. Instead of spending months building these mission-critical capabilities yourself, you can just use WorkOS APIs to gain all of them on day zero. run on WorkOS. WorkOS is the fastest way to become enterprise ready and stay focused on what matters most, your product. Visit WorkOS.com to get started. Rogo is the AI platform purpose-built for financial institutions, serving tens of thousands of bankers and investors at hundreds of leading firms worldwide. Rogo's AI agents autonomously execute large chunk of your firm's workflows. They can screen deals, draft sims, run buyer outreach, and do diligence on data rooms, all with full security and compliance handled. to make transactions stronger than the last. To learn more, visit rogo.com/invest. So the last time we did this, I think you had 35 million customers and I have 140 million customers. So well done. Busy five years. Yeah, busy five years. One would be tempted to think that with that kind of growth, we've saturated the potential of fintech as a category to create a new platform, digital, mobile first, this sort of thing. And that we're maybe in the later innings technology transformation. I know that you think it's the exact opposite, and I'd love you to detail exactly why and what you think we can expect in the next five years, not just for NewBank, but for digital financial services globally. If you look at global consumer financial services, it's about a $10 trillion industry and still something like 90% of that industry goes through banking branches and is owned by incumbent banks. So while there is a lot of neobanks and fintechs across the world in a number of different areas, we are all of those fintechs and all these neobanks together. We're less than 10 percent of that entire market. Still in financial services is still almost as if 90% of the world was going to blockbuster to get VHS and DVDs and Netflix like services still had something like 10% of that industry. So that transformation, that migration that you've seen in our industries, in media, with Netflix, in retail, with Amazon, with the Barnes & Noble, in transportation, with companies like Uber, that migration from incumbents to digitally native companies is still very, very early in financial services. I think Nubank is very unique. Specifically in Brazil, Latin America, because we are really the only new bank in the world where we fully surpass the biggest incumbent bank. In terms of scale, we now have in Brazil close to 120 million customers. So, up to 65% of the Brazilian adult population is a customer of us. So in terms of share scale, it's very large. But even more than that, we become the primary bank account of the entire country. So about 60% of our customer uses us as their primary bank account. We replace the incumbent bank. We're not a secondary wallet. We're not some money that you leave that. We are the primary institution where people are really banking day to day. And that is very unique to NewBank in Brazil. And increasingly, we're doing that also in Mexico and Colombia. We think that this is a little bit of seeing the future of what will happen around the world over the next 10 to 20 years, where. You would expect a lot of shift from market share from the still incumbent banks to the digitally native companies. And that is just migration of market share, but also you'll see a significant increase in market size as billions of people across the world entered the banking system. You still have. 3 billion people that are completely inside the banking system don't have access to credit. I'd love to break this into two components. You could imagine if you froze the product today, And just worked on getting it in more people's hands. That could be one dimension of expansion. But then, of course, I'm sure you're thinking carefully about all the wonderful things that AI will allow you to do and technology will allow you to do for customers that you don't yet do today. Can you talk through each of those levers to achieve this blockbuster to Netflix transition? The first phase of transformation that enabled a lot of immigration in our industries was [5:00] And this is the same thesis you see repeating through the smartphone is the moment that we get to serve consumers without all that physical infrastructure. We have such a gigantic cost structure advantage that we can pass that efficiency to the end customer via lower fees and better product. Just to give you data point, we have 140 million customers with 10,000 employees. Our competitors have 100 million customers with 100,000 employees and 6,000 branches. And you start thinking about why the cost of real estate of the branches is gigantic. So our cost to serve a customer is about 3% of what a traditional incumbent bank is. That means we have so much margin to give back to the customer via no fees, better experience, better interest rates, and overall a better product. That's pre-AI. We're still here in phase one of what a smartphone enabled. AI will be an additional multiplication of that efficiency to that because then you get to choose. whether you want to have much more efficiency and you could automatize a lot of areas of the bank. You could imagine banks that have 500, a billion customers with 5,000 employees. with insight, or you can just decide to increase your level of ambition. And for somebody like us, especially when we look at How early it is in the transformation for us is why don't we go to 100 countries and try to accelerate this transformation by Nubank being one of the leading digital financial services firm in the world. What are the craziest things that you can imagine the product or a bank like yours doing for the customer in three or five years because technology has made things possible that weren't before, just like mobile made a lot of new things possible? We've got a much bigger wave now. Like, if you really dreamed about what could be possible, what are some of those ideas? We like to categorize the problems that we see in financial services in three buckets. One is access. Second is cost. And the third is intelligence. On the access side, again, you still have 3 billion people in the world that don't have access to a credit product. It's crazy. It's almost half of the world does not have access to credit. Even countries like Mexico or Colombia, 13% credit card penetration. Why hasn't that happened yet? Because you needed the creation of a lot of data. You needed sophistication and credit underwriting. You needed the digitalization of the systems to happen. And this is happening. Cash goes away, disappears. enter a number of real-time payment systems, as you've seen in Brazil with PIX or India with UPI, So there is a lot of transformation and catalysts that are accelerating the the rails on which you increase access to consumer banking for the world. And this is going to be a massive opportunity. 10, 20 years from now, 100% of the population has access to banking services. 100% of the population has access to credit. 100% of small businesses have access to credit. This is an insane push to global GDP. I mean, I don't think it's really understood how big of a model this could be for every single economy as you get half of your entire population in some areas access to banking. So that's access side. Cost is the one that tends to be more hidden, especially in developed economies. U.S. is generally a well-served banking services, especially if you're a high-income consumer. But when you look at the cost of that banking system, it's pretty significant in ways that are not readily apparent. For example, there's about $15 trillion of retail deposits in the U.S. banking system. But banks pay, on average, about 40 basis points on yield a year. In a world where your interest rates are 4.5%, 5%, you look at the calculation of those 15 trillion, that's more or less $600 billion that are going to banks to pay for banking branches that in a digital banking environment would actually stay in consumers' pockets. So this is a massive wealth tax, the fact that banks don't pay any yield. If you're high income, you do. The lower you income, the less you get paid. That's a huge tax and that's a huge cost. And then you look at remittances, for example, in the U.S., $100 billion go out of the U.S., people pay 5% to 10% per transaction. That will go to zero. Remittances, moving money, you're moving bits. That will trend toward zero in a more competitive environment. monthly account. People are paying five to ten, fifteen dollars per month in the U.S. banking system for a bank account. And you look at Europe or developed economies, Japan, same thing. You see a huge amount of cost that consumers are paying that will eventually, with models such as ours that come to compete, will also compete away. And then finally, intelligence. I think this is also the one that is less understood and is beginning to be discussed with the advance of intelligence. If you are in the top 1% of income, you have access to the best financial advice. You probably will have a private banker. The private banker is understanding all the bank accounts that you have, is optimizing your deposits, is thinking about what is the right tax efficient structure. If you need a loan, they're going to go and compete a bunch of different banks with each other and get you the base yield. But if you're in the other 99%, especially the more you go in the bottom, [10:00] You have access to nothing. Historically, there's been a lot of conversation about education is the answer. You just got to educate people. People don't have financial education. That never worked. People don't sit down to do classes. You're not going to solve this problem of intelligence through education. You can solve it through AI. And that's where I think AI is really transformational in financial services. Because it's almost providing every single person around the world... super intelligence to be able to optimize their financial life on their behalf. And remove a lot of these profit pools that exist today in banking that, in an environment with intelligence will be competed away. - But flesh that last piece out a little bit more. WellFront used to have this idea of self-driving money, which comes to mind, like, okay, I've got some money. Either I don't have it in the banking system, which is the worst. So I come into the banking system, I'm earning 40 basis points instead of 4%. So I'm in a lousy starting position, but I certainly don't have investments or tax loss harvesting or all this other fancy stuff. What do you think that ends up being for the average person in a couple of years because of this intelligence bucket? The first obvious rearrangement is those 600 billion dollars. that banks are keeping in the balance sheet to pay for banking branches get rearranged get forwarded to all the banking consumers because suddenly they realize what am i doing with my money leaving it at zero especially when you think about compounding what that means for savings and compounding interest this is a lot of money being sent ultimately to pay banking branches you look at really any country around the world any structure you look at mexican banking balance sheets for example or brazilians where it's exactly the same thing the big incoming banks about 70 of their and 30% that pay some yield generally to the highest income population or the corporates. One of the things that you see in banking is in a perfectly competitive environment, the best product should win. If you're charging no fees, if you have the best experience, banking is a commodity product ultimately. So you should be able to get... 100% of the market share. That doesn't happen because there is a lot of inertia in the system. And with the combination, the multiplication of intelligence with open banking, for example, Plaid, A player can connect 120 million accounts and you put intelligence on top of that. You're effectively allowing each customer to know how to optimize their financial life and where are the fees coming. So I'll give you an example. We're doing this in Brazil today where, About 15 million of our customers every day. We're telling them, hey, your bank account, your credit card in that or bank is due tomorrow. You don't have any money there. and they're going to charge you a 15% interest rate per day. Let us give you some money so you can pay that credit card or refinance. You have this loan at 10% interest. Let us refinance that. at 5% interest. It's ultimately a drive towards optimizing and removing a lot of those profit pools on behalf of the consumer. In some products, like small businesses do want to sit down with a banker, with a human and have that conversation. If you have millions of dollars to invest, you still probably want to see a human on the other side of the table and have that conversation. For certain consumer segments and for certain products, if you want to get a mortgage, or certain sophisticated products, there is value in having a physical space. But that might be 5% of the cases today versus if you have a savings account, a credit card or a loan, you just don't need that anymore. So I think the rationale would be effectively, yeah, we're trying to serve these customers, but it's ignoring that for the majority of people, especially in mass market, increasingly you just don't need it and you don't go to that physical infrastructure anymore. One of the questions that I think lots of investors and founders are wrestling with is what will be the best competitive advantages or moats in this new era when the cost to build new stuff digitally is much, much lower? If you had to imagine it like you're some new founder, I strip everything away from you. And your only goal is to kill new bank is to like build a new company to attack new bank, let's say in Brazil or in Latin America. What do you think you would do? Like what would be the vector of attack on your own business? Which in a way was what we did to the incumbent banks 13 years ago, right? We were the disruptor going after incumbent banks. So I think the early stage startup today that is AI native, their biggest advantage will be the speed, the ability to move extremely quickly and launch a number of different products. Especially if you're born unregulated, then you're going to be able to move fast. By now, we are fully regulated. We have banking licenses in a number of different countries. And so we have to go at a certain pace. We have to be checking with regulators. Now, I think specifically in financial services, but if you're banking, that will only get you up to a certain point because eventually you're gonna need to be regulated as well. And so I think, In summary, there's been a lot of conversation about how AI disrupts software and that you can ask ChatGPT or Codex to rebuild an entire software strategy, but you still cannot really get ChatGPT to get you a banking license. And you still cannot really get them to have regulators give you licensing in six months instead of two or three years, specifically on credit or in certain products that we operate in. The amount of data that we have is pretty significant, right? So we have about 140 million customers. For each customer, we have something like... [15:00] into 30,000 data points across 13 different years. So this is a huge, massive amount of information. that allowed us to build our own foundation model. We build our own foundation model for financial services for credit. And this gives us a significant advantage in making credit decisions, for example, or collection decisions. The early stage startup entrepreneur, while they would have the speed, they would have a much harder time getting to the level of data. So data becomes a significant moat now for more the incumbents. Companies like us, we're sort of in the sweet spot of benefiting the most from AI because in one way we already have the scale and the data and the brand. and the trust and the licenses that will be hard for the early native startup to move. But also we still have the technology angle and technology infrastructure to allow us to benefit from AI and what AI is creating and move even faster than we were moving before. Vanta automates compliance so your team can spend less time on security reviews and more time getting customers. Vanta cuts audit prep by 82% and gives you instant up-to-date proof of trust without all the manual work, so you can close deals faster and with less friction. Customers report a 526% return on investment, and more than 16,000 companies use Vanta, including Ramp, Harvey, and Snowflake. Get started with $1,000 off at Vanta.com slash invest. ridgeline is the first end-to-end system of record with embedded ai for investment management firms running portfolio accounting reconciliation reporting trading and compliance on one unified platform firms are moving off legacy technology and on to ridgeline because of how far ahead ridgeline's ai features are compared to anything else in investment management software i've been hearing from a lot of investment managers about ai and they fall roughly into two camps with some system over a weekend. The reality is that running an investment firm will always require governance, controls, and a single source of truth for your data. And no amount of AI enthusiasm changes that requirement. Ridgeline is built on exactly that foundation, which is why I believe that the firms that come out ahead in the AI era will be the ones running on Ridgeline's unified platform. If you're serious about your firm's AI strategy, Ridgeline should be part of that conversation, and you can request a demo at ridgeline.ai. Can you build the pretty detailed timeline of when you first had the thought like, wow, AI is a thing that we need to take really seriously? And then as the leader of the business, how you have thought about rolling it out and being more aggressive with it over time? And I'm interested in the details. Like you said, you have your own foundation model. Like, how is that built? Is it open source that you've tuned based on your own data? Like, who do you work with? How did you think about it? I'm really interested in your personal timeline, like how far back it goes and what the major hash marks have been. We started using machine learning since we started the organization. Since 2013, and we looked at Brazilian interest rates for consumers paying 500 and 600 percent a year, we said there is an opportunity here to use better data and analytics and machine learning to price credit risk. And different than a lot of fintechs, we decided to take balance sheet, to take credit risk and to be capital intensive. by default. And so we had already an organization that was very optimized for making decisions based on models and data. When the chat GPT moment came about three, four years ago, we started paying significant attention around how technology was going to facilitate some of the decisions. And we acquired a company called Hyperplane in Palo Alto that was already working on doing foundation models for financial services. They were having such a hard time selling to the big incumbent banks. Sales cycles were two years. They were not really able to move while the technology itself was fairly forward looking. So we bought that company. We integrated the team. We think we were the first financial services firm in the world to build their own foundation model with its own proprietary data to do credit decision specifically. We even published all our own research projects. In this area, we developed an early advantage. We're now using our foundation model for fraud decisions, for Mexico, for Colombia. As we launched the U.S. and our international markets, we can day one already come in with our own foundation model and understand how to use that data for making basic decisions. This should accelerate the data. or ability to come in. The data thing is such a good example of the scale advantage. And you're an investor, obviously, before starting NewBank. And so I'm sure you've thought about this endlessly. It sounds like scale is maybe the most important thing that will make NewBank so competitively advantaged for the next 10 years or something. Is that right? Like, does that feel like the right one to pick? I think this case is important, but you need it also with the right culture and the right systems. So one of the things we heard a lot in the early days was, well, how are you going to compete with the banks? The banks have scale. They have a hundred year history with a hundred million customers. How can you replicate that data? And going back to the analogy of Blockbuster, my answer was always, well, Blockbuster had a huge amount of data. But imagine the amount of people coming in into the Blockbuster and seeing if they were looking at that VHS or the DVD or not or renting. Did the data help them build a better search algorithm? [20:00] No, because they were not capturing the data. They did not have the right systems to capture data. They didn't have the right culture, data culture, to be making decisions based on data or based on models. And somebody like Netflix was able to come in with no data and build a better algorithm to, and ultimately use that. Scale and data is important, but equally important is your systems and your culture. You ultimately be in a culture that is using a lot of this data to make the right decisions. And this, I think, what differentiates neobanks or more digitally native organizations like us versus the more traditional incumbent banks that are still operating with 15 different systems as a result of a huge amount of acquisitions that they did over the years. And they're still, if you look at a lot of the AI applications, how we use AI to transform COBOL? How do we get away from the mainframe? You're talking about 10, 15 year histories of going to the cloud. Thank you. of going and putting systems on the cloud. How do you run that tactically in the culture to make sure you don't fall in the same trap that like all the big players before you have fallen into, which is that you're suboptimally using your advantages and you don't have a good algorithm? Like, what is the tactics of you making sure the team doesn't mess that up and actually does the opposite? I think the biggest difference is that historically, traditional banks saw their basis of competition as a lot of banking branches. That's literally how consumers used to pick a bank is I'm going to pick a bank that has a banking branch very close. And so that's why there was a huge amount of effort and energy into real estate and physical infrastructure. And obviously a lot of focus on building brands and building trust with consumers. In financial service especially, you need to protect that trust with consumer. That's very critical. So when we began the company, we did not have to spend energy or focus on an offline world. It's more of an online world. And we saw technology and data as a way to build comparative advantage. And so since the beginning, we saw ourselves as a technology company, not a bank, technology company that happens to be in financial services. And therefore, the culture itself has a culture that uses data and analytics consistently for every single decision. And what our own foundation model has enabled us is. The cost of putting a feature into a model has gone down significantly. Pre-AI, you had to spend six months adding a new feature. If you wanted to test a new feature in a model, you had to be spending several weeks work testing this feature, putting it in the model, testing it around. With our own foundation model, you can test a number of different approaches much faster. And then the cost of creating a model, you have an email that has purple letters. or yellow letters, the cost of doing a specific experiment has gone down significantly. And so every decision in the organization can now be automatized and be done by model versus a human. That's what AI is also enabling. So it kind of let us do what we were already doing at a much more significant scale. But the culture itself already existed since we saw this as a way to differentiate ourselves since the very beginning. How do you think the global financial system will change the most or needs to change the most in the next five years, totally away from your business? Like, what are the exogenous changes that you're most carefully watching that you think matter? One of the most transformational technologies in the world in financial services is going to be the first one is the end of cash. And this is happening really everywhere. There is a lot of big cash-heavy economies like Mexico, for example, or Indonesia, or a lot of African countries. Brazil used to be like that, where that cash will go away and will digitalize. mostly through real-time payment systems. I think the implications of this... Some of them are obvious. You don't need banking branches or you need less. Some of them are less obvious, which is what happens to the economy, for example, the efficiency of banks when they don't need banking branches, the speed at which money moves in the economy, the amount of data that gets created. in an economy that then can be used, for example, to provide access to credit. So that's a big catalyst. The second one is open finance and open banking across the world. This is basically the concept that you as a consumer... own all the data that you have across all institutions, and you're able to grab that data and share it with anybody else. So one of the biggest sources of inertia in financial services, but also let's call it telecom, for example, or utilities, is that you have a history with a specific company and you cannot transfer that history with somebody else that comes new. And incumbents have benefited a lot from that inertia of data. The moment that you can connect else. you're reducing that inertia and you're enabling more competition for your data. And you can monetize that data as a consumer. You can grab your historical information with a bank and offer it to somebody else that is willing to pay for that. If you decide to do so, Right now, social media does that without you deciding to do that. increasing, you're going to be able to do that as a decision. And the reason why banking is important is because then it will decrease that inertia globally and will increase significant competition in banking. Banking today globally really is still very much [25:00] kind of legopoly structure. You have five, six banks in the US that own 60, 70% of all the assets, same in Europe, same in Asia, same in Latin America. So that will mean a less concentrated banking industry, much more fragmented, much more competition. Obviously, that translates into less fees and less cost. The trend that has a couple of decades is the role of crypto and stable coins around the world. What will eventually happen with increasingly people that are much more mobile in nature is that you operate in a combination of the fiat system in your country with a global offshore system. Your lives go in between without you even realizing this. For a lot of countries around the world, it's really important. You've seen with the usage of access to Bitcoin, for example, stable coins, increasingly, they want to bank in the offshore system. They don't trust their banking system. They don't trust their fiat currency. Today, there's about 190 currencies around the world. My view is that 10, 20 years from now, those 190 will be maybe 10 because the fiscal situation in so many countries around the world is so dire that a lot of that fiat money is not real money. This is basically crypto tokens. So increasingly billions of people are going to be banking on the offshore world US stable coin but also Bitcoin and other type of assets You see these two parallel systems interrelating with each other across the world If you had to think about the U.S. dollar and its importance, if it goes from 190 to 10 or something, do you think the U.S. dollar is the reserve currency 10 years from now? Is it one of a few? What do you think its role will be in the world? There are two sides of that argument. You can make the case U.S. stablecoins ultimately will be the big savior or one of the biggest saviors of the U.S. dollar because it democratizes access to the U.S. dollar. And while the fiscal situation in the U.S. is not great and it's deteriorating fast, it's still better than the fiscal situation for the other hundred and plus countries around the world. You look at the balance sheets of all of the central banks or countries in emerging markets in Latin America, it's way scarier and you don't have the sovereign capabilities. At least there is a lot of demand for the US stablecoin. There's very little for the Brazilian stablecoin, unfortunately. So the fact that stablecoin exists for US dollar will ultimately mean that the US dollar maintains its power as a reserve currency for a very long time, but it won't be unique. I think you'll have other five, six fiat currencies that people will be saving in and a few crypto tokens where you also be diversifying And I always been a Bitcoin fan and it plays a role in a diversified investment portfolio What the most surprising thing do you think about the move away from cash What would I not appreciate about the implications of that or how long it would take? What's surprising about going cashless? The main one is what this means for credit markets. For the majority of the world, in the majority of the world economy, still economies are very much offline. And just think about this situation where you have a taco stand in a corner in Mexico City and you're selling through cash. That transaction is not being recorded anywhere. That's a transaction that you're selling to somebody and it goes away. And that taco stand cannot get credit today. Now, that person has a history where every day they go and sell tacos and they get cash flow that's being recorded. But nobody can underwrite on top of that. The moment that that transaction goes into a digital ledger, that transaction happens digitally, then suddenly that taco seller can take that information to a bank and says, hey, look at my cash flow, give me access to credit. And if you give me access to credit, I can now have two taco stands or I can have five taco stands. Digitalization of the economy and what that enables for credit for a significant percentage of the world population is something that is not very well understood today and is a huge accelerant And again you look at economies like Brazil for example something like 70 of all the employment in the country is still in small businesses It's not a big corporate, it's small businesses. The same thing happens in India, for example, or Southeast Asia. A lot of these small businesses historically also dealt with cash. Now they're digitalizing. So this is trillions of dollars of credit being enabled for billions of people around the world. this could be a significant economic shift for economic mobility across the world. And it's a very exciting opportunity. Do you think that you can create creditworthiness versus just predict or measure it once you have all this data? Like, is that one of the goals to, like, help people become more creditworthy somehow? Yes, absolutely. I mean, the first thing you have to do is kind of measure or record a lot of those transactions into a ledger. And that's what the utilization of cash enables. And then with better algorithms, and AI plays a role here, and then also with better products, you can enable people to start building that credit history. So I'll give you an example. We were the first credit card for 30 million people in Latin America. When we used to ask banks why so many banks didn't give access to credit, what we heard is people don't want credit. People don't want a credit card. They are scared about paying taxes, and so they're never going to use the product. [30:00] But what really matters is what we realized is what they were really lacking was a product that really matched their needs. And so we started for a lot of people with a product that the first line was $20, a $20 credit line. When banks saw the product, they thought it was absurd. But in reality, people really appreciated that the $20 were better than zero. And those $20 became $100 and the $100 became $1,000 over a period of time. And so you enable them to create a credit history through the right product. And then you enable an AI copilot for that customer that tells them all the different moves and what they need to do in terms of spending and investing and savings to gain points and become more credit worth. And that's, again, the concept of you providing a coach to somebody to improve their ability to access credit and improve their financial picture. So absolutely, I think that there's going to be a big enabler. How siloed are the individual countries in terms of how useful what you learn in one is helpful in another? like when you went to Mexico from Brazil, how much of the learning from Brazil was portable versus something about Mexico that's just distinctive about that country? And then how do you think about all the countries you've learned in so far and exporting that and those learnings to other geographies It more exportable than we would have expected The part that was less exportable the biggest difference between at least a country like Brazil Brazil was actually much more banked than unbanked. And so when we went to Mexico, a lot of the conversation that we heard is like, people in Mexico like cash. They don't believe in digital payments or credit cards or debit cards. In the early years of Mexico, that was the biggest difference. When you go from a banked customer, a banked country to an unbanked country. It's harder to compete. I always say that we are a shark in digital waters when we are forced to compete in the offline world with banking branches. We're much less competitive. We don't have banking branches. We don't have ATMs. And so for a couple of years, it was harder for us to compete, even though we became the second largest bank in Mexico by now and the largest digital bank there. But as these countries digitalized also, including Mexico, then the world's looking closer to each other. And that's ultimately what you actually see across any other industry, right? You look at social media across the world is the same product. You look at Netflix, the same product. You look at Spotify, the same product. You look at Uber, the same product. Amazon, generally the same product. So I would say historically there were a lot of particularities and the cash part was a specific one the future looks much more homogeneous than the past even for financial services If I forced you to put your investor hat on thinking about fintech writ large and where there might be other opportunities beyond the ones that you're focused on addressing, if I were to adopt the phrase of our great mutual friend, Mickey Malka, who said a couple of years ago, fintech is dead. Maybe he's the best fintech investor ever. And now he's doing less of that and more things in AI and power and data and things like this. How would you frame the global opportunity to invest in fintech businesses of new companies, of startups versus companies like yours that had this scale, can go take advantage of a lot of this thing. Like if I think about that battle between if I want to invest in fintech, should I do it in the incumbents? You're a new incumbent, let's call you. And pure startups. Like, what do you think the opportunity set looks like? I actually asked this question, Ricky, last week because I saw him spend a lot of time in AI and investing in AI companies. I was like, what are you done with fintech really? And I liked his answer, which is no, we're not. We are just doubling down on the larger neobanks or neofintechs that are now running the sweet spot between maintaining the ability to transform themselves through AI but now have access to these modes of scale and data and brand and licenses And so he continues to be an investor of us He continues to be an investor of an umbrella of startups And so for the early entrant I think he going to be in areas that are potentially less obvious as an entry path It becomes a little bit harder and you probably are applications on top of AI models that you do So there'll be specific applications, I think, of an entry path. But I think building a consumer brand with hundreds of millions of customers at this stage, unless you come up with something completely fundamentally different, it's going to be harder. Can you talk about this big announcement that you just made and why you're so excited about it, kind of what it means for the business? I think most people still identify NewBank as a Brazil neobank or Latin American neobank, which is what we are. But when you go back in time and look at our seed deck in 2013 and the way we thought about the opportunity, there's nothing specific about Brazil or Latin America around the thesis. The thesis really is that global retail banking will be transformed by digitally native companies. And this is because digitally native companies can solve the problems that financial services have better. And these are the problems that cost the access and the intelligence. And so we're excited about this announcement as we launch the U.S. as a market. And we also announced launching to upwards of 35 different countries a stablecoin backed account. because this is the beginning of continuing to playing this thesis that is a global thesis in [35:00] past 13 years as proving this model in Latin America and the countries that we operate. And I think we did. We're now the largest private financial institution in the entire continent. We have the highest net promoter score. From a profitability perspective, we have the highest return of equity in a country like Brazil. So we kind of did all the checks. And now is a moment to continue executing this multi-decade vision of taking this model global. We're very excited. U.S. is obviously a very competitive market. It is the biggest market in the world. There is a lot of very strong neobanks and fintechs in the U.S. that we'll be coming to compete with. But in reality, it's a market that's still dominated by big incumbent banks. So while there is a lot of fintechs, you look at where is the market, the market is still dominated by a few incumbent banks and some community banks. Most of the banking in the U.S. is still happening through banking branches, as if most of their video and media were happening through blockbusters. They still need to charge a lot and they need to pay zero yield effectively to be able to fund this physical infrastructure. And there is a lot of comparative advantages that we bring to bear, like our ability to do credit for the mass market or ability to do payments globally. And there's something that we also announced with the feature. Ability to really make customer love fanatically, bring a consumer obsessed culture in an environment and industry where customers just don't really care about their banking relationships, especially for the U.S. mass market. What do you think is going to be the hardest part about competing in the U.S.? Building a very strong brand will be one of the hardest things. U.S. consumers mention brand as one of the key attributes to choose banks. And it's not cheap. And we want to do it in a way where we obviously have a lot of financial resources being a very profitable. But we want to do it in a way that is smarter and more capital efficient than simply come in and invest in marketing all over the place. So building an organic brand in an environment where there are so many strong banking brands. might be one of the most difficult challenges that we'll do. And that's critical to build trust with consumers because we want to become their primary bank account. That's what we aspire to do. If I had every buy side analyst that covers new bank here and you got to convince them of how they should think about your business through financial metrics, you mentioned return on equity earlier. Is that the metric that you would point them to as the barometer of how well you're doing? Are there other metrics that you would hope that investors would use to assess how good a job you're doing that are financial? How do you think about that? growth of number of customers and net promoter score and the cost to acquire that customer. So when you look at figuring out, have you hit a vein and are you growing in a way where you found an underserved market and you're able to offer a product that nobody else were able to offer. That's one of the key metrics, number user count, NPS. But then are you doing it at a sustainable pace? Because you can buy your way into growth by spending a lot of money on marketing. So what is the equation there? What is your... LTV to CAC, and ultimately, what is your return on equity? And the return on equity exemplifies the advantages that the business model could have. When you compare it with the big incumbent banks in the countries that we operate, We don't charge a lot of fees. We literally charge no fee. So we have less revenue. But then we end up with a return on equity that is 2 to 3x their return on equity. That basically means that you found a cost structure advantage. You found a way to offer a product at better quality at a much lower cost. And if you find that, in general, it's through technology, then you found yourself a moat. You found yourself a comparative advantage, especially in markets like banking that are in ways commoditized. So I think it would be also looking at that combination of return and equity. Talking about the U.S., one of the most interesting things, least appreciated aspect of it is that by now, Some of the U.S. incumbent banks have higher return on equity, are more profitable than emerging banking banks. There's some of the emerging market banks that we compete. That means your margin is my opportunity, right? The typical Amazon view or investor view of. that incredible ROI that even incumbent banks have, Even on top of the huge amount of cost that you're using to provide your product, that means there is a significant opportunity to undercut. Who are your biggest influences? Like, I know you've learned from some of the great entrepreneurs, some of the great investors out there. Like, when you think about your whole new bank chapter of your life, who are the people that jump out that have most actually influenced how you think about something? Who are they and why did they influence you the way they did? We began the company as a reaction of a consumer pain, of me realizing how bad consumer banking in Brazil was and how painful it was, and then realizing that this was a proxy for what the majority of the country was actually feeling. Thank you. So we'll always admire companies that are very consumer obsessed, that know that to win, they need to be able to offer the best product and best service in a way that creates almost fanaticism. And so from that perspective, we were very inspired by Amazon. Since the very beginning, Amazon being the more traditional consumer obsessed company and how they build this consumer obsession and how they thought. [40:00] about that. We studied this name very carefully because this is a company that wants to create magic experiences. Maybe it's not super obvious, but even in banking, a boring banking product. We want to create a magical experience for customers. What did you learn from Disney? That's an interesting one. I think it's the communication that you create with consumers and just simply the magic aspect of it. When you enter a Disney park, the setup of that entry path that you have, it's almost like you're entering into a kingdom. And the service that you get from everybody that works inside Disney and how they treat the young children, spectacles that they create and the fireworks that they create, they're really thinking about how do you create a product that is magical in nature. And that's something that we push our teams around is for something as boring as sending money to. How do you create a magic experience out of that? the welcome kit that we're sending to our customers here in the U.S., when they received that welcome kit of the first time, credit card, the first debit card, we're trying to create a magical experience in that moment. And we're very inspired by that. The customer obsession thing is something that sounds obvious now. Maybe Amazon really pioneered this way of thinking. Nothing's for free. What are the trade-offs associated with being so customer focused? It's become a popular thing to say. And the temptation is like, yeah, everyone's customer focused to varying degrees. But what does that actually mean? What is the hard part about being so customer focused? I think the hard part is, as with any cultural value, is to do what you say you do. Like you see it in any company that has very nice sounding values on the wall and you're Then the question is, are they actually executing that, especially when it's inconvenient? Most companies will be consumer obsessed. But then when they are suddenly seeing themselves making more money because they invented a new fee, then, oh, well, maybe I'm less consumer obsessed here and I want to make a little bit more money. And you're prioritizing the short term versus the long term. And actually, in the entire AI debate, you actually don't hear that much conversation about customers. Because there is so much energy and focus around how to use AI for efficiency and optimization. There is much less conversation about consumer obsession here. And this is where humans still are beating the AI. So the biggest complexity is actually prioritizing this value, even when it's inconvenient to do so, especially from a financial perspective. But also the AI bro, the tech bro is much more enamored today with the technology versus the consumer experience itself. I'm really curious how you think about your own product, like literally the thing I see on my phone versus your own agent versus new agents being introduced in the marketplace. Do you think there's a world where it actually matters that your marketing is to agents more than it is to humans? Like, how do you think about that world or whether to provide your own interactive agents versus a product that you're clicking around in on your phone? I think agents become a customer segment as important as consumers as a segment. And you have to provide a product for the agent and you have to market to the agent and you have to do SEO for the agents. It's the same way we have under 18 segment and we have a small business segment. This is an entire segment and we're treating it as a segment. We have a team dedicated to understanding that segment and understanding the needs of that consumer, which is the agent. So that will be equally important. But I don't think it's going to be necessarily 100% of that decision because I think still specifically in banking, and this might be different in other segments, for example, but in banking, you go back to trust and branding. And the human will want to know where your life savings are deposited. And there is the human aspect of that. And so you're still marketing to the human and you're still building the trust with the humans. You've got to do both. And also for some products, there'll be more delegation. And for some products, there'll be less delegation, specifically for areas like in Brazil, where we have 65 percent of the entire country as a customer. We have the opportunity to be that agent to the consumer and the customer is coming directly to our app and asking our agent. for support specifically with financial products. And we can do it in a way where it's actually an open agent. It even provides product from third-party banks and third-party financial institutions. So there'll be a mix of this happening at the same time. If I think back to our original conversation five years ago, and you just think about yourself, how you've changed the most in five years or what surprised you the most in the past five years, what comes to mind? One big change has been this view around five years ago, everybody was working remotely, including us. And over the five years, we've realized how important it is to at least spend some meaningful time in person, in an office. And that's, I think, a bit counterintuitive, even from what you were five years ago. also in a perhaps counterintuitive way, increases the argument to spend time in person more. Because increasingly, the work that AI can do is more or less the work that people at home individually could do. And so the individual at home will be competing with AIs. and we'll lose that battle. But the work around collaboration and coordination in presence physically, which is what humans need to do, is what needs to be happening inside the office. We went from remote to hybrid and we're spending significant, much more time in the office [45:00] That's been one change. I think the second part is how the organization changes through AI specifically. We start paying attention to the technology. But when we tested it, it just didn't really work. Just to give an example, we use a lot of Clojure as a programming language. And for 2025, we kept using a lot of the models for Clojure, for programming specifically, and it wouldn't work. It was really awful. As our most experienced engineer said, this is not going to work. So we were testing, And Opus 4.6 came in in December of 25. And that was sort of the moment that a lot of people suddenly realized, no, this actually works. We didn't catch that moment, that switch moment in December. We only cashed it 60 days after as a couple of my partners and I were in California and we started meeting with a bunch of. early stage entrepreneurs. And we said, well, something fundamentally happened. And we went starting to test a lot of the new models, the new GPT models from Clojure and said, holy shit, this is actually working out. It can do cloud-review very, very well. And we're 60 days where it made a huge amount of difference between us and best in class companies that were using AI. And so we went back to Brazil and we went kind of a cold red around the entire organization. And we took a step back and we're almost in the process of reinventing the entire organization. We mapped out. 120 different workflows. Across the entire organization, coding is one of them, but everything from how do you do product design, know your customer flows, internal audit, compliance, regulatory capital management. Every part of the organization, 120 workflows were reinvented natively using AI tools. We had to create a lot of eval infrastructure to be able to measure at every single step how these workflows get reinvented. And I would say we are more advanced in some, we're less advanced in other ones, but the entire organization is being reinvented. That's what's so exciting about this moment is we get to reinvent the way we do things with AI technology. Frankly, a lot of the times is that you see a huge amount of productivity gains. It's not even because of AI. It's simply there is always opportunity to be reinventing the way you do things actively. And if we were to do it now without AI, we would find a productivity gains. But AI just augments the possibility to do that. And that creates the opportunity, for example, for larger span of control. less manager, much more visibility. I personally went from having one direct report to 20 direct reports over that period because it just increases your ability to operate horizontally. It's less layers on the organization. We went from 10 layers to five. So the organization went very horizontal. This is a bolt of adrenaline to the organization as well in terms of our ability to execute. Can you say more about what made that possible? So what was happening at one direct report and 10 layers? And then now what's happening at 20 and five? I'm interested in literally like the day to day tactical thing. I'm sure most CEOs would rather have a flatter, higher bandwidth of information organization. So how does it work? Like what changed in that? I think this is partly AI related, partly is the way we manage our organization, I think for a period of time. As we start scaling significantly and we went from a million customers to 140, the organization just takes a little bit of inertia of itself and it starts adding layers. I think every fast scaling organization goes through the same process because you see things breaking all over the place, right? You have all these growth pains and the easy answer when things are on fire is to just add managers to solve the fire immediately. By sort of a natural organic process, we ended up very verticalized. And we ended up with a structure that didn't give a lot of visibility. Partly what AI creates is it allows you to take a step back and reinvent everything about your business. This was a catalyst, I think, for us to think about the way we were structured. Toby Luki has talked about this, how he can now ask the company stuff because of AI can tell him stuff. So the separation between him and ground truth based on data is gone. And that used to be this. I like this idea of the dictator's dilemma that as someone gets more powerful at the top of an organization, it's harder for them to know what is real and what is true. How do you stay up to date on what ground truth is or what is happening inside the business when it's gotten so big? Every part of that chain... there is a possibility for a message to get distorted and you get up the distorted vision by the time it navigates 10 different layers. Simply by removing a lot of those layers, you are much closer to the ground truth. And you reduce a lot of those layers by having information systems that have access to everything that's happening across the organization, the company brain. That is proactive and reactive. It's actually sending me information about what we're doing from changing our AI transformation to how we're doing around launching certain products or metrics or features. You get directly to the ground truth. There is more about a mentality of how do you actually also actively go directly to the people doing the work. It's very comfortable. The organization takes its own shape to respect the channels. You'll see this in bigger organizations. [50:00] a manager and ask the question to somebody that is actually doing the work. There is a specific chain of command that needs to be respected. This is more related, I think, of traditional organizations. And here is more your approach of being a CEO is like, no, I want to go talk to anybody that I have a question with. And there is no respecting of a chain of command. Literally, I'll go to whoever has access to the best information if I have a question or we're going to problem solve together if I see something that is not moving. So I think it's partly technology. Parley's culture and attitude of you're willing to be deep down integrated into the organization. How do you think all of this will affect your company or any company's ability to raise the overall average talent bar in the business? How can you use this to see how good each person is and help them get better or move people on that aren't as good? Have you thought about the performance culture management part of the data and the technology yet? I think it's going to be transformational. And I think this is an area that I'm surprised so few people have discussed about how AI is going to reinvent HR in the organizations. Because if you look at it the way it is today or for most organizations is your manager gives you feedback every six months, tells you what you're doing well, what you're not doing well. Generally, there's a lot of recency bias. So it's not really six months. It's like the last 30 days, what you did in the last 30 days, that's good or not. generally is not very actionable is, is like, You're doing this well, you're doing not, maybe you should do better. Imagine that's the type of feedback you would be giving an Olympic athlete. That would be crazy. You would never be able to succeed. Olympic athletes are somebody that literally every day, is coaching them and providing potential feedback. That is the opportunity that ultimately AI is going to create for every single person in an organization is you have Olympic athlete type of coaching. And so. Well, we're building a new bank. In our HR operations system is literally every day, depending on where you are, if you're an engineer or a designer or a product manager, you're going to be a business manager. Daily, you should be able to get feedback from the AI in your AI copilot if you're coding, you're not. He's reviewing your pull requests and he's telling you not only how you're using these tools, but how could you use them better? And it's very actionable and it's very specific to what you're doing. If you're a designer, is giving you feedback around how you're doing your job and how better you could do in your job. If you're a manager, managers right now have very little information, especially if you're managing a lot of people. That's the other thing about this large span of controls. there is a sense that you have to be having one-on-ones with every single person in your week. Like those one-on-ones tends to be, very low information density type of interchange. A manager that has access to all this information, this instrumentation is now seeing real time how every person in the team is actually doing and is allowing them to improve what they're doing. How do you stay up to speed on AI capabilities? Like who are the people you rely on? What are the things you like to read? I feel like I need to time box it every day. I get one hour to read the updates or something. It's like SportsCenter for business. How do you do it? I'll spend a lot of time on X. I think that's where you're saying a lot of the development also. I have an agent that collates and summarizes a lot of the developments of what's going on on details. I block specific time in my agenda. to use these tools and to build products. This is something that I hadn't done for a very long time. I coded a non-custodial crypto app, for example. And that's something that we've discussed at Nubank for a very long time to do. And I just went in and did a project. We asked, all of our senior managers to build products every quarter, our actual software and products, so that we are integrated and very aware of what's going on. I don't think necessarily the answer is that I should be spending 70% coding. There are better people that will be much more productive coding than I am. But I think it's important for everybody in the organization to be actually not only reading, but actually using the tools actively to understand how fast the technology is going. Your finance team isn't losing money on big mistakes. It's leaking through a thousand tiny decisions nobody's watching. Ramp puts guardrails on spending before it happens. Real-time limits, automatic rules, zero firefighting. Try it at ramp.com slash invest. As your business grows, Vanta scales with you, automating compliance and giving you a single source of truth for security and risk. Learn more at vanta.com slash invest. away the complexity across portfolio accounting, reconciliation, reporting, trading, compliance, and more, all at scale. Schedule a demo at ridgeline.ai. Every investment firm is unique, and generic AI chatbots don't understand your process. Rogo is built by finance professionals. It's connected to your data, learns your process, and quickly produces client-ready outputs. Check them out at rogo.com slash invest. The best AI and software companies from open AI to cursor use WorkOS to become enterprise ready overnight, not in months. Visit WorkOS.com to skip the unglamorous infrastructure work and focus on your product. If you think really carefully about it, [55:00] How has what motivates you to keep doing this and keep fighting and keep building changed over the course of the new bank journey? Like if you think back to the start, you and actually last time talked about this, like what the original motivations were. And I mean, not just the business opportunity, but the personal sources of drive. How has that evolved and what's it like today? I'm motivated by impact, by making it count, by doing something. We're investing time and energy in something that has a meaning and that has some kind of impact. I went through a phase, even with Nubank, where I started thinking, is there more stuff that I should be doing beyond Nubank? I think all founders get starting on the company. The company then has a significant growth. And then you start asking your questions. What else could I be doing? Ultimately, I went through the other side of that analysis for about a year or two years, realizing that there's nothing else that I can imagine myself doing that could have more impact than taking Nubank from where it is today to making it the next larger and going from three countries to 100 countries or from 100 million customers to a billion customers. What this means in terms of impacting financial services impact for people around the world for billions of people access to financial services And with that impact out there it impact internally too It like how I can also understand myself through bigger and newer challenges and how I can develop better as a leader to motivate an organization that has big ambitions and with big technology shifts like AI can either decide to stay behind or stay in front. I think this technology shift forces every organization to think in which side of this disruption do I want to be? I find that I'm more of a wartime general than a peacetime general. And so when a disruption, a change in the world happens, it suddenly is a source of motivation. I get more energized to actually continue to execute that mission. What's an example of understanding yourself better because of growth? Like in what ways has that happened? One of the questions that I personally had as I scaled the business is, will we need to bring much more professional managers at some point? Because I didn't manage anybody before starting NewBank. And then suddenly found myself being the CEO of one of the largest banks in Latin America and asking myself, well, what does the organization need right now? Do we need to go and hire a much more experienced manager? And we did. We went and hired a lot of external talent because I concluded or thought that I was going to need that But also then quickly realized that actually that not what the organization needed What the organization needed was more of what I had already brought into the organization which was this clear leadership the motivation the willingness to think really big and to dream very big and get people excited about doing something that feels impossible at the beginning. Once something clicks and says, no, this is the right answer for me and for the organization, then I can continue to do what I'm doing with much more confidence, especially when there are changes and shifts. There was always a little bit of a latent question, am I pushing too hard? Am I asking too much from the team? Is this the right pace? And the answer after 13 years of doing this and being a little bit more confident is yes, I'm pushing enough. It's my job to push very hard and to dream very big and to shoot very high. And then the organization will react and will push back if needed. But it's helpful to have somebody that is actively pushing the boundaries of what perhaps the company is ready to do. What is the hardest part of doing that for a long period of time? The imposter syndrome. It's fine to be a founder at the beginning, but then suddenly when you see yourself running a very large organization, it's like, am I the right person leading this organization? You have to through some thinking and through some experiences and through some feedback conclude answer that question for you at some point and answer it definitively on maybe for some reason you're not, but probably you are. That doesn't mean that necessarily the founder will have all the answers and will have all the skills. You will need to continue bringing people that add a lot of complementary skills to you. But I think at least for me was surpassing that phase of having a voice in the head thinking, are you the right CEO for the business? And you're making the shape between founder and CEO, because generally you will see founders having both titles, but they're very different functions. The founder and the CEO are very different functions. They don't necessarily fit with each other. What's the key energy of each? The founder is impatient, huge amount of sense of urgency. The founder will want to go run very, very fast, will move fast, extremely fast. The CEO will need to be more patient. The CEO will need to be more systematic in how it makes decisions. We'll have to bring generally much more systems thinking into scaling the organization. And sometimes those things collude. If you have to be impatient on one end, you have to be patient on the other end, there might be conflicts in that nature. Generally, you would think that CEOs are better managers. [60:00] Founders don't tend to be good managers. Founders don't have a lot of patience or willingness to do one on ones and sit down and provide coaching to a team. They just want to run. A lot of the time, they tend to be much more ICs, individual contributors and managers. CEOs tend to be better managers because you have to build systems. You have to be thinking more from a system perspective. I think the key, though, for me has been to realize that The CEO title itself and the founder title, you can redesign in different ways, too. And it's being able to do both. It's being able to have a little bit more patient when you need a little bit more patient. But generally, you want to maintain more the founder than the CEO. And I think that's why founder-led companies continue to outperform for decades, because if you have to choose, you choose more the founder than the CEO mode. especially in environments where there are big technology shifts, where a lot of the advantages, the assets of the founder come to bear and are more obvious than necessarily the advantages of the CEO. What was your lowest confidence moment personally in the entire history of the business? How'd you get through it? And I think you go through a lot of different phases. And I think scaling a business is hard. Especially because you're consistently making decisions about situations that you don't really know. Like, for example, I wanted to build a technology company, but I was not an engineer myself. So I needed to go and hire great CTOs and build a great technology organization. How did I know how to pick a great CTO at that point if I didn't have the experience or credit? Whether it's a lot of specific to financial services and credit expertise. So there was a lot of questioning inside internally, a lot of the decisions that I made about building the team. That made me ask myself the question, do we need to bring somebody with more experience from a big bank or from a big organization to run the organization? For a period of time, is this imposter syndrome that is asking you my question is like, Are you the right driver of the car as the car becomes more complex and goes faster and faster and faster and faster? What was the most emotional moment in the history of the business for you? Passing a number of different incumbent banks in Brazil was very emotional because for 10 years, all we heard is, again and again and again, especially from the local experts, was that it was impossible to compete with any bank, that we were going to get crushed, that we were going to die, that These were the most trusted brands in the ecosystem and that we had no shot at being successful. Becoming the largest private bank in the entire country was something incredible. That was emotional in terms of what we were able to achieve from a point of scale. Personally, I was born in Colombia and I had to leave Colombia when I was eight because of all the security situation in Colombia as we grew up and I had to move to Costa Rica. And Colombia was then our country number three, the moment that I was able to launch Nubank in Colombia, which we did in a big event with the president of the country at the same time. I was very emotional as well because it was a bit of coming back home and delivering a product and a service that I thought was going to be very great for the country. in an environment where I was being recepted by my people. And it was videocasted and there were 100,000 people connected to the launch of Nubank in Colombia. It was very emotional, too. It was very exciting. With all this experience now running the business, how would you define great leadership? And I'm especially interested in the subtle points of what defines, in your opinion, a great leader. I think first is vision. You have to be able to see something that few people see, a world that is not as obvious right now. company about how things are going to move. So you have to have some kind of vision about the future that is differentiated. But then you have to be able to sell that vision. And selling is about inspiring people. And a lot of the time, that vision will be very hard or almost feel impossible to And the natural reaction of most people when they are presented with an impossible vision is to push back and to declare it non-realistic, right? This phrase, it's not realistic that this is going to happen. What is not realistic, right? I mean, look at every single meaningful thing. Development in the history of the world, at some point it was defined by somebody to be non-realistic. The amount of confidence that a lot of the experts or people use when they say it's not realistic. You have to be able to sell that vision in a way that a lot of people get motivated by that. That drives following and motivation. But then you have to be able to be a good listener and understand that sometimes you're actually going to be wrong. That vision that you have might not be fully complete. There is a mode in which you go against the consensus and you completely declare consensus as wrong. and you are the complete revolutionary. Nothing about the consensus makes any sense. And you will [65:00] Invent absolutely everything. That is one mode. The other mode is you're consumed by consensus. You think about it. Progress has been unrealistic and things will be the way they are today. What's hard, I think, about a great leader is you navigate in the middle. Consensus exists or status quo exists because... You think about it from an evolutionary perspective. The vision of the world competed with other visions of the world and it won. And so the world we see today was superior to our possible wars that happened. And so there are elements of that status quo that exists because of a reason. And so you have to be able to understand what are those elements. And they will continue to be here. But also... You have to be able to fight that confidential wisdom to see where the elements will need to be changed. And I think leadership is about knowing what doesn't have to change and pushing people. to change the things that have to change. That's a combination of conviction, but also active listening and humility to know that you might not have all the answers and that you're going to need to change your mind actively when you do that. Combining all those different factors, I think is pretty critical for good leadership. If we think over the next five years for NewBank specifically, I'd love you to paint what you think is like the bear case for the business and the case you worry about and are actively making sure it doesn't happen. And then if I think about my rose colored glasses on like the best possible story for NewBank. over the next five years. I'd love to hear the two ends of the spectrum from your perspective. The bare case is we are in these two countries in Latin America. We've saturated them completely. growth declines. We haven't really been able to find any other growth alternatives from a customer perspective. I mean, we have 65% of Brazil. Very soon, we're going to run out of Brazilians. Ideally, we would see the natality rate in Brazilian current people having more children, but that's not happening. So we're going to run out of Brazilians. We did not succeed internationalizing. We ended up being restricted to these three countries and growth saturated. That's the bare case. Partly of that would also mean that we could not take advantage of a lot of the AI opportunities that enables also. And so that's one reason why we're not able to internationalize. The bull case is the opposite of that, is we... truly took advantage of this thesis that we have around how much market share is going to be transferred. from incumbent banks to detailing native companies over the next two to three decades. And how that is going to be accelerated by technology shifts like real-time payment systems, open finance and AI. We're able to ride those tailwinds from technology shifts. And we were able to execute through a multi-country platform. a global vision of Nubank taking us to 100 plus countries over the next few decades. And we were able to solve a lot of these big problems that we still think that will be solved. Three billion people don't have access to credit, cash is becoming digital, more people need access to banking. If we have this cost structure advantage versus our competitors versus the industry players that still own 90% of the $10 trillion profit pool, if we have a cost structure that is 50x theirs, and this is pre-AI, then that meant we were able to pass a lot of the efficiency back to the customer via higher yields, lower fees, better experience, fully digitalized, integrating advice in our products, and truly taking advantage of these internationalization opportunities. So that is the bull case. When I had you last time five years ago, I already asked you my traditional closing question. So I have a different one for you today. Again, across NewBank and the incredible experience that you've had building it so far. What is the most interesting lesson that you've learned about entrepreneurship? how to quickly ignore the imposter syndrome and how The fact that you are the person that had the vision and the energy to build something out of nothing, to give birth, that already is giving you a level of care for that company so high that it will make you become the right leader for a very long time. I don't think anybody in the world cares as much for this company as I do, right? To a level that is almost miserable. You're a father, right? You have children. Nobody will ever care so much about this company, about this company being successful. Then I will. And that will make you put so much energy and focus to make it successful to the point that that ultimately is what is going to make you be the right leader for the organization for a very long time. independently of your expertise, or your knowledge, or whether you have the right experience, just that insight, just that energy will make you the right person. It seems like you could decompose that into have more faith in yourself. and be really persnickety about the vision that you want to go work on. Because if it's not something that's like coming deep from your heart, from your gut, you won't get the benefit of this thing you just laid out. Do you think those are the right two parts of that really interesting insight? Yeah, especially the second part you mentioned around the mission. It's a luxury and a blessing to find a mission that you truly are excited about. because then that will really make you go the long way. Scaling a company won't be easy. It will be much [70:00] painful than anybody anticipates. And if you don't have the level of conviction for the mission, it will be just harder to go. How would you encourage other people to do a good job on that search? Like if someone's searching for a mission that's perfect for them, Any advice that you would give them or any sign that they've got the right one? There is a level of serendipity there and luck. There is a level of search, of listening carefully, of feeling that this is a project, this is a problem that you truly care about. This is something that if you really care about, you are not sleeping because you're thinking about it. You cannot imagine your life Continuing if you do not decide to tackle it. I think AI makes it harder, not easier, just because the bar to start a company is now so much easier. You were probably just going to go and pursue problems that are easier to solve ultimately. I think that's why some of the best founders that we're seeing right now are going for the impossible. Truly, right. There is this shift now into hardware and space and things that seem completely impossible because. You're looking actively for those opportunities that will be harder and unique to you that you're going to get to believe that only you can really solve. Incredible place to close. What an awesome closing thought. Interesting lesson. Thank you, as always, so much for your time. Thank you so much, Patrick. It was awesome. If you enjoyed this episode, visit Colossus.com. You'll find every episode of this podcast complete with hand edited transcripts. and companies that we admire most. Learn more at Colossus.com slash subscribe. VC. Outro Music. You know how small advantages compound over time? That's true in investing and just as true in how you run your company. Your spending system is your capital allocation strategy. Ramp makes it smarter by default. Better data, better decisions, better economics over time. See how at ramp.com slash invest. As your business grows, Vanta scales with you, automating compliance and giving you a single source of truth for security and risk. Learn more at vanta.com slash invest. The best AI and software companies from OpenAI to Cursor to Perplexity use WorkOS to become enterprise ready overnight. not in months. Visit WorkOS.com to skip the unglamorous infrastructure work and focus on your product. Ridgeline is redefining asset management technology as a true partner, not just a software vendor. They've helped firms 5x in scale, enabling faster growth, smarter operations, and a competitive edge. Visit RidgelineApps.com to see what they can unlock for your firm. Every investment firm is unique and generic AI doesn't understand your process. Rogo does. It's an AI platform built specifically for Wall Street, connected to your real outputs. Check them out at rogo.ai slash invest.