Sacerdote Sells Enterprise Software Short as Anthropic Drives AI Stack

Anthropic推动AI技术栈变革，Sacerdote做空企业软件 随着自主AI智能体取代按席位付费的许可模式并促使企业IT预算转向基础模型算力，传统的SaaS商业模式正面临结构性萎缩。 Alex Sacerdote深入剖析了Whale Rock Capital Management历时20年建立的投资框架，该框架旨在通过锁定早期S曲线、强大护城河及巨大的盈利潜力来把握技术平台变迁中的机遇。他解释了在企业级编程热潮将该领域转变为规模达5000亿美元的市场之前，Whale Rock为何优先布局芯片以及Anthropic等基础模型。本次访谈详细探讨了推动供应商价值重估的硬件复兴、AI对传统按席位付费SaaS构成的严重生存威胁，以及Whale Rock如何将深度的第一手草根调研（scuttlebutt）与现代AI工具相结合，在公开与私募市场中发掘Alpha收益。 Whale Rock 依托沉淀 20 年的投资框架，专注于挖掘技术 S 曲线、持久竞争优势以及被低估的长期盈利能力，并按照“电力→芯片→云基础设施→基础模型→应用层”的需求传导顺序布局整个 AI 生态栈。 (3:08) 基础模型市场已从最初的 60 多家初创公司整合为由 Anthropic、OpenAI 和 Google Gemini 主导的寡头格局；以 Claude Code 为代表的自主 AI 编程工具正成为关键解锁点，对应约 5000 亿美元的全球潜在市场空间。 (5:00) 目前全球企业级 AI 应用渗透率尚不足 1%（仅占全球知识工作者的约 10 个基点），需求呈现近乎直角上升的“反 L 型曲线”，主流模型正通过构建 SDK、编排层和软件外壳来形成长期粘性、防范商品化。 (10:00) AI 算力负载每年 10 倍的爆发式增长推动芯片、液冷及高层数 PCB 等硬件供应链全面“去商品化”；与此同时，传统 SaaS 的席位订阅模式面临严峻冲击与预算挤压，促使 Whale Rock 战略性做空传统软件并重仓算力基础设施。 (35:00)

Whale Rock Capital founder Alex Sacerdote argues that identifying technology S-curves and underappreciated earnings power allows investors to capture exponential compounding before market consensus adjusts. In the current cycle, Sacerdote sees enterprise value concentrating in dominant foundation models and specialized hardware, while legacy seat-based software faces structural disruption. Exploiting S-Curves and Underappreciated Earnings Power (19:08) Whale Rock's investment philosophy rests on three pillars: technology S-curves, durable competitive moats, and underappreciated earnings power. Sacerdote explains that emerging technologies endure long gestation periods marked by high prices, friction, and poor usability until catalyst events remove adoption barriers—such as the $200 price point and touchscreen for the iPhone, or the $40,000 price and 300-mile range for Tesla's Model 3. Once these barriers fall, adoption shifts from linear to exponential growth. Because traditional financial analysts frequently rely on linear, short-term forecasting, markets consistently undervalue companies positioned at the steep inflection point of an S-curve. Sacerdote notes this dynamic allowed Whale Rock to purchase Nvidia in 2023 at 4 times forward earnings, Tesla in 2019 at 5 times earnings, and Amazon's AWS with minimal implied market value. According to Sacerdote, software and internet S-curves typically compound uninterrupted until reaching 30% to 40% market penetration, at which point consensus estimates catch up and earnings surprises normalize. Foundation Model Oligopoly and Enterprise Coding (3:08) In Sacerdote's view, compute infrastructure cycles develop along a strict hierarchy starting with power, advancing to chips, cloud platforms, foundation models, and finally enterprise applications. While roughly 60 startups competed in foundation models in early 2023, high compute requirements rapidly consolidated the sector into a three-provider oligopoly: OpenAI leading consumer mindshare, Google Gemini leveraging multimodal search and document ingestion, and Anthropic focusing on enterprise workflows and software development. Sacerdote identifies autonomous AI coding as the pivotal unlock of this platform shift. With approximately 20 million professional software engineers globally spending an estimated $20,000 to $30,000 annually on compute tokens, Sacerdote models the coding addressable market at roughly $500 billion. Rather than facing commoditization, frontier model providers are entrenching moats through proprietary software harnesses, developer toolchains, and enterprise security frameworks, even as current enterprise agent adoption remains under 1% of global knowledge workers. Legacy SaaS Disruption and Hardware De-Commoditization (35:00) The rapid advancement of AI models presents an existential threat to legacy Software-as-a-Service business models, prompting Whale Rock to pivot from a 40%–50% software portfolio allocation to a net short position. Sacerdote contends that traditional seat-based licensing faces severe pressure as enterprise clients freeze hiring, replace routine workflows with autonomous agents, and reallocate IT budgets directly into foundation model APIs. To assess incumbent viability, he proposes an AI evaluation metric that sums the percentage of total sales generated from AI with the company's category market share. Conversely, hardware supply chains are undergoing rapid de-commoditization. While legacy server workloads grew at a predictable pace, compute demand compounding at 10x per year forces physical components to extreme engineering tolerances. This shift benefits suppliers of mission-critical hardware, including high-bandwidth memory, liquid cooling systems, power supply units, specialized optical cabling from Corning, and high-density 40-layer circuit board materials from manufacturers like Elite Material. Primary Scuttlebutt Diligence Across Technology Markets (13:23) Whale Rock operates a unified research framework spanning public equities and late-stage private companies, exemplified by its $100 million secondary purchase of Stripe at a $35 billion valuation in 2020 and its entry into Anthropic at an $18 billion valuation. Sacerdote maintains that while AI models accelerate initial data synthesis and background filing reviews, high-conviction investing still requires Philip Fisher's traditional 'scuttlebutt' method, built on 2,000 to 3,000 face-to-face management and customer interviews annually. To establish maximum position sizing, Sacerdote applies a 'tripod of conviction' requiring alignment across three independent inputs: the portfolio manager's thesis, thorough primary validation from an internal sector specialist, and independent thesis confirmation from a respected external peer. This rigorous process informed early, non-consensus investments in platforms such as Adyen and AppLovin before broad market recognition. While Sacerdote projects substantial runway for AI compute, he highlights key macro and operational risks, including public skepticism, regulatory restrictions on data center infrastructure, and potential hyperscaler capital expenditure pullbacks if near-term monetisation slows. Additionally, Whale Rock maintains minimal exposure to the enterprise application layer, estimating that durable software moats will require another one to two years to become visible.

Whale Rock Capital创始人Alex Sacerdote指出，科技投资的核心在于识别技术S曲线突破普及阻碍后的拐点，捕捉未被市场线性思维充分定价的非线性盈利能力。他认为，当前AI浪潮正驱动硬件供应链走向深度“去商品化”，基础模型收敛为少数具备软件生态壁垒的寡头，而传统席位制SaaS软件正面临商业模式的根本性冲击。 S曲线拐点与数字护城河 (19:08) Whale Rock二十年的投资框架由技术S曲线、持久护城河与被低估的盈利能力三大支柱构成。Sacerdote强调，核心技术往往经历漫长的潜伏期，直到核心阻碍消除（如iPhone通过200美元定价与触控屏消除体验门槛，特斯拉Model 3通过4万美元售价与300英里续航破局），采用率才会进入指数级陡峭阶段。由于市场普遍按季度线性推演，极易低估高经营杠杆企业的远期盈利爆发，Whale Rock曾借此在4倍预期市盈率买入NVIDIA、5倍买入特斯拉、4倍买入苹果，并在早期以极低隐含估值买入未单独披露业绩的AWS。 然而，缺乏护城河的S曲线押注往往导致资本损失，正如早期的RIM、Palm与诺基亚。数字护城河（网络效应、行业事实标准、开发者生态、核心IP与零获客成本品牌）通常比传统实体壁垒更持久。例如Oracle与彭博终端依靠行业标准掌控定价权，ASML与高通凭借不可替代的IP垄断核心环节，而AWS则在竞争对手反应过来前便确立了十倍于对手的研发与规模优势。 算力爆发驱动硬件去商品化 (44:22) 过去四十年，在摩尔定律驱动下服务器硬件高度商品化，工作负载年增25%至40%。然而AI引发的计算需求正以每年10倍的速度复合增长，将物理硬件逼至极限，带动整个硬件供应链走出商品化泥潭。Sacerdote指出，NVIDIA机架功耗激增50%至125%，单个AI机柜成本从传统服务器的5000美元跃升至20万至30万美元，使得液冷、高带宽内存（HBM）和精密供电成为刚需，推升了台达电子与Advanced Energy等电源供应商的平均售价。 这一物理瓶颈使具备专有工艺壁垒的厂商获得了强劲定价权。例如，由IBM超算基因演化而来的Celestica成为Google TPU服务器的独家代工方，并占据白盒以太网交换机50%至60%的市场份额；由于AI服务器PCB由传统的10层攀升至40层，覆铜板龙头台光电子（Elite Material）实现50%至60%的出货复合增长并扩张毛利率；微软单个数据中心所需的特殊光纤用量甚至可绕地球4.5圈，使康宁（Corning）在专用光纤领域巩固了主导地位。 模型层寡头格局与代码Agent (5:00) 基础模型层从2023年初约60家初创公司激烈竞争，迅速收敛为OpenAI、Google Gemini与Anthropic形成的三足鼎立格局。Sacerdote指出，开源模型由于缺乏海量算力，难以跨越基准测试从80%到85%的关键准确率门槛，而这恰好是企业实际生产力的分水岭。Anthropic专注于企业复杂推理与代码自动化，其日活跃用户约1400万至1500万，年化收入正向300亿美元迈进；Whale Rock借助Claude Code制作了90页的深度报告分析全球开发者情绪，并以180亿美元估值参与其融资。 Sacerdote将AI代码生成视为“真正的释放点”。随着编程工具从每月20美元的代码自动补全升级为自主Agent工作流，知名技术专家Andrej Karpathy的代码编写已从80%手工编写逆转为自然语言Prompt驱动。按全球2000万名专业软件工程师、每人年消耗2万至3万美元算力Token测算，仅代码自动化就对应着约5000亿美元的潜在市场空间（TAM）。 传统SaaS危机与新AI评估法则 (35:00) 面对AI范式转移，Whale Rock彻底逆转了此前软件占组合40%至50%的仓位，于2023年初重仓半导体硬件并对传统SaaS采取净做空策略。Sacerdote分析，传统软件以僵化数据库界面与席位计费为核心，正遭遇四重压力：企业IT预算直接流向模型API与算力采购；裁员与招聘冻结削减席位数量；软件商无法再维持每年5%至10%的惯性涨价；非技术专家利用前沿模型构建内部工作流，大幅拉低了定制软件的研发门槛。 针对传统软件巨头，Sacerdote提出评估AI软件的“新40法则”：即AI业务收入占总营收比例加上在对应AI细分领域的市占率（两项相加达到60即为优质标的）。例如Salesforce在400亿美元营收基数上仅产生5亿至7亿美元的AI ARR（占比仅1%至2%）。如果Agent直接在底层数据库上执行任务，传统软件极易被降级为“无头数据库”，除非它们能作为Agent操作与工作流的核心载体维持不可替代性。 草根调研体系与公私市场套利 (61:03) Whale Rock将自身定义为以10人团队为核心的“学习机器”，每年开展2000至3000场面对面调研，坚守菲利普·费雪（Philip Fisher）的“草根调研”（Scuttlebutt）方法。Sacerdote指出，AI工具可高效起草背景纪要，但无法取代对真实产业生态的洞察；例如团队早在AppLovin上市前便深度追踪其业务，在行业峰会开展高频访谈，在公开市场形成共识前挖掘出阿尔法。在确立高确信度时，他依赖“三脚架”法则：自身独立逻辑、行业分析师的严格论证，以及机构外顶尖同行的相互验证。 这种跨公私市场的统一研究体系使Whale Rock能利用市场低效进行套利。由于大学捐赠基金等机构通常低配大盘科技股，误以为大市值缺乏Alpha，Sacerdote认为在数字化飞轮下，头部科技平台的全球利润池往往呈指数级膨胀，反转数百位公募经理的认知需要漫长周期，反而提供了丰厚的超额收益。其2020年通过交叉验证Adyen与Stripe的200余家客户数据，在恐慌期以350亿美元估值受让1亿美元Stripe老股，正是该研究体系的直接成果。 Sacerdote同时列举了AI多头逻辑面临的三大核心风险：一是公众对AI好感度偏低（仅约20%受访者持乐观态度）可能招致监管限制与数据中心审查阻力；二是若前沿实验室撞上性能天花板，开源赶超可能引发模型恶性价格竞争；三是超大规模云厂商削减资本支出也将对算力链条造成阶段性冲击。此外，他指出企业AI应用层目前仍处于极早期（知识工人渗透率仅约0.1%），其可持续的竞争壁垒与投资机会预计仍需1至2年才能真正显现。

## Background 本期节目来自知名投资播客《Invest Like the Best》（第 477 期），由 Colossus 频道于 2026 年 6 月 9 日在 RSS 及网页端发布。节目简介重点介绍了受访嘉宾——掌管逾 170 亿美元资产、过去三年复合年化收益率约 44% 的顶级科技对冲基金 Whale Rock 创始人 Alex Sacerdote，深入探讨其历经二十年锤炼的科技周期投资框架，并以其最核心持仓 Anthropic 为切入点拆解整条 AI 产业链。 ## TL;DR Alex Sacerdote 完整拆解了 Whale Rock 的核心投资框架：捕捉 S 曲线黄金爆发拐点、锁定高壁垒竞争护城河，并挖掘被市场严重低估的长期非线性盈利潜力。他详细阐述了前沿基础大模型如何走向三足鼎立的寡头格局，代码生成与 Agent 革命如何打开 5000 亿美元级别的新市场，并反手推演出算力暴渴如何引爆光纤、高层 PCB 与电源等底层数据中心资产的“硬件去商品化”超级复兴。与此同时，他剖析了为何传统 SaaS 软件正面临预算挤占与席位制瓦解的四重窒息，分享了在 Stripe 与 Anthropic 上以深度研究打动创始人的非共识下注经历，并直面反驳了机构投资者低配超大市值科技股的盲区。 ## Atmosphere & style 整场对话散发着华尔街顶级科技操盘手高密度、极度清晰的分析气场，充满穿透财报与产业链微观细节的硬核洞见。Alex 与主持人 Patrick 的交锋干脆利落，在剖析技术扩散壁垒与商业模式颠覆时兼具工科层面的精微还原与宏观历史长周期的宏大视野。而当对话在尾声转入其已故父亲在高盛的传奇生涯及长达六载的创业护航时，原先理性冷峻的投资推演瞬间沉淀为深沉动容的温情传承，为整集充沛的商业智慧增添了极为动人的人性底色。 ## Bulletin - `02:29` Whale Rock 目前信心最高的标的是 Anthropic；基础大模型赛道已收敛为类似于云计算早期的三足鼎立寡头格局（Anthropic、OpenAI、Google），开源与尾部模型难以跨越前沿闭源壁垒。 - `05:00` 代码生成与自主 Agent 开启了 5000 亿美元的工程师生产力市场，Anthropic 在实现数倍级营收跃升的同时已达到逃逸速度，并构筑起极高的企业端生态与 IP 护城河。 - `37:31` 传统企业级 SaaS 正面临 IT 预算分流、失去提价权及席位制（Seat-based）瓦解的四重窒息，促使 Whale Rock 将此前占仓位 40%-50% 的软件持仓清仓并反手净做空。 - `44:22` AI 算力负载年增 10 倍引发严重短缺与“硬件去商品化”，HBM、高阶 PCB、定制光纤及液冷代工（如 Celestica、台光电子、康宁）正从大宗商品蜕变为高毛利核心壁垒资产。 - `01:03:40` 主流机构投资者对超大市值科技股存在系统性、结构性的严重低配，未能理解数字经济赢家通吃带来的非线性利润爆发与深度专家捕捉 Alpha 的巨大空间。 ## Key takeaways - `00:00` Whale Rock 的核心投资哲学建立在三大支柱之上：寻找技术创新 S 曲线的黄金爆发拐点、甄别牢不可破的竞争壁垒，以及挖掘被市场严重低估的潜在盈利能力。 - `02:29` 基础大模型层最终演变为类似公有云（AWS、Azure、GCP）的三足鼎立寡头垄断，Anthropic 专注企业级与代码推理，OpenAI 占领消费级大众心智，Google Gemini 具备全栈技术底蕴。 - `05:00` 代码生成是 AI 终极能力的商业化质变：全球 2000 万工程师的垂直场景蕴含 5000 亿美元 TAM，算力置换高昂人力薪酬带来了十倍级研发效率飞跃与极高 ROI。 - `10:00` 基础模型公司不仅输出 API，更在构建 SDK、编排层等软件外围“马具”（Harness）闭环，客户锁定效应堪比早期的 AWS。 - `10:00` 全球知识工作者中深度使用 AI 原生工作流的比例尚不足 10 个基点（0.1%），但算力已严重告急，未来 4 年随着企业端渗透率向 1%-15% 攀升，算力短缺将成为唯一确定的物理瓶颈。 - `13:23` 公募基因机构参与顶级未上市独角兽竞争，必须依靠超越寻常行研的“投名状”深度报告与第一手产业链交叉尽调来赢得管理层信任与超额配售额度。 - `20:00` S 曲线爆发的关键在于消除价格、易用性与供应链等“采用壁垒”（Barriers to adoption）；当渗透率从 0% 走向 30%-40% 时往往伴随巨大的业绩超预期，而超过 40% 后指数增长通常宣告结束。 - `25:08` 企业级采用呈现依赖后端集成的“洗碗机模式”，初期由于数据安全与合规顾虑推进缓慢；而消费端如同“收音机模式”可极速铺开，AI 的无摩擦体验使其爆发速度超越以往一切技术浪潮。 - `30:00` 数字经济的六大核心护城河包括网络效应、行业事实标准、光速规模效应、关键知识产权、生态底座与极低获客成本的品牌力，缺乏护城河的高景气风口玩家最终将归零。 - `37:31` 传统 SaaS 厂商未能借 AI 提升客单价，反遭 CIO 预算挤占、优先级边缘化、提价权丧失及按人头计费模式动摇的四重挤压，商业模式面临根本性重塑。 - `40:00` 投资团队提出 AI 时代的“新 40 法则”（AI 业务营收占比 + 该 AI 垂直赛道市场份额），以此筛选真正具备 AI 纯度与业务壁垒的软件受益者。 - `43:18` AI 代理进入企业工作流将引发两极分化：悲观情形下传统软件被“无界面化”（Headless）降级为底层单纯数据库；乐观情形下成为不可或缺的核心记录与执行底座。 - `44:22` 数据中心计算负载年增 10 倍推动硬件去商品化（De-commoditization），高带宽内存（HBM）、40 层高阶 PCB、铜箔基板及特种光纤等供应链环节正在经历前所未有的景气与结构性供给短缺。 - `46:38` 以 Celestica 为代表的高端制造服务商，凭借攻克液冷工程难关、深度绑定博通及贡献开源交换机系统 SONiC，已蜕变为不可替代的关键安全供应商与高壁垒核心资产。 - `53:56` 公开市场多数投资者之所以错失硬件超级周期，根源在于“恐高”心理、对周期性波动的动摇，以及单点分析师无法看透基础模型层对硬件架构提出的重构要求。 - `55:36` AI 终局叙事的核心风险包括监管逆风与大众恐惧、模型能力提升遭遇天花板导致降级为价格逐底竞争、超大规模云服务商削减资本开支，以及巨头自研芯片的替代威胁。 - `57:47` 应用层的爆发在历史上永远滞后于基础设施建设，当前 AI 应用层格局混沌、护城河模糊且易受上游大模型能力扩展的直接侵蚀，因此投资应保持审慎克制。 - `01:00:00` AI 工具可以胜任初稿撰写与信息搜集等速记工作，但无法看透未来；投研的核心 Alpha 依然根植于菲利普·费雪式的面对面草根调查法（Scuttlebutt）与人类深层信任网络。 - `01:02:53` 高确信度投资决策依赖“三脚架”（The Tripod）交叉检验：CIO 的 S 曲线大局观认同、垂直行业分析师的财务基本面论证，以及受人尊重的外部顶尖同行的独立看多判断。 ## Detailed walkthrough ### `00:00` 播客开篇与顶级对冲基金 Whale Rock 的投资哲学 播客开篇提及了赞助商工具对现代高效企业运作的支撑，随后主持人 Patrick O'Shaughnessy 正式引出了本期对话的核心嘉宾——Whale Rock Capital Management 创始人 Alex Sacerdote。Whale Rock 是一家专注于科技领域的顶尖投资机构，旗下管理着超过 170 亿美元的资产，横跨对冲基金、纯多头（Long-only）及混合投资策略。在过去三年中，该基金凭借惊人的业绩成为全球表现最卓越的对冲基金之一，年化复合回报率高达约 44%。 Alex 的投资体系在过去二十年间不断沉淀和迭代，其核心框架建立在三大关键支柱之上：寻找技术创新的“S 曲线”（S-curves）、识别具备持久竞争壁垒的护城河企业，以及挖掘被市场严重低估的潜在盈利能力（underappreciated earnings power）。本期对话从 Alex 当前信心最高（highest conviction）的重仓标的——Anthropic 切入，顺藤摸瓜剖析从底层芯片、中间层基础大模型到顶层应用软件的完整 AI 技术栈。 ```mermaid flowchart TD accTitle: AI Tech Stack Architecture accDescr: The four tiers of the artificial intelligence stack from bottom infrastructure to top application layer. subgraph Infrastructure["算力与基础设施层"] Power["电力与数据中心"] Chips["半导体芯片 (如英伟达)"] Clouds["云计算巨头 (AWS / GCP / Azure)"] end subgraph Models["基础模型层 (三足鼎立寡头格局)"] Anthropic["Anthropic (聚焦企业端与代码)"] OpenAI["OpenAI (占领消费端先机)"] Google["Google Gemini (全栈技术实力)"] end subgraph Applications["应用软件层"] Agents["自主智能体 (Claude Code 等)"] Enterprise["垂直行业深度落地"] end Power --> Chips --> Clouds --> Models --> Applications ``` ### `02:29` 押注算力底层与大模型格局的“三足鼎立”演变 当 OpenAI 于 2022 年 11 月推出 ChatGPT、行业变革的发令枪打响时，Alex 带领 Whale Rock 仅有 10 人的精英投研团队，迅速对整个技术生态展开了地毯式的深度研究。每一次底层计算范式的迁移，都不可避免地催生出一套崭新的技术栈，并在旧技术栈上重塑赢家与输家。正如黄仁勋频繁强调的，整个 AI 架构最底层是电力和芯片，往上是云计算平台，再往上是基础大模型（Foundation Models），最上层才是应用。在 2023 年初，Whale Rock 的研判策略是“率先卡位芯片与基础设施层”：因为底层硬件最先承接爆发的需求红利，且市场竞争格局最为清晰透明；无论未来上层谁能胜出，对海量算力的吞吐需求是确定无疑的。 然而在随后的两到三年里，基础模型层的竞争迷雾逐步散去。回溯到 2023 年 4 月，Whale Rock 曾在一场研讨会上分析模型层的四种终局可能：赢家通吃、开源廉价化导致的价格恶战（race to zero）、彻底沦为大宗商品化（commodity），抑或是形成三到四家主导的寡头格局。现实演化印证了寡头垄断的判断——当初角逐的 60 多家初创企业绝大多数以凋零告终；即便是资本雄厚的科技巨头也难掩颓势，亚马逊始终未能在前沿自研模型上占据核心生态位，Meta 的初代努力遇挫后不得不推倒重来。最终，赛道呈现出高度类似于昔日三大公有云（AWS、Azure、GCP）托起整个 SaaS 时代的“三足鼎立”格局： - **OpenAI**：率先俘获了广大的消费级市场和大众心智； - **Google Gemini**：坐拥深厚积累，全栈实力不容小觑（Google 也是 Whale Rock 的核心重仓股之一）； - **Anthropic**：作为纯粹聚焦企业端需求的黑马初创公司，异军突起。 面对中国及全球开源模型的追赶压力，Alex 指出前沿闭源模型的优势依然坚不可摧。在基准测试中，从 80 分提升到 85 分所跨越的不仅仅是 5 个百分点，而是实现商业级解锁质变的鸿沟。开源阵营缺乏顶级算力集群支撑，能够逼近前沿却始终无法完成反超；而伴随着模型扩展定律（Scaling Laws）的延续及强化学习反馈回路的迭代，头部领先阵营的技术飞轮依然拥有极为深厚的演进空间。 ### `05:00` 代码革命与 Agent 爆发：开启半万亿级全新市场 真正让 Whale Rock 确立对 Anthropic 极致信心的关键催化剂，是“代码生成”（Code）这一 AI 终极能力的商业化质变。Alex 坦言，在技术浪潮初期，尽管他们对大模型抱有长期信心并投入巨资，但对于实际能落地多少企业收入、能否真正替代人力劳动仍存怀疑态度，彼时企业客户对早期模型的幻觉与局限性反馈颇多。早期的辅助开发工具（例如每月 20 美元的微软 GitHub Copilot）功能偏向于基础的代码语法纠错或简单补全一段逻辑，远未触及核心生产力的变革。 转折点发生在 2025 年 Claude Code 等新一代工具的亮相，AI 真正迈入了自主执行任务的“Agent（智能体）”时代。工程师们的使用方式从受限测试走向全天候不间断调用： - 在 Anthropic 内部，已有工程师单日消耗的 token 成本达到 100 美元左右，换算成年化支出高达 20,000 至 30,000 美元； - 全球约有 2000 万名软件工程师，仅按这一体量计算，单纯“编写代码”这一垂直应用场景所孕育的市场空间（TAM）就高达 5000 亿美元； - 这种高额投入对企业而言极为合算，因为这相当于用极具性价比的算力成本置换了数倍于此的高昂人力薪酬，实现了十倍级的研发效率跃迁。 伴随这一爆发，顶级软件先驱的认知发生了剧烈反转。传奇极客 Andrej Karpathy 与 Linux 之父 Linus Torvalds 彻底改变了此前的保留态度；Karpathy 曾指出过去 AI 只能写 20% 的辅助代码而 80% 需人工手写，但在最新模型发布后这一比例完全倒置，他如今甚至“除了用英文撰写 Prompt 以外，不再亲自写一行代码”，更不用说大模型还将彻底解放数以亿计完全不懂编程的普通人群的开发潜力。 在商业模式与护城河层面，基础模型完全不同于传统公有云纯粹的服务器与存储大宗商品化（commodity）。每一家基础模型都因独特的训练配方与对齐方法形成了显著的专业壁垒：Anthropic 在私募股权金融分析与复杂代码逻辑推理上具备压倒性优势，而 Google 则在长文本、PDF 文档吞吐解析上独具特长。这种深厚的核心知识产权构成了天然的差异化竞争壁垒，即便市场上存在各类动态路由调度工具（Model Routers），以及无数试图攻破其代码领地的竞争对手，Anthropic 依旧牢牢焊死了自身在代码与企业级生态中的统治地位。 - **极速飙升的财务曲线**：Whale Rock 在 2025 年 8 月以约 180 亿美元估值重注 Anthropic 时，其年化营收正经历从 1 亿美元到 10 亿美元、并向着更高量级光速攀升的罕见增长曲线，展现出科技投资历史上前所未见的商业化爆发力。 ### `10:00` 基础模型的生态飞轮与 AI 的 S 曲线渗透率 基础模型公司的价值远不止于对外输出一个纯粹的 API 或裸模型。以 Anthropic 为例，他们正在围绕核心 API 打造一整套闭环生态与“垄断级”产品矩阵——包括软件开发工具包（SDK）、面向企业的 Claude 产品、编排层（Orchestration Layer）以及配套的全套工具链。Anthropic 内部将其称为一种“马具”（Harness），即环绕在 API 外围、能够最大程度榨取底层模型性能的软件系统。 这让 Alex Sacerdote 联想起 2013 年初识 AWS 时的情景：当时绝大多数人认为云计算不过是堆在远程仓库里的廉价大宗商品服务器，根本不值一提；而亚马逊早已看透这是一种全新的计算范式，抢先一步发明并堆叠了所有关键配套组件，从而一步步构筑起极高的客户锁定效应。 ```mermaid flowchart TD accTitle: AI渗透阶段与算力短缺反馈 accDescr: 展示从AI 1.0到深度Agent应用的发展路径，以及企业渗透率跳升对底层基础设施算力造成的巨大压力。 subgraph Adoption ["应用渗透阶段 (S 曲线演进)"] A["AI 1.0 阶段<br/>超强搜索引擎 / 浅层问答<br/>用户规模：数亿级大众用户"] --> B["AI 2.0 / 原生形态<br/>Claude 接管电脑 / 构建自动化技能<br/>当前渗透：知识工作者约 10 bps"] B --> C["企业级智能体爆发出海<br/>从极客扩展至早期大众<br/>预期 4 年内渗透率奔向 1% ~ 15%"] end subgraph Infra ["基础设施压力 (L 曲线直线上升)"] D["底层基础设施 S 曲线<br/>当前渗透率仅约 10%"] E["算力严重短缺<br/>Anthropic 仅获所需算力的一半<br/>未来 4 年算力瓶颈持续收紧"] end C --> D D --> E ``` 站在 S 曲线的角度审视全局，底层基础设施层的渗透率目前仅在 10% 左右，但这依然是捕获 AI 红利最顶级的路径之一。眼下虽然有两亿到八亿用户在接触 AI，但绝大多数人仍停留在“AI 1.0”时代——把它当成一种“加了类固醇的增强版搜索引擎”。真正的范式转移在于全新原语（Primitives）的诞生：例如让 Claude 直接接管并操作电脑桌面、打通底层系统、自主沉淀专业技能，进而让大企业能够开发出真正的自主 AI Agent。 但谈到真正掌握这种原生工作流的深度用户，Sundar Pichai 曾指出这还不足全球知识工作者的 10 个基点（0.1%）。以 Anthropic 为例，其日活跃用户（DAU）大约在 1,400 万至 1,500 万之间，其中只有极少部分人在以极致方式调动 AI 潜能。这是极其经典的 S 曲线起点：先是一小撮极客折腾探索，接着蔓延到早期采纳者，随后走向早期大众。这个数字在未来四年内将以惊人的斜率从 10 个基点跳升至 1%、3%、5% 直至 15%。 今年以来，企业端仿佛被“按下了开关”，所有高管瞬间意识到必须争分夺秒完成转型。当下的局势酷似 1998 年的“互联网 1.0”前夜：人人都知道自己必须拥有一个网站，但亲手搭建一个网站却极其艰难。正因如此，企业级 AI 应用的市场渗透率目前还不足 1%，其需求曲线几乎是一道垂直向上的“L 曲线”。这种前端爆发直接撞上了后端的物理天花板：全球目前仅有 10 个基点的人在深度使用 AI，基础设施的算力就已经彻底告急。Anthropic 目前拿到的算力甚至只有其刚需的一半。正如 Marc Andreessen 所言，未来四年内唯一百分之百确定的事情，就是算力永远不够用。 --- ### `13:23` 公募基因突围一级市场：重仓 Anthropic 的破局之战 作为一家长期深耕二级公募市场的千亿级长仓与对冲基金，想要在估值惊人的私有独角兽（如 Stripe、Databricks、OpenAI、Anthropic）中拿下理想规模的份额，面临着完全不同的游戏规则。二级市场只需在交易终端轻点买入键，而在顶级一级市场，投资必须是“双向奔赴”，初创公司必须心甘情愿为你分配额度。 Whale Rock 首次接触 Anthropic 时是在其 10 亿美元融资轮次。当时分析团队虽然认识其财务团队的成员，但对公司理解不够透彻，加之公司毛利率为负，代码生成市场的爆炸式增长也尚未完全显现，因此并未仓促出手。公募投资者的优势在于能够在漫长的时间跨度中跟踪一家企业，按自己的节奏伺机而动。 转折点发生在 Alex 与 Dario Amodei 的面对面深谈。Alex 敏锐地意识到这支管理团队极具非凡特质：全神贯注、执行力极强、团队离职率近乎为零，且模型输出的代码质量傲视群雄。当公司的商业化路径真正开始兑现——将收入从 1 亿做至 10 亿美元已是奇迹，而他们正在朝着 90 亿美元的惊人目标狂奔时——Whale Rock 发起了全力冲刺： - **极致的行研投名状**：为了赢得会面并打动核心高管，Whale Rock 的分析师利用 Claude Code 全网抓取并分析了整个代码市场的海量用户反馈，提炼出 Anthropic 产品真正的核心长板与尚待补齐的短板，并附上了对未来代码生成软件市场规模的完整预判，最终凝结成一份长达 90 页的专业级商业汇报 PPT。 - **高管信任与超额配售**：这份报告彻底征服了 Anthropic 团队，他们向 Whale Rock 敞开了投资大门。随后，Whale Rock 与其首席财务官保持着极度紧密的协作伙伴关系，最终拿下的投资份额远远超出了市场对其规模的预期，打出了一场“以小博大、超水平发挥”的漂亮全垒打。 --- ### `15:00` 独角兽新版图：从 Adyen 到 Stripe 的“可乐与百事”打法 如今的超级独角兽一级市场总体量已经超越了除少数巨头外的整个欧洲股票市场，甚至单挑德国或英国的股市也绰绰有余。在涉足一级市场之前，Whale Rock 每年就要与上市公司管理层进行 2,000 到 3,000 场面对面深度调研。随着未上市巨头对二级市场行业格局的颠覆性影响日益加深，他们将 10% 到 15% 的调研精力直接倾注在这些顶级私有企业身上。 Whale Rock 在一级市场的初试锋芒始于 2020 年对 Stripe 的押注，而这完全脱胎于他们在二级市场对欧洲支付巨头 Adyen 的深度重仓： 在 2017 至 2020 年间，Whale Rock 持有大量 Adyen 股票。Adyen 作为新一代云端支付代表，正从 WorldPay 等传统老牌机构手中抢夺全球高达 80 万亿美元支付盘子中的份额（现代云支付渗透率当时仅约 5%）。Alex 意识到，“除非你对 Stripe 了解得如同手心手背，否则你根本无法真正把脉 Adyen”。为此，团队与 Adyen 的 200 余家客户进行了地毯式访谈，每次询问 Adyen 时也必定深挖 Stripe，最终确信这两家公司就是数字支付领域的“可口可乐与百事可乐”，必须想方设法进入。 在 2019 年结识 Collison 兄弟后，转机在 2020 年 4 月疫情恐慌最严重时降临。一家持有大量 Stripe 老股的风投机构合伙人（Alex 的好友）主动打来电话寻求流动性。凭借多年积累，Whale Rock 在缺乏详尽官方财报的情况下迅速完成了底牌穿透： 1. **真实交易规模（TPV）重估**：当时 Stripe 官方披露的估值约为 350 亿美元，声称其总处理交易量（TPV）超过 5,000 亿美元。但 Whale Rock 对比 Adyen 25-30 个基点的抽成率（Take Rate）以及 Stripe 40-50 个基点的抽成特征，并结合其实际员工规模测算其盈利能力，推算出 Stripe 的对外口径极为克制保守——其实际 TPV 已经逼近 1 万亿美元大关，实际抽成率也显著更高。 2. **果断吃进巨额老股**：在确定各项指标远好于假设模型后，Whale Rock 直接从该卖家手中接盘并顺势追加，锁定了高达 1 亿美元的老股大宗交易（Block Trade）。 对于许多风险投资基金而言，他们受制于基金周期必须退出变现；而对冲与交叉型基金的独特优势恰恰在于“既能在一级市场买入陪伴，又能跨越上市周期在二级市场长期持有”——Whale Rock 随后在拉美数字银行巨头 Nubank 上同样成功复制了这一策略。 --- ### `19:08` 核心投资三部曲：S 曲线、竞争护城河与被低估的盈利能力 随着技术周期的快速轮转，Alex 首次完整阐释了 Whale Rock 过去二十年千锤百炼的核心投资分析框架。整个体系严密咬合为三个核心支柱： ```mermaid flowchart LR accTitle: Whale Rock 投资框架三部曲 accDescr: 展示S曲线的正确位置、强大的竞争护城河与被低估的盈利能力三者如何共同驱动科技股投资回报。 S["1. S 曲线拐点<br/>捕捉技术爆发期<br/>获取指数级单元量增长"] --> M["2. 竞争护城河<br/>技术巨头的商业模式韧性<br/>防止利润被竞争者侵蚀"] M --> E["3. 被低估的盈利能力<br/>非线性利润爆发<br/>估值重塑带来的超额回报"] ``` - **第一步：锁定 S 曲线的黄金爆发期**：必须在技术采纳生命周期从极客探索跨向早期大众的关键拐点介入，唯有处于这一特定区间的赛道，底层业务才能获得“指数级的单元量增长”（Exponential Unit Growth）。 - **第二步：构筑牢不可破的竞争壁垒（Moats）**：科技行业孕育着大量极其优异且类型多元的商业模式与护城河。唯有具备持久竞争优势的企业，才能避免激烈的同质化内卷吞噬行业规模扩张的果实。 - **第三步：捕获被市场显著低估的盈利潜力（Underappreciated Earnings Power）**：在 S 曲线高斜率爆发与强力商业护城河的共同催化下，企业的最终盈利表现绝对不会呈现平缓的线性增长，而是会爆发出惊人的指数级盈利杠杆。寻找那些真实变现潜能被华尔街和大众市场严重误判的公司，是获取超额回报的终极胜负手。 ### `20:00` 技术 S 曲线框架与长期盈利能力 Alex Sacerdote 阐述了他核心的投资框架：通过理解技术 S 曲线（S-curve），发掘企业被严重低估的长期盈利潜力（earnings power）。很多时候，一家优秀公司的每股收益能从 1 美元暴增到 10 美元，这种爆发发生的频率远超外界想象。正因华尔街和大多数投资者极度短视，只死盯下一个季度或来年的财报，无法理解指数级增长的威力，Whale Rock 常常能在这些全球最顶尖的企业估值极低时进场建仓： - 2023 年大举买入英伟达（NVIDIA）时，对应其未来的实际盈利，市盈率（PE）仅为 4 倍； - 2019 年抓住电动汽车 S 曲线买入特斯拉（Tesla）时，估值仅为 5 倍 PE； - 早年持有苹果（Apple）时，对应实际盈利仅付出了 4 倍 PE； - 在 AWS 刚崛起时买入亚马逊（Amazon），按其后续带来的巨大价值算，AWS 业务简直是“免费白送”。 虽然市场普遍认为两到四年后的发展无法预测，但只要深刻理解 S 曲线规律、看透护城河并掌握正确的财务建模方法，就能捕捉到这些超级红利。任何新技术在爆发前往往都有漫长的孕育期：智能手机在 iPhone 诞生前 10 年就存在了；互联网在 Netscape（网景浏览器）出现前已经发展了 20 年；AI 在各家巨头内部隐秘迭代多年，直到 ChatGPT 横空出世才真正引爆；特斯拉上市 15 年后，才在 2019 年迎来近乎垂直的爆发式普及。这些拐点的核心在于“破除采用壁垒（barriers to adoption）”： - 早期智能手机笨重、没有触控屏、缺乏无线数据网络，售价高达五六百美元；直到乔布斯将价格压到 200 美元，配合 AT&T 的 3G 网络、直观易用且“连老奶奶都能轻松上手”的触控屏及成熟生态，所有障碍一扫而空，瞬间形成需求龙卷风（tornado of demand）。 - 特斯拉同样如此：马斯克消除了里程焦虑（提升到 300 英里续航），将车型售价降至 4 万美元区间，并彻底打通量产供应链，才促成了从平缓到垂直飙升的转折。 ```mermaid flowchart TD accTitle: 技术S曲线发展与拐点演进机制 accDescr: 展示技术从低渗透率孕育期、打破壁垒引发垂直增长，到渗透率达标后增速放缓的全过程。 subgraph Phase1["漫长孕育期 (0% - 几%渗透率)"] A["高昂价格 / 体验简陋"] B["配套基建缺乏 / 供应链受阻"] C["潜伏多年 (如iPhone前10年/Netscape前20年)"] A & B --> C end subgraph Trigger["关键拐点：消除壁垒"] D["价格降至大众临界点"] E["易用性与网络/基建就位"] F["供应链实现规模化量产"] D & E & F --> G["需求龙卷风 (Tornado of Demand)"] end subgraph Phase2["垂直爆发与成熟期"] H["爆发期 (0% - 40%渗透率) 业绩暴超预期"] I["成熟期 (40%+渗透率) 卖方预期赶上 / 指数增长结束"] G --> H --> I end Phase1 --> Trigger ``` --- ### `22:20` 动态评估 S 曲线的高度、斜率与退出时机 理解 S 曲线不仅要看它何时启动，更要研判它“有多高（TAM 潜在市场空间有多大）”，从而确定承销期以及何时离场。S 曲线本身是动态变化的。以亚马逊 AWS 为例，它曾只是零售分析师眼中的一个隐蔽账目行，但 Whale Rock 意识到其目标市场是整个企业 IT 支出——此前该领域被路由器、内存、存储设备及 Dell、EMC 等巨头割裂切分，而 AWS 将这些全部囊括，直面 6,000 亿美元的系统市场。起初他们假设云基础设施会带来 50% 的通缩效应，因此推算渗透率仅有 1% 到 2%；然而后续事实证明，上云的总成本与自建几乎相当并没有出现大规模通缩，这意味着实际的 TAM 比最初预想的还要庞大得多。 S 曲线也存在超级 S 曲线（Mega S-curves）与子 S 曲线之分，从早期的 Internet 1.0、移动互联、云计算、电商，一直层叠演进到如今体量最大的 AI 时代。然而，S 曲线的发展并非完全顺遂，投资者必须紧密跟踪其实际走势。例如电动汽车普及原本被认为能达到 40% 到 50%，但在渗透率达到 10% 到 15% 时却撞上了现实之墙，并未如预期般贯穿到底。 关于何时卖出，通常当某项技术渗透率达到 30% 至 40% 时，其指数级增长阶段便宣告结束。此时卖方分析师的预期已完全追平实际业绩，不再有令人惊喜的大幅超预期（big beats）。 - 在苹果公司的投资上，Whale Rock 在 2012 年其美国智能手机渗透率达到 50% 左右时选择卖出。 - 尽管错过了苹果后续依靠 App Store 30% 抽成应用生态以及多品类配件持续平稳复合增长（约 20% 年化），但其爆发力最强、年化回报达 50% 至 70% 的黄金阶段，确实集中在渗透率从 0% 到 50% 的 S 曲线前半段。 --- ### `25:08` 捕捉拐点：非共识直觉、草根调研与滞后进场策略 面对技术曲线前期动辄长达数年甚至十年的漫长横盘，如何准确抓住买入时机？Alex 引用了英特尔传奇 CEO 安迪·格鲁夫（Andy Grove）的观点：“当战略转折点到来时，你不能依赖常规数据。”在这个阶段，依靠的是直觉与轶事证据（anecdotal evidence）。Alex 推崇《Tao Jones Averages》一书中关于“全脑投资”的理念——最顶尖的投资者必须擅长运用右脑的视觉化感知与点状信息连接能力。 - 投资移动端视频游戏 S 曲线时，初期受制于屏幕尺寸与手机算力，市面上全是轻度休闲游戏；直到 Alex 在中国亲眼看到一个 12 岁男孩拿着大屏手机流畅游玩重度大型游戏，瞬间顿悟重度手游时代已全面来临。 - 相比消费端直观可见的体验，企业级软件更为隐蔽，必须依赖一线现场调查。在拥有 3 万名美国 CIO 参会的 Gartner IT Symposium 上，Whale Rock 团队捕捉到多次爆发现场：Splunk 讲解数据库时展厅被围得水泄不通；30 年前 VMware 演示服务器虚拟化时全场爆满；而调研 AWS 时，其主宴会厅在早上 9 点、10 点、11 点连续几个时段均全场爆满，这种需求井喷在财务数据显现之前就已经肉眼可见。 Alex 强调，**哪怕晚了一两年入场也完全没有问题**。如果 S 曲线的顶端高达数千亿美元，其增长坡道将极为深远，错过最初的 100% 涨幅并不会影响大局。Alex 回忆早年在富达投资（Fidelity）时，传奇基金经理彼得·林奇（Peter Lynch）非常乐于指导年轻人，林奇曾亲自告诫他：“把历史走势图涂白抹掉，一切只看未来。” ```mermaid flowchart LR accTitle: 两种技术扩散速度对比模型 accDescr: 对比消费级技术收音机模式与需要后端集成的洗碗机模式的不同扩散周期。 subgraph RadioModel["收音机模式 (Radio) / 纯消费端"] direction TB R1["即买即用 / 零部署阻力"] R2["7 年达到近 100% 极速普及"] end subgraph DishwasherModel["洗碗机模式 (Dishwasher) / 企业级B2B"] direction TB D1["依赖后端水电/排污/复杂集成"] D2["企业安全顾虑 / 组织文化摩擦"] D3["跨越周期漫长 (如SaaS耗时20年)"] end ``` 针对扩散速度，Whale Rock 曾委托前克莱顿·克里斯坦森（Clayton Christensen）合作者 Horace Dediu 研究过去 100 年的重大技术采用规律，并将其制成 S 曲线图挂在办公室墙上。研究揭示出两种鲜明的采纳范式： - **“收音机（Radio）模式”**：消费级产品落地极快，当年收音机仅耗时约 7 年便实现了近 100% 的极高普及率。 - **“洗碗机（Dishwasher）模式”**：必须与下水、管道等后端系统完成对接改建，普及过程极其缓慢。这与企业级 B2B 软件高度相似，因其必须深度接入企业旧有繁琐系统。 Alex 回忆自己在富达作为互联网分析师覆盖的第一只股票就是亚马逊，当时市场狂热鼓吹 B2B 互联网前景，但因底层基建缺位，这一浪潮直到 20 年后的 SaaS 时代才真正成熟落地。这同样是当前企业级 AI 面临的现实阻碍：大企业极度注重数据安全、行事审慎，组织内部往往需要少数热心布道者和最高管理层力推，但 IT 部门仍会因合规风险百般迟疑——正如早期企业普遍担忧云端不安全，直到美国中情局（CIA）带头采用、第一资本（Capital One）CIO 公开表示“云端比本地自建更安全”后，云计算才得以真正放量。 SaaS 和云计算因系统接入的摩擦，往往呈现出 30% 到 50% 相对平稳的年化复合扩张；而 AI 最令人震撼之处在于，对于消费者乃至许多企业用户而言，只要打开浏览器页面就能即刻调用其全部威力，免去了漫长的集成阻碍，从而迸发出前所未有的爆发式采纳速度。 ### `30:00` S 曲线与数字护城河：如何在竞争群体中识别真正的赢家 在技术普及的初期，真正的采用率往往只从 10 个基点（0.1%）刚刚起步向 2% 到 5% 渗透。Alex Sacerdote 将这种极度陡峭的需求拉升形态形容为“反向 L 型曲线”（Backwards L curve），其增长跑道极其漫长。然而，仅仅押注高景气度的 S 曲线需求是不够的，每个风口初期都会有数十家玩家激烈厮杀。Alex 的核心投资方法并非在早期混战中盲目下注，而是在行业出现明确的分化信号、领跑者脱颖而出之后，再进行彻底而详尽的产业链尽调，重仓拥有极强竞争壁垒的赢家。 许多传统价值投资者（包括沃伦·巴菲特）过去对科技行业避之不及，主要原因在于科技迭代速度过快、未来难以预测，且容易遭受颠覆式打击。但 Alex 指出，S 曲线正是预测科技演进的精确路线图。历经二十余年的投资实战，Whale Rock 发现数字经济中的竞争优势不仅不逊于传统线下行业，甚至由于数字属性而变得更加坚不可摧。 ```mermaid flowchart TD accTitle: 数字世界的竞争壁垒框架 accDescr: 展示 Alex Sacerdote 总结的六大科技核心竞争壁垒及其代表公司 subgraph Moats["核心数字护城河"] N["双边/多边网络效应<br/>(LinkedIn, Meta, 阿里巴巴)"] S["行业事实标准与锁定<br/>(Oracle, Bloomberg)"] Sc["极速扩张的规模壁垒<br/>(Amazon 5年走完沃尔玛40年)"] IP["不可绕过的核心专利<br/>(Qualcomm, ASML)"] Pl["生态底座与平台效应<br/>(AWS, iOS)"] B["极低获客成本的品牌力<br/>(Google, Tesla)"] end Winner["领跑者复合优势<br/>赢家通吃 / 估值复利"] N --> Winner S --> Winner Sc --> Winner IP --> Winner Pl --> Winner B --> Winner ``` - **网络效应（Network Effects）**：如 LinkedIn、Meta、阿里巴巴，用户越多，产品价值呈指数级上升。 - **行业事实标准（Industry Standard）**：像 Oracle 和 Bloomberg。即便市场上存在免费开源的数据库，Oracle 依然能维持极高定价，因为全球所有数据库管理员的技能、企业软件的适配全围绕其构建，牢牢锁定了关系型数据库市场。 - **光速规模效应（Scale at Speed）**：借助互联网与云计算，Anthropic 的年化收入能在极短时间内冲向数百亿美元；亚马逊仅用了 5 年时间就建立起了沃尔玛花费 40 年才沉淀出的规模壁垒。 - **关键知识产权（Critical IP）**：如同高通（Qualcomm）在基带通信领域让手机厂商不得不缴纳专利费，或 ASML 在先进光刻机领域独步天下，没有它们的技术授权或设备就无法生产硬件。 - **内生品牌与零买量（Brand Power）**：谷歌、亚马逊或埃隆·马斯克的企业几乎无需依靠传统付费广告拉新，获客成本（CAC）极低，使得客户生命周期价值（LTV）模型具有压倒性优势。 Alex 回忆起 Whale Rock 在 2013 年 Robin Hood 投资者大会上的经典战役：当时他们向市场力推亚马逊 AWS。在绝大多数多头尚未看懂云计算潜力时，Alex 就断言“这场战争还没正式打响就已结束”。当时 AWS 拥有整整 7 年的先发优势，研发投入是竞争对手的 10 倍，形成了“只有可口可乐、没有百事可乐”的独占格局。相反，如果身处历史级的顶级 S 曲线中却没有护城河，下场将极其悲惨——正如黑莓（RIM）、Palm、诺基亚、HTC、LG 和摩托罗拉，最终从巅峰迅速跌落归零。 --- ### `34:04` 基础大模型层的大逃杀：Anthropic 与 OpenAI 的逃逸速度 大模型层面的残酷洗牌完全印证了这种赢家通吃的法则。最初有大约 50 家公司雄心勃勃地试图构建基础大模型，如今绝大多数已被淘汰出局，行业彻底收敛至两到三家领头羊。尽管谷歌拥有顶尖人才，但其 Gemini 业务受制于庞大而复杂的既有基本盘；作为纯粹的大模型标的（Pure Plays），Anthropic 和 OpenAI 正在建立起外人难以逾越的复合壁垒。 面对微软与谷歌这两头拥有无限现金牛的万亿巨头，OpenAI 和 Anthropic 展现出了惊人的资本运作能力，在极端资本密集的赛道中成功融得数百亿美元。特别对于 Anthropic 而言，凭借连续实现十倍级（10x）的营收暴增和持续的融资造血，它已经迎来了自身的“逃逸速度”（Escape Velocity），迈过了最危险的生存关口。 - **代码领域的绝对核心 IP**：Anthropic 牢固统治着代码生成与推理市场，并构筑了强大的企业级品牌。Alex 直言，去硅谷或任何大型企业问 CIO，对方第一反应脱口而出的必然是 Claude。 - **飞轮级的自我迭代（Recursive Improvement）**：Anthropic 将其强大的代码模型反哺于自身的模型研发与基础设施迭代，形成“用 AI 写 AI”的递归飞轮，导致其技术创新步伐不仅没有放缓，反而呈现出指数级加速。 - **企业级 vs 消费级商业化**：企业客户付费意愿极强，因为企业采购大模型能够直接替代高昂的人力成本、交付即时 ROI；消费级虽拥有海量用户眼球，但目前盈利模式仍局限于广告或特定个人助手订阅。 - **算力资本门槛**：互联网世界的铁律是“领跑者越大、跑得越快”。下一代前沿大模型不仅需要百亿美元级别的算力集群，更需要持续支付天量推理成本，全行业只有极少数跨过规模门槛的领头羊能够留在牌桌上。 --- ### `37:31` 企业级软件的四重窒息：从 50% 仓位到全面反手做空 AI 对传统企业级软件（Enterprise Software / SaaS）的颠覆，是 Alex 投资生涯中观点反转最剧烈、执行最坚决的领域之一。五年前，SaaS 曾占据 Whale Rock 整体投资组合的 40% 到 50%。在 2023 年 4 月的机构交流会上，团队最初曾逻辑推演：老牌 SaaS 巨头拥有现成的分发渠道、庞大的直销铁军（Direct Sales Forces）和海量企业私域数据，理应能通过接入 AI API 顺理成章地大赚一笔。 然而，现实的演进迅速打破了这一幻想。传统 SaaS 厂商赶工推出的 AI 插件与功能体验平庸，既不能打动客户，也根本无法向企业额外收费提价。Whale Rock 迅速调整航向，不仅清仓了几乎全部软件持仓，甚至在步入今年时转为“净做空”（Net Short）传统软件板块，这一激进调仓在第一季度为其贡献了丰厚的回报。 ```mermaid flowchart TD accTitle: 传统企业软件面临的四重挤压机制 accDescr: 展示 AI 崛起如何从预算、ROI、提价权和用工规模四个维度挤压传统 SaaS 厂商 AI["生成式 AI / Anthropic Tokens<br/>极致的高 ROI 与代际差距"] --> B["CIO 预算挤占<br/>优先保障 AI 算力与 Token 支出"] AI --> P["软件优先级暴跌<br/>SaaS 采购在排期表中退位"] B --> Squeeze["传统 SaaS 厂商利润表承压"] P --> Squeeze Squeeze --> NoHike["每年常规提价能力丧失"] Squeeze --> SeatLoss["企业岗位自动化导致<br/>按人头收费（Seat-based）模式瓦解"] ``` Alex 形象地比喻道：传统软件的交互与构建模式，在 AI 时代就像是用“笔与纸”或是开着“马车”；而原生 AI 工具则是“喷气发动机”，甚至是科幻电影《星际迷航》里的“瞬移传送机”（Transporter）。即便最终软件公司不被 AI 彻底覆灭，它们眼下也面临着残酷的四重打击： 1. **CIO 优先级断崖式下跌**：在所有企业技术负责人的待办列表中，传统软件升级已被彻底边缘化。 2. **IT 预算的硬性分流**：企业预算是存量博弈，资金被大笔抽调去购买 Anthropic 等底层模型的 Token，因为大模型能立竿见影地替代人工、带来极高回报率。 3. **长期提价权的丧失**：过去 SaaS 软件每年定期涨价 5%-10% 的商业模式走到尽头，客户对平庸升级不再买单，软件厂商提价底气尽失。 4. **席位制商业模式的隐形杀手**：随着 AI 代替初级白领岗位，按员工人头（Seat-based）收费的软件商业模式正面临底层逻辑的动摇。 ### `40:00` 席位制受挫、行动力迟缓与AI时代的“新40法则” 在探讨传统企业级软件（Enterprise Software）所受冲击时，市场正陷入聪明人之间的多空激烈博弈。悲观的现实首先体现在人员规模上：大量企业正在冻结招聘甚至裁撤岗位，这直接打击了依靠员工人数计费的“席位制”（Seats）商业模式。如果从乐观角度看，老牌SaaS公司反应迟钝或许只是因为AI底层技术原语（Primitives）发展尚早，商业化落地需要时间；但深层短板在于人才与销售模式的错位——过去卖的是标准化固定软件，而如今如果要交付能够替代人类工作的系统，必须派遣“前线部署工程师”（FDE, Forward Deployed Engineers）驻场贴身实施，现有软件巨头缺乏这种人才结构。 另一方面，虽然多头辩称大企业绝不会冒着风险自己重写一套复杂的ERP系统，且历史上老技术具备顽强的“黏性”（正如移动端游戏并未杀死主机游戏，平板和手机也未终结PC），但AI原生代代码工具正在以指数级进化，三到五年内完全可能催生出正面击溃老牌玩家的全新AI原生企业。一旦传统巨头的核心数据优势被技术抹平，替换成本将急剧下降。老牌巨头想要“换挡”更是步履维艰——以Salesforce为例，在高达400亿美元的庞大总营收盘子中，真正的AI ARR（年度经常性收入）可能仅有5亿到7亿美元，占比微乎其微。 为了在AI转型期甄别真正的受益者，投资团队对软件行业经典的“40法则”（Rule of 40，即营收增速加上营业利润率）进行了重构，提出衡量AI渗透率的“新40法则”： - **传统40法则（SaaS）**：营收增速（例如20%）+ 营业利润率（例如20%）= 40%，代表企业兼具成长性与盈利质量。 - **AI时代的“新40法则”**：AI业务在总营收中的占比（如30%）+ 在该AI垂直赛道的市场份额（如30%）= 60%。如果能达到甚至突破这一数值，意味着该标的既享有极高业务敞口，又坐拥强大的壁垒；而当前多数传统软件公司的AI营收占比仅为可怜的1%到2%。 ```mermaid flowchart TD accTitle: AI时代的软件商业模式分化 accDescr: 对比传统SaaS老牌巨头面临的困境与AI代理时代系统地位的变化路径 subgraph IncumbentThreat ["老牌SaaS面临的阻力"] A["企业冻结招聘与裁员"] --> B["席位制 (Seats) 收入承压"] C["AI编程工具成熟"] --> D["自研系统与AI原生竞争者崛起"] E["高基数拖累"] --> F["AI ARR仅占总营收1%-2%"] end subgraph AgentEvolution ["Agent时代的定位演变"] G["AI Agents 工具调用"] G --> H["悲观路径：无界面化 (Headless)"] H --> I["沦为纯底层数据库，丧失交互价值与溢价"] G --> J["乐观路径：深度集成与执行底座"] J --> K["工作流程完全固化在既有记录系统内部"] end ``` --- ### `43:18` 网络效应、系统底座与AI代理的“无界面”博弈 对于现有软件工具在未来的存续空间，关键可能在于工具是否具备强大的“网络效应”以及是否为“核心记录系统”（System of Record）。当企业引入Claude等尖端AI模型时，最先采取的往往是将其直接接入Slack——Slack本身的代码实现并没有高不可攀的门槛，其核心壁垒在于“所有人都在上面”。如果AI Agent未来将Slack视为组织内的核心信息中枢，反而会彻底固化其不可替代的底座地位。 然而，企业软件向AI代理时代的迁徙是一柄双刃剑。AI Agent极有可能直接运行在Workday、Salesforce等大型HR与CRM底层系统之上，这带来了两种截然相反的发展路径： - **悲观场景（Headless CRM / 无界面化）**：软件被彻底边缘化为单纯的后端数据库。传统系统赖以生存的是人类操作交互界面，一旦AI代理不再需要任何前端UI，而是通过API直连数据并完成操作，软件公司将丧失与客户直接互动的价值沉淀与溢价能力。 - **乐观场景（固化为执行基座）**：AI Agent直接进入CRM并在其生态内部执行全部业务逻辑，这将使CRM成为AI行动不可或缺的操作土壤，反而巩固了其作为关键记录系统的生存确定性。 --- ### `44:22` 硬件复兴浪潮：数据中心40年未见之大变局 过去40年里，数据中心硬件行业几乎是一片乏味而沉寂的死水。自90年代Intel x86架构统领数据中心以来，即便经历了波澜壮阔的云计算时代，整体计算负载年均增长也仅维持在25%到40%之间；由于摩尔定律（Moore's Law）以相同的节奏稳步推进，硬件不需要突破极限的技术飞跃即可满足需求。这种平稳导致服务器所有环节——从PCB印制电路板、内存颗粒到机箱和网络交换机——全都彻底沦为廉价的“大宗商品”（Commodity），行业升级周期极其漫长，从1G以太网迭代至10G甚至需要耗费整整七年时间。 而如今，生成式AI带来了爆炸式的需求跃迁，数据中心计算负载正以每年10倍（10x）的速度狂飙，直接将服务器每一个物理部件推向了物理学极限。正如三年前红杉资本（Sequoia）的肖恩·马奎尔（Shaun Maguire）所感叹的那样：“真希望重回那个由硬件主导的对冲基金黄金时代，所有公司都是上市公司且拥有极为强劲的知识产权（IP）。”当年的苹果与思科开启了硬件的神话，如今AI基础设施正在上演波澜壮阔的“硬件去商品化”（De-commoditization）与硬件文艺复兴。 - **高带宽内存（HBM）的非同质化**：过去被视为纯周期大宗品的内存，转变为通过先进封装将10层芯片立体堆叠的HBM，I/O传输吞吐量较此前提升了10倍。三星耗费数年攻坚仍困难重重，供应商必须提前三到四代与NVIDIA深度联合研发锁定规格。 - **以太网与网络迭代提速**：过去交换网络从100G、400G演进到800G往往是一个漫长的7年大周期，如今在算力集群狂渴的带宽压力下，升级周期被极限压缩到了几乎每年一更。 --- ### `46:38` 从代工弃儿到天堑壁垒：Celestica与高端PCB的价值重估 硬件去商品化最生动的案例莫过于代工制造企业Celestica。自1999年科技泡沫破裂及产业链向海外转移后，电子代工（EMS）长期被资本市场视为毫无定价权的苦生意，三年前Celestica的市盈率一度被压到仅8倍。然而，该公司骨子里保留着源自IBM大型超级计算机的顶尖工程基因，最终成为Google TPU服务器的独家代工厂商，并牢牢掌控着向头部云厂商供应白盒以太网交换机的业务。 传统数据中心服务器单价仅约5,000美元，损坏后直接整机淘汰；而如今单台运行功耗极高的AI服务器造价高达20万至30万美元，核心故障将直接导致昂贵的高密度集群宕机瘫痪。Celestica攻克了极高难度的“液冷技术”（Liquid Cooling），使其产品角色从“易损件代工”一跃转变为类似“民航客机关键安全零部件供应商”，客户根本无法承受换掉它的风险。不仅如此，Celestica早期工程师深度参与编写了开源网络操作系统SONiC（Software for Open Networking in the Cloud），并与Broadcom（博通）保持极紧密的软硬件配合，如今在云端以太网交换机市场斩获了高达50%至60%的庞大份额。 类似的技术价值裂变同样在被忽视的印制电路板（PCB）及其上游原材料行业中上演： - **PCB板层数暴增**：普通服务器PCB仅需10层左右，而AI训练服务器的主板需求飙升至40层，制造复杂度和良率门槛将绝大多数厂商淘汰出局，仅有极少数顶级PCB供应商能够量产。 - **上游核心材料垄断（台光电子 Elite Material / EMC）**：作为PCB最关键的基础材料，铜箔基板（CCL, Copper Clad Laminate）同样迎来了出货数量与单板层数双升的乘数效应，推动相关产品出货量达到50%至60%的复合年均增长率（CAGR）。 - **供应链地位彻底逆转**：过去客户对上游配件的态度是“下周需要再打电话给你”，而现在必须提前四年协同制定下一代架构路线图（Roadmap）。原本年化增速5%、利润率微薄的周期大宗品制造商，彻底蜕变为未来三至四年顶级营收CAGR可达35%至50%、壁垒森严的核心资产。 ### `50:00` 硬件与供应链的连带繁荣：从光纤、电源到短缺风暴 整条 AI 硬件供应链正在经历一场前所未有的景气周期。在当下的供应链全链条中，不仅毛利率持续扩张，而且几乎所有元器件都在面临全面短缺。即便某些环节在传统认知中被视作周期性“大宗商品”（commodity），在当前的严重供不应求下也将迎来超级周期。以生产光纤的康宁（Corning）为例，他们在数据中心特种光纤领域占据了极高的市场份额。微软近期新建成的一座数据中心内部署的光纤总长度，足以绕地球整整四圈半；康宁的光纤更细、更耐弯折，且完全按照最严苛的定制规格生产，不仅毛利率更高，更是公司目前增速最快的业务板块。 在网络架构层面，需求正在向三个维度纵深爆发：一是连接机架之间（rack-to-rack）的 **Scale-out**；二是由于单体数据中心难以满足极端电力需求、必须跨数据中心拉通训练集群的 **Scale-across**，这要求通信线缆的厚度和传输能力实现 10 倍级的跃升；而最大的爆发点则在于机架内 GPU 与 GPU 紧密互连的 **Scale-up**。目前 Scale-up 主要依靠铜缆，但行业普遍预期未来必将转向光纤互连，一旦该转变发生，康宁所面对的市场空间将直接翻上两到三倍。与此同时，整机柜中的每一个供电组件都承受着巨大压力——英伟达每一代新芯片或新机柜的功耗都会增加 50% 到 125%，直接拉动了台达电（Delta Electronics）和 Advanced Energy 等电源厂商的平均售价（ASP）。Alex 甚至惊叹于这种反直觉的爆发力：“当我听到他们的故事时我几乎难以置信：你们的平均售价在未来四年里每年要递增 40%，而且利润率还在上升？” 放眼全局，只要这波以“L型/S型曲线”为代表的 AI 真实需求逻辑成立，整条电子供应链就已经被彻底拉爆——当前的 DRAM 内存市场、NAND 闪存市场以及高规格 PCB 市场，实际上已经存在约 30% 的结构性供给缺口。在衡量 AI 敞口比例（% AI）和市场份额时，Alex 极为看重“变化率”（Rate of change）而非绝对值。在 2024 年 Whale Rock 内部的一次分析中，团队尝试让 Claude 绘制各家公司的市场份额走势，但模型最初未能准确表达，原因就在于它漏掉了变化率维度的加速效应。当一家公司的 AI 业务占比从 10% 迅速蹿升至 30% 时，其整体营收增速与利润率都会呈现非线性的加速度。 ```mermaid flowchart TD accTitle: AI 数据中心互连架构演进与硬件瓶颈 accDescr: 展示从跨数据中心、机架间到芯片间互连的传输介质与扩展机会 subgraph DC_Network["数据中心互连三层演进"] SA["Scale-Across：跨数据中心互联<br/>解决单点电力受限，线缆需加厚 10 倍"] SO["Scale-Out：机架间互连<br/>海量高耐弯折定制光纤（如康宁）"] SU["Scale-Up：机架内 GPU 互连<br/>当前使用铜缆，未来将向光纤演进（潜在市场扩容 2-3 倍）"] end subgraph Supply_Pressure["供应链核心瓶颈与溢价"] Power["电源供应（Delta / Advanced Energy）<br/>功耗剧增 50%~125%，ASP 年复合增 40%"] Shortage["核心元器件短缺<br/>DRAM、NAND、高阶 PCB 存在约 30% 缺口"] end SA --> Supply_Pressure SO --> Supply_Pressure SU --> Supply_Pressure ``` - **单体数据中心用纤量**：微软近期落成的一个大型 AI 数据中心内，光纤用量足以环绕地球 4.5 圈。 - **Scale-up 光互连催化剂**：机架内部 GPU 互联若从铜线切换至光纤，康宁的潜在市场规模（TAM）将实现 2 至 3 倍的扩张。 - **元器件短缺广度**：即便当前产能全开，全球 DRAM、NAND 闪存和高端 PCB 市场的实际供应仍落后于潜在需求约 30%。 --- ### `53:56` 为什么公开市场极少有人能把握这种科技周期？ 许多人困惑于 Whale Rock 的投资逻辑——既然核心理念无非是“S曲线 + 持续竞争壁垒 + 被低估的盈利能力”，这套剧本在过去二三十年里早已上演过多次，为什么公开市场上能够真正赚到大钱的基金依然寥寥无几？Alex 笑称自己的母亲也曾问他：“你为什么要到处跟别人讲你的秘诀？”他的回答是：“这就像赌场公开教所有人怎么玩二十一点一样——规则大家都懂，但要真正执行到位、忍受波动并做对决策，难度极其巨大。” 这背后的根本门槛在于认知深度与全局视野（Holistic view）： 1. **对硬件与半导体的长期漠视**：在长达十几二十年的软件主导周期中，绝大多数科技投资人对芯片与底层硬件缺乏积累，市场突然转向硬件复兴时，涌入的大多是缺乏经验的新手，市场上唯有极少数像 Whale Rock 和 Gavin Baker（Atreides Management）这样深耕底层硬件周期的团队能够游刃有余。 2. **“恐高”心理与估值偏见**：这类领军企业的股价走势图往往处于历史最高点，对普通投资者而言，在连续大涨后“敢不敢买”是一个巨大的心理关卡。 3. **缺乏底层逻辑带来的周期动摇**：过去四年里，英伟达每一次交出惊人业绩后，市场都会充斥着“这绝对是周期顶点/泡沫”的质疑声，随后股价经历长达半年的横盘整理，令信念不足者仓皇出逃；紧接着英伟达又会迎来更加强劲的一年。 4. **单点分析师的“管中窥豹”**：纯粹跟踪半导体行业的分析师往往会错过这一波大牛市，因为他们只盯着传统的芯片出货量与库存指标，根本看不懂基础模型层（Foundational model layer）的技术突破正在怎样倒逼计算架构发生范式转移。只有拥有宏观视角的科技通才，横跨应用、算法与芯片，才能穿透喧嚣建立坚定的持股信念。 > “如果你只是一名半导体分析师，很多半导体分析师彻底错过了这一轮行情，正是因为他们看不到基础模型层正在真实发生什么，以及整个宏观图景如何铺展。” --- ### `55:36` AI 终局叙事的核心风险：从模型停滞到巨头退场 在一片极度乐观的情绪之中，Alex 依然保持着敏锐的风险意识。首要的外部风险来自于公众与监管层面的逆风。当前普通大众对 AI 抱有较强的负面与恐慌情绪，相关调研显示仅有约 20% 的人对 AI 持明确乐观态度；美国缅因州近期甚至出台了限制数据中心建设的禁令。尽管“AI 这一阿拉丁神灯的精灵已经从瓶子里放出来了”，但监管的收紧依然是一个不可忽视的变量。 技术与商业层面的关键风险则集中在以下维度： - **模型能力的迭代是否会撞墙（Hit a wall）**：即使大模型自此停止进步，仅凭现有能力也足以支撑海量的产业落地应用。但黄仁勋多年前针对图形芯片说过一句警醒的名言：“如果‘足够好’真的就足够了，我就不会再有生意可做（If good enough is good enough, I won't have a business）。”人们之所以愿意不断为新硬件买单，是因为黄仁勋每年都能做出性能更强的大芯片，而用户永远在追求极致。假如 Anthropic 或 OpenAI 的前沿模型提升停滞，开源模型便会迅速追平差距，行业将演变为残酷的“逐底竞争”（Race to the bottom），这对模型公司的估值将是沉重打击（不过对英伟达等芯片商而言却并不完全是坏事，因为黄仁勋在 GTC 上反复强调他极度乐见开源生态的繁荣，卖铲人不在乎最终哪家模型胜出）。 - **云厂商与超级大买家的资本开支退潮**：若有核心玩家在军备竞赛中掉队并骤然削减算力需求，整个供应链将被迫承受巨大缺口。尽管此前甲骨文（Oracle）取消了一笔巨额算力订单后 Meta 瞬间接盘将其吸纳殆尽，但设想一下，如果 Meta 某天突然宣布“我们跟不上了，继续军备竞赛纯属浪费资源”并退出竞争，将会对算力需求造成何等冲击。 - **自研芯片的替代威胁**：以微软为代表的超大规模云服务商（Hyperscalers）正在不遗余力地加速推进自研芯片，试图摆脱对第三方算力生态的单一依赖。 --- ### `57:47` 为何当下对应用层保持克制：历史滞后性与护城河之争 在黄仁勋所定义的“五层蛋糕”（从芯片、系统、网络、模型到应用）中，科技史的一贯法则是应用层（Applications）最终捕获整个生态体系中绝大部分的市值与商业利润，正如智能手机时代的苹果与移动 App。然而，Whale Rock 现阶段却极少出手 AI 应用层，这一反常选择源于其对科技周期的深刻洞察。 首先，**应用的爆发在历史上永远是滞后的**。回顾 iPhone 问世的前三到四年，硬件基础设施铺设才是主旋律，真正的超级杀手级移动应用直到基础设施高度普及后才开始大规模涌现。其次，**当前 AI 应用层的生态格局极为混沌且充满危险**： - **护城河的不确定性**：模型层的边界究竟停留在哪里？应用层的能力从哪里开始算起？上游基础模型能力的快速演进，极容易通过“向下集成”直接抹杀纯套壳应用的存在价值。 - **传统 SaaS 巨头的防御之困**：市场原本预期能够看到 Salesforce（CRM）等传统企业级软件巨头率先在 AI 应用上兑现爆发式增长，但现实进展却相对平缓，目前仍未在企业级领域看到决定性的胜负手。 - **标杆案例的早期观察**：Alex 团队正在密切跟踪由前 Salesforce 联席 CEO、Google 地图编写者、前 Facebook 首席技术官 Bret Taylor 创立的企业级 AI Agent 平台 **Sierra**。Alex 认为 Sierra 正在将 AI 落地推进到“真刀真枪的决胜期（where the rubber hits the road）”，展现出了极高的工程水准和商业潜能，但这类公司能否最终沉淀为不可撼动的庞然巨头，仍需时间验证。 --- ### `59:40` AI 时代的投研本质：Whale Rock 奖项墙与费雪的“草根调查”法 在 Whale Rock 的办公室里，有一面极具象征意味的巨大“研究成果奖项墙”（Award wall）。在过去二十年的历程中，每年都会有一位或多位分析师凭借最具洞察力的年度投研项目将自己的名字刻在墙上。随着 AI 工具的爆发，今天借助 Claude Code 等工具可能在一小时内就能完成 2009 年足以拿奖的量化建模与报告撰写工作，这让外界十分好奇 AI 究竟如何重构了 Whale Rock 的投研底座。 Alex 坦诚而清醒地表示，AI 固然帮助团队大幅提升了信息收集与快速入门的速度，但这并没有从根本上颠覆分析师的核心职能。Whale Rock 坚守的投研精髓，本质上依然直接继承自菲利普·费雪（Philip Fisher）在 1950 年代名著《怎样选择成长股》（*Common Stocks and Uncommon Profits*）中提出的**“草根调查法”（Scuttlebutt Approach）**： - **穿透噪音的面对面交流**：分析师必须走进真实世界，尽可能多地去拜访每一家能见到的产业链公司，与所覆盖的管理团队建立深厚信任纽带。 - **交叉验证的网状访谈**：深入走访供应商、客户以及直接竞争对手，从产业链上下游各方的微观反馈中寻找关于技术产品、定价能力、交期拉长与管理层诚信的第一手拼图。 - 无论算法和生成式 AI 演进到何种程度，这种基于商业常识、人类信任网络和第一手产业调研构建的核心壁垒，依然是识别超额阿尔法（Alpha）不可被机器取代的根基。 ### `01:00:00` AI 在投资研究中的边界：是记录员，而非预言家 Alex Sacerdote 坦言，生成式 AI 正在重塑对冲基金的投研日常，尤其在面对像 ABF 载板（ABF substrates）或印制电路板（PCB）这类复杂晦涩的新硬件供应链时，研究员可以借助 AI 在极短时间内完成基础知识的补课。Whale Rock 如今也广泛使用 AI 来起草备忘录、梳理财报季度表现。AI 产出的初稿甚至比许多初级分析师写得还要好，但它绝不可能代替人去选股。 Alex 强调，如果分析师的工作仅仅停留在基本的操作战术（blocking and tackling）和事实罗列层面，那么这类工作很容易被 AI 替代。分析师最不可替代的核心价值是凌驾于信息之上的智慧（wisdom）与洞见：这份财报究竟意味着什么？它如何影响我们的核心投资假设？核心变量发生了什么转变？"不要只做一个事实记录员（Don't just be a reporter）。AI 可以成为一名极其出色的速记记者，但它永远无法看透未来。" 为了说明这种人类特有的“非共识深度研究”，Alex 举了团队对移动广告技术巨头 AppLovin 的经典下注。移动广告技术（AdTech）在二级市场历史上往往被视为商业模式较差的泥潭行业。然而，Whale Rock 的两位科技分析师 Michael 和 Sam 凭借极深的行业认知，在 AppLovin 仍处于一级市场私有状态时就长期紧密跟踪。他们飞赴拉斯维加斯参加移动应用广告大会，专程前往戛纳国际创意节（Cannes Lions），与数十上百位行业一线人士进行深入访谈，彻底吃透了复杂的行业术语与竞争格局，并在模型上做足了功夫。不仅如此，他们还与 AppLovin 的联合创始人兼 CEO Adam Foroughi 建立了极其深厚的互信关系。Alex 坚信，这种建立在人与人深度连接、现场实地尽调以及第一性原理洞察之上的超额认知，绝非任何大语言模型所能模拟。 --- ### `01:02:53` 投资圈的人脉网络与“三脚架”审视法则 谈及与其他顶级投资机构交流的作用，Alex 提到了成长股投资先驱菲利普·费雪（Philip Fisher）的经典方法论：在全国范围内结识 10 到 15 位志同道合的高水平同行并相互分享思路。多年来，Alex 在业内结下了许多深厚的友谊，这些交流绝非单向的信息索取，而必须是双向奔赴的智识互惠。 Alex 将建立高确信度的验证机制形象地称为“三脚架（The Tripod）”模型： ```mermaid flowchart TD accTitle: 三脚架决策模型 accDescr: Whale Rock 内部用于确立重仓股高确信度的三方交叉验证架构。 subgraph Tripod["三脚架确信度模型 (The Tripod)"] A["创始人 / CIO 认可<br/>(Alex Sacerdote)"] B["资深行业分析师深度验证<br/>(Internal Analyst)"] C["高度尊重的外部同行共鸣<br/>(Respected Peer)"] end A --- Decision(("坚不可摧的重仓确信度<br/>(High Conviction)")) B --- Decision C --- Decision ``` - **第一根支脚**：CIO（Alex 本人）从 S 曲线与大局观出发认同该逻辑； - **第二根支脚**：团队内部专注于该细分领域的分析师通过底层基本面与财务模型深度论证并看好； - **第三根支脚**：一位在外部市场中令其极度尊重的同行也独立得出了高度看好的结论。 当这三根支腿稳稳立住时，一个投资想法的确信度就会发生质的飞跃。 --- ### `01:03:40` 多产品架构演进与超大市值科技股的“结构性低配” Whale Rock 创立至今已有 20 年历史。在前 15 年里，公司为了保持极致专注，只运营旗舰的多空对冲基金（Long/Short Fund）。随着资金规模增长与策略成熟，大约在第 10 年左右，投资人开始主动寻求纯多头（Long-Only）策略；到了 2015 年前后，公司正式引入一级市场投资选项，允许 LP 自主选择是否参与以及配置比例（如 15% 或 25%），直到 2020 年才正式在一级市场落子，进而演化出覆盖对冲基金、纯多头与混合型产品的多产品矩阵。 在拓展产品的过程中，Alex 洞察到了全球机构资金配置上的一个重大历史级盲区：**对全球超大市值科技股（Mega-cap Tech）的系统性、结构性严重低配（structural underweight）**。 像大学捐赠基金（Endowments）等主流大型机构资金，其资产配置往往存在以下扭曲现象： 1. **重仓一级、轻视二级**：大量资产配置于风险投资（VC）和私募股权（PE），导致公开市场权益仓位本身就不高； 2. **地域分散带来的科技稀释**：在有限的公开市场股票组合中，往往有近半数资金投向了缺乏原生大型科技巨头的国际市场； 3. **“大盘股无 Alpha”的思维定势**：绝大多数捐赠基金笃信超大市值股票研究充分、不存在超额收益（Alpha），因此在二级市场更青睐委托中小型股票精选基金（SMID managers）； 4. **对冲基金的风控限制**：即便其配置的多空对冲基金带有做多偏好，受限于风险集中度敞口规则，它们也很难单票给 NVIDIA 这样的巨头配置 15% 以上的高权重。 然而，Alex 认为数字经济的底层规律已经发生了根本性转移：在网络效应与软件飞轮的驱动下，数字科技领域的赢家往往呈现爆发式扩张，以极快的速度吞噬超高市场份额，且其产品能够无阻力地行销全球。这必然会导致史无前例的超级利润池与超大市值公司。全球大型捐赠基金系统性低配这些巨头，实质上是在与科技演进的客观经济规律背道而驰。 反击“大市值没有 Alpha”的论调，Alex 给出了极具杀伤力的对比：小盘股的上涨可能只需要一两个精明买家发现其价值就能被推上去；但像 Alphabet 这样万亿美元级别的庞然大物，如果市场普遍误认为它是大模型时代的“输家”，那么要逆转这种偏见并推升股价，需要让市场上 100 位掌管分散化投资组合的通才型基金经理（Generalist PMs）彻底扭转认知。Alex 充满信心地反问：“我们能否比市场上 95% 的通才经理更早看清 Google 绝非输家而是大赢家？我们做到了。在这一领域，深度科技专家的 Alpha 概率极高。” 更重要的是，这些巨头拥有无与伦比的深厚护城河，且同样能搭乘超级 S 曲线的东风——NVIDIA 正在经历狂暴的 S 曲线扩张，台积电（TSMC）、SK 海力士（SK Hynix）以及 Whale Rock 刚刚建仓数月的阿斯麦（ASML），都与这波不可逆的算力浪潮紧密绑定。 --- ### `01:07:58` 复合认知网络：10 人构筑的 Whale Rock“学习机器” 在谈到 Whale Rock 的组织本质时，Alex 将整个投研团队定义为一台精密运转的“Whale Rock 学习机器（Whale Rock Learning Machine）”。整个核心团队由 10 位经验极其丰富的资深专业人士构成。 沃伦·巴菲特（Warren Buffett）靠海量阅读书籍来学习，Whale Rock 也读经典、研读前沿技术博客，但面对瞬息万变的科技世界，纸上得来终觉浅，投资人必须深入产业最前线。Whale Rock 团队每年要与企业管理层进行多达 200 到 300 次高强度的面对面交流。巴菲特和查理·芒格（Charlie Munger）终生强调“知识的复利效应（compounding knowledge）”，而 Whale Rock 已经围绕科技这一垂直领域不间断地进行了长达 20 年的知识复利积累。 团队的长期稳定性是这一复利飞轮得以运转的基石： - 核心合伙人 Andrew 和 Michael 已经与 Alex 并肩作战长达 18 年； - 哪怕算上最近加入的新人，整个投研团队的平均投资从业经验也高达 10 年左右； - 统一的研究底座同时赋能所有公开市场与一级市场产品——团队不会盲目去翻找每一个早期的 A 轮或 B 轮项目，但一旦市场上出现符合 Whale Rock 投资框架与认知半径的标的，同一套研究引擎便能以极高的确定性迅速开火。 --- ### `01:09:15` 终极传承与最深善意：父亲在身后的六载护航 在访谈的固定保留问题——“有人为你做过最友善的一件事是什么？”面前，Alex 动容地谈到了自己的父亲。 Alex 的父亲毕业于康奈尔大学电气工程专业，随后转型进入华尔街，并在高盛（Goldman Sachs）度过了长达 41 年的辉煌职业生涯。在 1980 年代，父亲掌管高盛的企业融资业务（Corporate Finance），到了 90 年代则出任高盛私募股权业务（PIA）的主席，并长期坐镇关键的承诺委员会（Commitments Committee），凭借出色的投资直觉与审慎判断屡次帮助投行避开灾难性的交易。 ```mermaid flowchart LR accTitle: 父亲在高盛与 Whale Rock 的角色传承 accDescr: Alex 父亲的履历演变及其在 Whale Rock 初创时期的导师与护航角色。 subgraph GS["高盛 41 年光辉岁月"] G1["康奈尔电机工程背景"] --> G2["80年代：掌管企业融资部"] G2 --> G3["90年代：掌管私募股权投资 (PIA 主席)"] G3 --> G4["兼任承诺委员会核心把关人"] end subgraph WR["Whale Rock 创业启航 (2006-2011)"] G4 -. 退休加盟.-> W1["出任 Whale Rock 首任非执行主席<br/>(提供声誉背书与宏观指引)"] W1 --> W2["Alex 在波士顿招兵买马、专注投研与操盘"] W1 --> W3["父亲在纽约利用声望协助募资并化解疑难"] end ``` 2006 年 Alex 决定创立 Whale Rock 筹集启动资金时，第一通电话打给了父亲。出乎 Alex 意料的是，刚从高盛功成身退的父亲平静地对他说：“你在高盛看了我 41 年。不如我加入你吧？我来做你的白发长者（the gray hair），我来当董事长负责监管把关。你在波士顿招揽团队、搭建公司、专注管钱；我来帮你募资并做宏观指导。” 从创立到父亲于 2011 年不幸病逝，父子二人并肩战斗了整整六年。Alex 感慨，经营对冲基金是一项压力极大的艰巨事业，但在共同打拼的六年里，父亲从未提高过嗓门对任何人发过火。父亲离世后，Alex 收到了雪片般飞来的吊唁信，信中无一不在诉说父亲是如何深刻地影响了他们的一生。父亲极其聪明睿智，却始终保留着惊人的谦逊、温和与幽默感；无论是初级员工遭遇生活困扰还是团队面临重大业务分歧，只要走进他的办公室，他总能以极大的优雅与从容化解难题。 Alex 动情地总结道，拥有这样一位父亲与导师是自己此生最大的幸运：“如果我能成为像他那样的人的一半，我的人生就已经彻底赢了。” ## Notable quotes > `10:00` 未来四年内唯一百分之百确定的事情，就是算力永远不够用。 — Marc Andreessen > `25:08` 把历史走势图涂白抹掉，一切只看未来。 — Peter Lynch > `25:08` 当战略转折点到来时，你不能依赖常规数据。 — Andy Grove > `30:00` 这场战争还没正式打响就已结束。 — Alex Sacerdote > `44:22` 真希望重回那个由硬件主导的对冲基金黄金时代，所有公司都是上市公司且拥有极为强劲的知识产权（IP）。 — Shaun Maguire > `50:00` 当我听到他们的故事时我几乎难以置信：你们的平均售价在未来四年里每年要递增 40%，而且利润率还在上升？ — Alex Sacerdote > `53:56` 这就像赌场公开教所有人怎么玩二十一点一样——规则大家都懂，但要真正执行到位、忍受波动并做对决策，难度极其巨大。 — Alex Sacerdote > `53:56` 如果你只是一名半导体分析师，很多半导体分析师彻底错过了这一轮行情，正是因为他们看不到基础模型层正在真实发生什么，以及整个宏观图景如何铺展。 — Alex Sacerdote > `55:36` 如果‘足够好’真的就足够了，我就不会再有生意可做（If good enough is good enough, I won't have a business）。 — 黄仁勋 > `01:00:00` 不要只做一个事实记录员（Don't just be a reporter）。AI 可以成为一名极其出色的速记记者，但它永远无法看透未来。 — Alex Sacerdote > `01:09:15` 你在高盛看了我 41 年。不如我加入你吧？我来做你的白发长者（the gray hair），我来当董事长负责监管把关。你在波士顿招揽团队、搭建公司、专注管钱；我来帮你募资并做宏观指导。 — Alex Sacerdote 的父亲 > `01:09:15` 如果我能成为像他那样的人的一半，我的人生就已经彻底赢了。 — Alex Sacerdote ## Mentioned Alex Sacerdote, Patrick O'Shaughnessy, Whale Rock Capital Management, Anthropic, Claude, Claude Code, OpenAI, ChatGPT, Google, Google Gemini, 黄仁勋, NVIDIA, AWS, Microsoft Azure, GCP, Meta, Andrej Karpathy, Linus Torvalds, Dario Amodei, Stripe, Adyen, Databricks, Nubank, Tesla, Elon Musk, Apple, Peter Lynch, Andy Grove, Horace Dediu, Clayton Christensen, Oracle, Bloomberg, Salesforce, Workday, Slack, Celestica, Broadcom, 台光电子 (EMC), Corning, 台达电 (Delta Electronics), Advanced Energy, Gavin Baker, Bret Taylor, Sierra, Philip Fisher, AppLovin, Adam Foroughi, ASML, TSMC, SK Hynix, Warren Buffett, Charlie Munger, Goldman Sachs ## Video description My guest today is Alex Sacerdote, founder of Whale Rock Capital Management.  Whale Rock is a technology focused investment firm that manages more than $17 billion across hedge fund, long only, and hybrid strategies. Over the past three years it has been one of the best performing hedge funds, compounding at roughly 44 percent a year. Alex invests through a single lens that he has refined over twenty years. He looks for technology S-curves, durable competitive advantages, and underappreciated earnings power.  This conversation is a tour through how he applies that framework right now. We start with his highest conviction position, which is Anthropic, and use it to work through the entire AI stack from chips to models to applications.  Please enjoy my conversation with Alex Sacerdote. 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.  ----- 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 ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ridgelineapps.com⁠. ----- Editing and post-production work for this episode was provided by The Podcast Consultant (⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://thepodcastconsultant.com⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠). Timestamps: (00:00:00) Welcome to Invest Like The Best (00:02:29) Alex Sacerdote (00:03:08) Anthropic: Highest Conviction Position (00:13:23) Investing in Private Markets at Scale (00:19:08) S-Curves: The Full Framework (00:25:08) When to Buy Tech Companies (00:30:20) Identifying the Leader from the Pack (00:34:04) Anthropic & OpenAI's Competitive Moats (00:37:31) AI's Threat to Enterprise Software (00:43:18) Network Effects in the Agent Era (00:44:22) The Hardware Renaissance: Chips & Infrastructure (00:53:56) Why So Few Investors Get This Right (00:55:36) Key Risks to the AI Bull Case (00:57:47) The Application Layer (00:59:40) How AI Is Changing Research at WhaleRock (01:02:53) The Role of Investor Networks & Idea Sharing (01:03:40) Building a Multi-Product Firm (01:07:58) WhaleRock as a Learning Machine (01:09:15) The Kindest Thing [Source](https://traffic.megaphone.fm/CLS6537493219.mp3)

[10:00] Marc Andreessen: 未来四年内唯一百分之百确定的事情，就是算力永远不够用。 [25:08] Peter Lynch: 把历史走势图涂白抹掉，一切只看未来。 [25:08] Andy Grove: 当战略转折点到来时，你不能依赖常规数据。 [30:00] Alex Sacerdote: 这场战争还没正式打响就已结束。 [44:22] Shaun Maguire: 真希望重回那个由硬件主导的对冲基金黄金时代，所有公司都是上市公司且拥有极为强劲的知识产权（IP）。 [50:00] Alex Sacerdote: 当我听到他们的故事时我几乎难以置信：你们的平均售价在未来四年里每年要递增 40%，而且利润率还在上升？ [53:56] Alex Sacerdote: 这就像赌场公开教所有人怎么玩二十一点一样——规则大家都懂，但要真正执行到位、忍受波动并做对决策，难度极其巨大。 [53:56] Alex Sacerdote: 如果你只是一名半导体分析师，很多半导体分析师彻底错过了这一轮行情，正是因为他们看不到基础模型层正在真实发生什么，以及整个宏观图景如何铺展。 [55:36] 黄仁勋: 如果‘足够好’真的就足够了，我就不会再有生意可做（If good enough is good enough, I won't have a business）。 [60:00] Alex Sacerdote: 不要只做一个事实记录员（Don't just be a reporter）。AI 可以成为一名极其出色的速记记者，但它永远无法看透未来。 [69:15] Alex Sacerdote 的父亲: 你在高盛看了我 41 年。不如我加入你吧？我来做你的白发长者（the gray hair），我来当董事长负责监管把关。你在波士顿招揽团队、搭建公司、专注管钱；我来帮你募资并做宏观指导。 [69:15] Alex Sacerdote: 如果我能成为像他那样的人的一半，我的人生就已经彻底赢了。

[41:40] Alex Sacerdote: Andy Grove says, when you have strategic inflection points, you can't trust the data. Strategic inflection points are about intuition, anecdotal evidence. [44:22] Alex Sacerdote: I met with Sean McGuire like three years ago, and he said, 'I wish I could come back and be a hardware hedge fund because all the companies are public and they all have powerful IP.'... We are in this renaissance of chips. [53:56] Alex Sacerdote: My mom said, 'Why do you tell everyone your secret?' It's like, why does the casino teach people how to play blackjack? It's really hard to do.

[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 customers grew revenue 3.2 times faster than the average American business. Visa, Vercel, Cursor, Stripe, Notion, 11Lab, Shopify, and 70,000 other businesses all run on Ramp. Mine does too, and so should yours. Learn more at ramp.com slash invest. Felix by Rogo is a personal finance agent that turns a single prompt into finished client-ready work using your firm's own templates, context, and standards. Send Felix an email like, take these comments and turn them for me, or update my tracker with the context of these emails, or run the ability to pay math on this buyer, and Felix sends back finished PowerPoint decks, Excel models, and sourced research. Felix works the way your team already does, delivering work quickly and accurately around the clock. Learn more at rogo.ai slash Felix. 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. Hello and welcome everyone I Patrick O and this is Invest Like the Best This show is an open exploration of markets ideas stories and strategies that will help you better invest both your time and your money If you enjoy these conversations and want to go deeper, check out Colossus, our quarterly publication with in-depth profiles of the people shaping business and investing. You can find Colossus along with all of our podcasts at colossus.com. Patrick O'Shaughnessy is the CEO of Positive Sum, all opinions expressed by Patrick and podcast guests are solely their own opinions and do not reflect the opinion of Positive Sum. This podcast is for informational purposes only and should not be relied upon as a basis for investment decisions. Clients of Positive Sum may maintain positions in the securities discussed in this podcast. To learn more, visit PSUM.VC. My guest today is Alex Sacerdote, founder of WhaleRock Capital Management. WhaleRock is a technology-focused investment firm that manages more than $17 billion across hedge fund, long-only, and hybrid strategies. Over the past three years, it's been one of the best-performing funds, compounding up roughly 44% per year. Alex invests through a single lens that he has refined over 20 years. He looks for technology S durable competitive advantages and underappreciated earnings power This conversation is a tour through how he applies that framework today We start with his highest conviction position which is anthropic and use it to work through the entire AI stack from chips to models to applications Please enjoy my conversation with Alex Sacrodote. Alex, you were saying that your highest conviction position is anthropic right now. Can you tell the story of discovering it, making the investment, using this anecdote as an excuse to talk about all the things that I think you and I are mutually interested right now. Investors like you investing in private markets, entropic to business, AI, everything. It's a great way to zoom in. Why is it your highest conviction and how did you get started? When the gun went off with OpenAI, ChatGPT in November 2022, we immediately took the firm and did a massive deep dive with our 10 person team. Anytime you have a new compute paradigm, there's a new stack and that creates new winners and losers on the old stack. Now Jensen talks a lot about it, but it's power at the bottom, chips at the bottom, the clouds, and then the foundational models, and then the applications on top. And at that time, this was 2023 early, we said, we want to be in the chips and the infrastructure first And not only do they get the demand first but we know who the winners are And no matter who wins above which we weren sure at the time we know we going to need tremendous amounts of compute And we did a deep dive into that, which we can talk about later. But over the next two or three years, we started to get more clarity on how the foundational model layer would evolve. And at the time, two or three years ago, there were 60 different companies going after it. OpenAI was kind of in the lead. And we did a webinar in April 2023. And we said, look, this might be a winner take all. It might be a total commodity because there's open source players. It might be a race to zero, or it might be an oligopoly where there's three or four leading players. And what we saw over the following three years was that almost all the startups fell away and died. And then some of the largest companies in the world, including Amazon and others and Meta, Amazon never really showed up. We'll see what happens with Meta, but they came in strong. Basically, their effort faltered and they had to do a total reboot. In the meantime, Anthropic, Co-Cup, [5:00] and it was this dark horse candidate, the startup, they focused really purely on the enterprise. OpenAI had kind of won the consumer, and then Gemini can never be counted out. We love Google as well. It's one of our largest positions. So it really started to look like a three-horse race and somewhat of an oligopoly. Very similar to how the cloud market evolved, where three companies underpin the entire SaaS cloud world and have really excellent businesses. And then we also were aware of the open source risk. From China, we started to get comfortable that the quality of the tokens from the leading edge were superior because if you're 80% close to the top of the benchmarks, going from 80 to 85 is a huge unlock. The open source guys, they don't have as much compute, so they can come close to the leading edge, but they can't leapfrog it, and then they kind of falter. Meanwhile the scaling laws and other means of improving the models the feedback loops etc we saw that there was a very strong runway and everyone we talked to close to the industry saw that the scaling laws would continue We developed this thesis that it would be a three-horse race. The big kicker was code, and this is the true unlock of AI. In the first few years, we knew AI would be big, but we were skeptical also. We made large investments because we knew the training would be there, but we weren't sure how much revenue might come and if it could truly replace labor. Because if you remember, the early versions of the models were good, but there was a lot of negative feedback from corporates. And could they be truly agentic? In 2025, the first Claude code and the coding tools really began to explode. You saw the first gen was like Microsoft Copilot, which is like $20 a month. And that could sort of improve your grammar of coding, maybe find a bug, maybe make a block of code like a paragraph. And then Anthropoc came out sometime in the middle of the year, and it could do so much more. And it started to get to this point where it could run agentically, and the coding market just exploded And then we started hearing people who could use it unfettered We heard that even within Anthropic at that time people were spending a day on tokens which if you do the math comes out to $20,000 or $30,000 a year. And if you think about how many coders there are in the world, 20 million, you've got a half a trillion dollar market just from coding alone. And mind you, that was on seven, eight, nine-month-old technology. We could see just on the coding market alone that Anthropic had a tremendous opportunity ahead of it. So we made the investment at the 180 valuation, we said, and I think they were hoping to get to a 9 billion. Yeah, 1 to 9, yeah. One to nine. Yeah. And then the numbers were like nothing we'd ever seen before. One hundred to a billion on the way to nine. But when we did it in August of 2025, nobody had any idea what 2026 could be. The second big unlock lately is that Claude Code has gone to almost completely agentic. You had Andre Karpathy and Linus Torvalds, two of the smartest people in coding, and they completely flip flopped And Karpathy said last year code tools could write 20 and 80 would be handwritten That flipped when the latest model came out and now he hasn't written a line of code except in English. And not to mention the pure unlock that we're going to get for the people that never knew how to code. So just coding alone has completely taken off. One difference between the cloud, GCP, AWS, and the AI companies is the clouds, generally it's commodity. They're selling you servers and storage. They have a lot of software on top, and there is stickiness to it. But in the AI models, everyone thought it would be pure commodity, but there's tremendous differentiation within. There's different training methods and different skills that they're good at. And a lot of people have routers that switch in between, which sort of makes it sound like their commodity. But Anthropic, they're very good for anything that has to do with private equity and finance. Google's very good for ingesting PDFs. There's a lot of like differentiation, critical IP, which is a great competitive advantage. Many companies have come after the coding franchise and Anthropic has been able to keep [10:00] ahead. The other thing that's good about the foundational models and Anthropic is it's not just the API or the model. They're building a whole monopoly or a whole ecosystem of products around the API. So they've got the SDK, Cloud for Cowork, Orchestration Layer, and all the tools. They call it sort of a harness, which is the software around the API that gets the most out of the model. This was one of the things we saw with AWS really early on in 2013 was, oh, people thought it was a commodity server up in a warehouse, big deal. They saw this was a new way of doing computing. So they invented all these products that they could see before everybody else that slowly built lock-in. The other way we think about this is where are we on this S-curve? And we have this infrastructure layer S curve, which we think is 10% penetrated. And by the way, we think it's still one of the best ways to play AI. And we'll talk about how that feeds back through. But if you think about it 200 or I don know how many 800 million people are using AI They just using AI 1 which is like a search engine on steroids But now with these new primitives where you have clawed on your computer linking it in then you build skills, and then they're going to build true AI bots. Big corporations are going to build much larger. But where are we in terms of the amount of people doing that? I mean, Sunder said it's 10 bips of the knowledge workers of the world. So Anthropic has something like 14 or 15 million DAUs, probably a small portion of those are truly doing AI the way you can do it. So that 10 bps, it's classic S-curve where these are the tinkerers, and then it's going to go to the early adopters, then it's going to go to the early mainstream. But you're going to go from 10 bps to 1% to 2% or 3% to 5% to 15% in the next four years. And kind of a light switch this year went off in the enterprise where everybody realizes they need to do this now and do it fast. But it's still Internet 1.0 when it's like you knew you needed a website in 1998, but it's like hard to build that website. But this is coming together fast. And so the enterprise AI or enterprise application AI market is less than 1 penetrated And you know we talk about S curves we call this an L curve just straight up We take this to the infrastructure. We're at 10 basis points of people really using AI. And there's not enough compute in the world. So Anthropic has half of what they need right now. And that's before this huge take up. Mark Andreessen said in the next four years, one thing he's sure of is there's not going to be enough compute. I'm so curious when an investor like you, who historically was a public markets investor, you could hit buy and buy whatever you want, is now operating in lots of the most important private market companies. We can talk about Stripe or Databricks or OpenAI or Anthropic. How do you get the positions at the size that you want coming from the legacy of being able to just buy. How much of it is creativity just directly with the company? If it is directly with the company, so it's a double opt-in, they have to decide to let you in too. How do you do that? What have you learned about getting the allocation you want or the amount of equity you want in a private company, given that that wasn't your original background? We got to know the company One of our analysts knew people in the finance group there We had a look at the billion round and we didn know the company as well The gross margins were negative And frankly we hadn seen coding explode the way it had. And one thing about public markets is you get to know companies over a long period of time, and you can kind of invest on your own schedule. Then I got a chance to spend some time with Dario. I started to realize these guys, their management team is excellent. The focus, the dedication, they had almost no turnover, the quality of code. And then the business plan was really starting to play out. It's one thing to grow from 100 to a billion, but it's another to do nine. And then so we reached out to the company as much as we could. They took a meeting with us. We did a 90-page PowerPoint deck where we used Cloud Code to scour the internet for all the feedback we could about the coding market and what their products were good at, where they might need to improve. And we also did our whole overview of what the coding market would be. They welcomed us into this round, and then we stayed close with the CFO. It's been great to build a relationship with them, and I think we punched above our weight in terms of the allocation. So that one was a, [15:00] total home run. We are in this period where the unicorn market is bigger than most stock markets in Europe, maybe even combined. It's definitely bigger than Germany. It's definitely bigger than the UK. Even before we invested in privates, the first one was 2020. We have to know these companies and you really have to know them now because sometimes they're the biggest companies in the space and have huge impacts. So we do 2,000 to 3,000 face-to-face meetings with management teams a year, and about 10% or 15% of those are with privates. And then we kind of focus in on the companies that we really want to learn about and find ways to meet with them, get involved in the rounds. Our first one was Stripe. We had a large investment at the time. This is 2017, 18, 19, and 2020. We own Audion, which is a fantastic payments company. And they're a next-gen cloud payments company taking from WorldPay and the cloud modern payments was 5% of total $80 trillion market or what have you. But you can't invest in Audion unless you know Stripe like the back of your hand So we did tremendous amounts of due diligence talked to 200 customers in Audion But when we asked about Audion we asked about Stripe and we realized this is Coke and Pepsi We said we got to find a way to invest And I finally got to meet the Carlson brothers in 2019. And so that was our first one. We weren't really known for privates. I've got a friend who's involved with a venture firm that has tremendous amounts. And I talked to him about it and I said, let me know if you ever want to sell some. And then I get a call from him during COVID in April of 2020, we knew a lot about Stripe. We didn't have the full financials, but we knew enough that at that valuation, I think it was $35 billion, they disclosed we had over half a trillion of TPV. And we knew that Audion's take rate was 25 or 30 bps, and we knew Stripe's was 40 or 50. And we knew how many employees they had, so we could kind of get at the profitability. It turned out the take rate was higher. It turned out they were being modest about their TPV. It was much higher than the 550. It was closer to the $1 trillion. We underwrote the thing under our assumptions, and it was much better. And then we were able to upsize that from the seller to a $100 million block. The VCs are going to own and then most of them are going to sell They like it that we own and own in the public market which we did with Nubank as well owned it for a long period of time in the public market as well Vanta automates security and compliance for over 16,000 fast-moving companies like Ramp, Cursor, and Harvey, keeping them audit-ready around the clock. It's the number one agentic trust platform, and it now helps companies like yours watch for the risks that show up between audits across your vendors, your AI tools and your whole environment. Every new tool your team signs up for, every vendor that turns on AI features is an opportunity for something to go wrong. And most security programs weren't built for AI's pace of growth. The Vanta agent works like a 24-7 GRC engineer in the background, finding issues, drafting fixes for you, and cutting vendor assessment time by up to 50%. Whether you're a fast-growing startup or a global enterprise, Vanta helps you earn and prove trust. Invest like the best listeners get a special offer for $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 onto Ridgeline because of how far ahead Ridgeline AI features are compared to anything else in investment management software which is why I believe that firms that come out ahead in the AI era will be the ones running on Ridgeline unified platform If you serious about your firm's AI strategy, Ridgeline should be part of that conversation. You can request a demo at ridgeline.ai. Maybe now's the right time to lay out everything you've ever learned about S-curves. Obviously, your firm is sort of predicated on this idea of technology adoption life cycles and investing in companies at the right time amidst a certain platform change or S-curve change. I'd love you to go into the super deep detail of what you've learned since this is the lens through which you've viewed markets and stocks for a long time. Bring us into like the nitty gritty, fine grating, nuanced detail of why S-curves can be so useful for investing. We have an investment framework. It's S-curve, competitive advantage, and then underappreciated earnings power. And when you get the right part of the S-curve, you get exponential unit growth. If you have a very strong business model, which in tech, there's so many of those for so many different types of moats, your earnings don't grow linearly. They grow exponentially. And that's the last piece. Invest when there's underappreciated, [20:00] long-term earnings power. And very often the earnings can grow from $1 to $10, and it happens way more than you think. And it allows you to buy some of the best companies in the world for extremely low PEs. When we were buying NVIDIA in 2023, we were paying four times earnings. When we bought Tesla in 2019 for the car S-curve, we were paying five times earnings. When we were owning Apple, we were paying four times earnings. When we bought Amazon for AWS, we were getting it for free. The world doesn't think exponentially, and they're so focused on the next year, the next quarter. Very few people believe you can accurately predict two, three, four years out. But if you follow and understand the S-curve, you know the moats and you know how to model, you really can predict these great things. So let's go to the S-curve. So the S-curve is crucial because every technology follows this pattern where it comes out. The smartphones were out 10 years before the iPhone The internet was out 20 years before Netscape AI has been out hidden inside of these companies but it wasn until ChatGPT took it public and ignited what it was So electric vehicles Tesla went public 15 years before 2019 when it went vertical because there were so many barriers to adoption. The first smartphones, they were clunky. They didn't have touchscreen. There wasn't a wireless data system. And then they were too expensive. They were $500 or $600. Steve Jobs got the price to $200. There was AT&T at a 3G network. It was touchscreen. It was so easy. Your grandmother could do it. And he built an ecosystem and made it simple. So all the barriers to adoption were eliminated. And then you rocket when those barriers are removed. That's the tornado of demand that everybody in the world knows they need this right away. And so that's the flip that happens. It happened with electric vehicles. The price was too high. Elon got the price to 40,000. Range anxiety was there. He got the range to 300 miles. The supply chain was finally in place so he could churn out millions of these things. That triggers the inflection. Now, the other nuance, it's not just, oh, it's taken off now. It how big is this S how tall it is so you know when to sell how long to hold on because we underwriting out two or three years We have to know what the growth looks like thereafter And these S-curves can be dynamic. So when Amazon had AWS and it was a hidden line item inside of Amazon covered by retail internet analysts, we realized the TAM for this, it was the largest TAM in enterprise IT ever, because previously the TAM was routers, memory, storage, Dell, EMC, but they were doing it all. You want to know how tall the S-curve is. So we figured out they were addressing 600 billion of IT systems directly addressing that. And then we said, it's probably going to be 50% deflationary, and then therefore we're one or 2% penetrated. But then over time, we realized it actually wasn't deflationary. If you talk to anybody now, I say if you build it yourself, it's about the same price. So that means the TAM was so much bigger. There's mega S curves and there's sub S curves. We been lucky that we had Internet 1 mobile cloud e and now AI which we can confidently say is the biggest And all these things build upon one another With the electric vehicle S you have to pay attention too because we thought probably maybe 40 to 50 percent of the cars would go electric, but it did hit a big wall at 10 or 15 percent. Usually the S-curves go kind of all the way, but in this case, for a variety of reasons, it didn't. So you have to adjust and you have to stay on top of it. And generally, when something gets to sort of 30, 40 percent penetrated, then you stop having exponential growth, which means the sell side catches up and there's no longer big beats. And is that when you sell? Generally, we like the high growth. And it was a mistake with Apple because in the first five or six years of Apple, it was awesome. I mean, it was our largest position. It would go up 50, 70 percent a year, except for 08. And then we sold in 2012 when it got to sort of 50% of the US had a smartphone. And with Apple, they maintain their leadership position. It had a couple of years of underperformance and then the multiple got low and they added several ancillary things. And then they also got to play in the application because they get 30% of the app. So they were able to compound very nicely. [25:00] say 20 percent. But the big years were in zero to 50 percent part of the curve. I'm so fascinated by this sometimes decade plus long flat line at the beginning of one of these curves, which makes me wonder what you've learned about the right moment to buy or even start paying attention before you buy. How do you measure that? Is it always different? What are the pitfalls that you've fallen into? How do you know when start thinking about buying in one of these things? Andy Grove says, when you have strategic inflection points, you can't trust the data. And strategic inflection points are about intuition, anecdotal evidence. I love this book called The Tao Jones Averages, A Guide to Whole Brain Investing, which is right brain and left brain. And the best investors have the right, the creative side where it's visual, it's connecting the dots. we invested in the mobile video game s curve for so long the screens were small on the phones and the processing power wasn't good so you had all these casual games but then i was in china and i saw this little 12 year old boy with a huge phone and he was like playing an awesome video game i'm like oh my god it now coming to the phone so it visual enterprise is hard because you can see it Smartphone you can see oh my God I got it It amazing With AI there some intuition there But enterprise, you don't really get to do that. We go to the Gartner IT Symposium, 30,000 American CIOs go there. We saw this happen with Splunk, where that used to be an amazing database company. And their room where they were explaining it was like standing room only. We saw that with VMware 30 years ago where they virtualized a server. There was standing room only, and you could just see the corporate demand just beginning. And with AWS, we went there, and their grand ballroom was completely packed. And that was at 9 o'clock. And at 10 o'clock, the grand ballroom was completely packed. 11 o'clock, so you could see the demand exploding before it happened. We look for all kinds of clues and there's a whole pattern recognition that happens. And by the way, it's OK to be late. It's OK to miss the first one, two, three years in a lot of cases, because if the top of the S curve is half a trillion the growth can go on for a long time So you don always have to be right there It OK to miss the first 100 Peter Lynch I started at Fidelity and he loved to mentor the kids So I got some time with him He said, white out the chart. It's all about the future. So it's okay to miss. But what helps about the S-curve is sort of how long it goes for. Then there's the slope of the S-curve, which is important. And a lot of people think, because we're in a modern world, everything's so fast. but there's a lot of factors that determine the pace of the adoption and we commissioned this gentleman Horace Dadu used to work with Clayton Christensen to go look in history and we have the big s-curves on our wall over the last hundred years and the radio s-curve is one of the fastest ever it took seven years to reach like a hundred percent penetration but the dishwasher s-curve is like that because it needs to be plugged into the back end. B2B stuff can take a long time because it needs to be plugged into the existing systems. Consumers generally tend to go a lot faster. I love that, the radio and the dishwasher, the two models for adoption. I covered internet at Fidelity My first stock was Amazon That a whole other story which is a lot of fun But I also did B2B internet and there was a whole huge bull case on that The underlying infrastructure wasn in place for B2B to happen Ultimately it happened 20 years later with SaaS That is a risk with AI in that these big companies are very security conscious, can be slow to move. There's a lot of cultural issues with AI where you really need a few evangelists to push it through. The top management needs to push it through, but then the IT is saying this is risky. And that happened with cloud, too. That was one of the big things with cloud where everybody was afraid it's unsecured to have your data in the cloud. And then we saw the CIA do it. And we saw Capital One. And we talked to the Capital One CIA. I always said it's more secure in the cloud. And then it really started to take off. Those takeoffs, maybe because SaaS is like the dishwasher and because cloud, it's got to be plugged in. And it meant that, yeah, it was growing, but it was sort of a 30 to 40, maybe a 50% growth rate. But what's amazing about AI is you just, at least with consumers or even business, you just open up the browser and it's there. And so that's why we're getting... [30:00] this straight up. And I think there's enough runway in the near term going from 10 bips of people really using it to two to five or whatever, which is going to cause it to keep on going straight up. This we call it a backwards L curve. So it's really pretty exciting. What have you learned about when the group that ends up being the leader separates itself from one of these competitive packs? So you're talking there mostly about overall growth of the S curve demand. There's always multiple players fighting for it. It seems like you kind of invest after someone has separated themselves from the pack, not try to pick the winners from the pack. You look for the S curve. Then we do an exhaustive study of everybody with exposure in that area and try and find the one with a very powerful competitive advantage. And a lot of people didn't like tech. Warren Buffett didn't like tech because he couldn't predict the future. Change too fast Yeah And so the S curve is our map for looking in the future Now a lot of people were worried about tech because they thought there was so much disruption You could never trust a company to be a long lived asset What we found over the years is some of the competitive advantages within the digital world are more powerful if not equally or more powerful than in the offline world You got the network effect that was so powerful for LinkedIn, Facebook, Alibaba, you name it. Then you can become an industry standard. Oracle and Bloomberg are the industry standard. Oracle charge a lot and there's free versions. There's open source Oracle, but they had all the database administrators. They had all the software that was tuned to work with them. They basically had a chokehold on the relational database market forever. You can get to scale very quickly because these S-curves grow and all of a sudden Anthropik is doing 30 billion in sales or Amazon had so much scale and they got it quickly. So they got a Walmart size scale advantage in five years versus 40 years for Walmart. So you can have network effects, scale, you can become industry standard. You can be a platform that people build on top of. You can have critical intellectual property, which was what Qualcomm had You couldn make a phone without paying them Or ASML has critical intellectual property You can make a chip without their lithography You can also have brand And brand very important because Google Amazon they got to grow They never had to advertise. Elon's never had to advertise for anything. And cost to acquire versus lifetime, it's the whole business model. Almost all the companies I mentioned have all of these rolled into one. Sometimes we can notice these things before the rest of the world. One of our high points was we pitched Amazon for AWS at 2013 at the Robinhood Investors Conference. And we said, the bulls have no idea what they're sitting on. Amazon's won the war before it even started. And at that time, we said there's Coke and there's no Pepsi. Did turn out there was Pepsi, but it was big enough to last. And we could see they had a seven-year lead. So first mover is important. Then they became a whole ecosystem and a platform. Then they got scale. so they were 10 times the size of everybody else. Nobody could invest in the R&D to catch them. But you're right that if you don't have a competitive advantage, you can be in the best S-curve of all time. And still lose out. But if your name was RIM, Pong, Nokia, HCC, LG, Motorola I can go on forever zero zero zero negative negative negative negative And that what we saw at the foundational model layer where there like 50 companies trying to do that And they all have fallen away And two or three have emerged at the top. There's a lot of reasons to think they will continue to hold their position. Google's a little trickier because they have this other huge, massive, complex business attached to the Gemini business. But if you take Anthropic and open AI as pure plays and you dig through those and you reason through their competitive advantages, why aren't they susceptible to erosion of those things? Of all the S-curves we've done, AI is by far the most complex and the fastest changing. We have to keep in mind that there are risks. The rewards are the highest because we're talking about a market in the trillions, maybe clouds 800 billion. And this might be, we now think three to five, but there's higher risk, higher reward. But let's just say with Anthropic now, it looks like they have critical intellectual property. Generally, they've been able to maintain their high market sharing code. Number two is they've built a strong brand for enterprise to where go talk to any CIO. The first thing they'll say is clawed. [35:00] Three, they're going to have escape velocity and scale. And what was scary for OpenAI and Anthropic fighting these big companies like Google was they had these huge cash cows. And to both of the management teams credited OpenAI and Anthropic, they were able to work in these super capital intensive industries and find ways to raise capital. and certainly with Anthropic, with their 10x sales growth and their fundraising ability, it looks like they've reached escape velocity. So now they have scale. And the other thing that Anthropic and OpenAI could have is Anthropic, now that they're leading in code, they set that code back onto their model. And it's this concept of the recursive improvement. And if you look at the pace of their innovation, it's accelerating. maybe they can have this liftoff stage open ai they were focused on so many different other sectors they're starting to do better in enterprise and their coding tools good and they're starting to see accelerating growth on that side and then look the consumer franchise it looks like enterprise right now is much better because you and I we willing to pay a lot because it replacing human beings Now consumer maybe you can get advertising but maybe they would pay for a claw bot type assistant if you could make that perfectly well for them But they have a gazillion eyeballs there. Things do shift. We have this charts that we almost do for all of our pitches. On the Internet, the leader goes bigger, faster and wins. most of the time the leader gets it. Shopify becomes the leader, it just keeps on going. Amazon, the leader keeps on going. SaaS company, XYZ, you get the lead, it compounds on itself. And another thing is you need to be big. Another is scale. You need the compute, and you got to pay for the compute. So there's only so many people that can do that. Those are some of the moats that we think are now showing up in this business. Now, there are some exceptions to that rule, usually with the paradigm shift, AOL, and then dial-up went to broadband and they didn't make the change. Netscape came out early and it wasn't as strong of a business model. And if you talk to anyone in the Valley or any startups, they'll tell you that they're building on top of these three and the world's a huge place and the economy is a huge place that they be able to differentiate within those I so curious then what you think all of this means for software When I look through your portfolio I don see a ton of big software companies enterprise software companies I don know if you once had them and sold them but it hard to have the experience of building really useful, cool little tools, even if they're still toys and not have the thought of, wow, if I spend enough time on this, even if I'm not technical, Maybe I could build a ERP equivalent replacement or something for my company. There doesn't seem to be a fundamental reason why that's not possible. And then those companies could be in lots of trouble. Seems like everyone has a strong view on this one way or the other. I'm curious how you've approached those sorts of companies, given that you don't seem to own a ton of them. Five years ago, we might have had 40 or 50 percent of our portfolio in software. And early on in our April 2023 webinar, we said definitely invest in chips first. But at the application layer, initially, we thought these companies are huge. They have huge sales forces. They can take these AI APIs and build products and they have the data. This is going to be amazing for software. Pretty quickly we realized their AI products were not very good They weren moving the needle Nobody could charge for them We basically sold almost all of our software Entering this year we were net short It really helped us in the first quarter. There's so many layers to this. I mean, the old way of software is like using a pen and paper, or it's like a horse and buggy. The new way of software is like a jet engine or, frankly, the transporter from Star Trek. It's so revolutionary changing that it feels like it has to be disruptive. If it's not disruptive now or right away, the software companies have another problem, which is their list on the to-do list or priority list of any CIO has fallen a lot. So even if AI is not going to be disruptive, they're spending it on Anthropic tokens because there's faster ROI there. Second, if they're spending all that money over there, it pushes on the budget. So that hurts them. Third, a lot of software companies were able to raise price every year. And now they're probably nervous about doing that. Then fourth, we'll see what happens with Jobs. [40:00] because there's smart people on both sides of that. But we are seeing some companies really gut their jobs. Freeze hiring or whatever. Freeze hiring. And so that hurts on seats. In terms of them building their own apps, if you want to be optimistic, it's taken them a while to do that. We talked about how early the primitives of AR. So maybe they have just taken a while to get something they can commercialize, but they might not have the right people. It's a different selling motion from selling a fixed system versus if you're installing something that does human work, you got to be right at the side to make sure it's really getting done. So you need the FDEs for deployed engineers. They might not have the right people internally to do that. Then, of course, there's the risk of you can build it yourself. The bulls will say, well, they're never going to build their own ERP system. And that's probably right. And it is true that old tech is very sticky. Mobile video games didn't hurt console games, and the tablet didn't hurt the PC, and the smartphone didn't hurt the PC. There's a lot of integrations and work that goes into these software That all true And companies do like to buy They don like to build themselves that much That all true But you can imagine a world where in one two three four five years you could have a brand new AI native company going after each one of these very strong incumbents. And if their data advantage could get obviated, it might be easy to take it out and put the new want in with AI. The valuations are very high and everybody knows they're under pressure. Some people are tempted to buy these, but the AI coding tools are just getting better and better. We'll have to wait and see. We're watching these software companies very closely to see if they're getting any revenue that can change that trajectory. But it's hard because if you're a company like Salesforce, you've got 40 billion in sales. You might have 500 of ARR, 700 of ARR of AI. So you've got this huge base. Now, maybe this starts to work, but it takes a while. In software, there's the rule of 40, which is your growth rate plus your operating margin. And if you have 20 growth rate 20 that good For AI we have a new rule of 40 What percent of your sales are AI 30 And what your market share in that category Say 30 You be 60 That a great place to look because you got exposure and you got a strong market position Problem with software is their AI is 1% or 2% at this stage, and it's a long way to go. One thing we are picking up though now lately, and this is half-baked, but AI could make some of these software platforms more important because what's the first thing you do with cloud, you plug it into Slack. If that can become a key repository, that will make Slack a permanent fixture within the organization. And so maybe the next wave of AI will be these agents that use tools, and they might operate inside of the existing incumbent software tools to use them like a human being would. Just to pull in that thread, it seems like the commonality of the tools they might use that are the most sticky would be network-based tools. So Slack is a great example of the software in Slack itself, the software is not the special part. It's that everyone is there. I'm curious what kinds of things you would want. Is it just the presence of a network effect? Is that the only thing that really matters Still early in our thinking here but even maybe Workday or the HR systems or the big systems of record the agents may be running on top of them It good and bad I mean CRM is going headless or they making a headless version And that the bare case, too, that you get relegated to just being a database. There's a human interface to it. Then they need to make the AI interface, which is no interface. It's just them going right into the data, you lose that customer interaction. But if the agents are going right to CRM and doing the work inside of CRM, that will solidify CRM. So you won't have to think it's going away. Can we talk about chips? You've referenced them a few times. Infrastructure chips, everything around the data center. Maybe I don't know how you conceive of it. Why is this so interesting to you? I love the modified rule of 44 percentage that's AI and percentage market share in the category that's interesting stat. What companies shine on that? What are laggards that are surprising? For the past 40 years, nothing has changed in the data center. Even with cloud, Intel x86 became the data center chip sometime in the 90s and compute grew in the cloud era. And compute workloads grow 25 to 40% every year. [45:00] Moore's Law is improving at that rate. It didn't require tremendous innovation. And there really was almost no growth in hardware for years and years and years. And the whole industry basically commoditized every part, every chip, every part of the server, the printed circuit board to the memory, to the enclosures, to the networking. there was no innovation. You would go from 1 gig to 10 gig. That would take seven years. And when you do switch in the first year, it does take some innovation to get to 10 gig and would create a little cycle, but then it would commoditize. Now you go to AI, the workloads are growing 10x every year, and they're pushing every single aspect of this hardware to the physical limits of what it can do. Not only are you creating tremendous unit growth, but we call it the decommoditization of the hardware industry. I met with Sean McGuire like three years ago, and he said I wish I could come back and be a hardware hedge fund because all the companies are public and they all have powerful IP And Sequoia made some of their best investments back in the hardware day with Apple and Cisco and others And we in this renaissance of chips. So not only do you have tremendous unit growth, it's requiring tremendous innovation. That means every aspect of the server, memory, which used to be a pure commodity, this high bandwidth memory is stack 10 chips on top, input outputs or 10x what they were before. Took Samsung for years to do it. And it's a critical piece. And then that is constantly upgrading. So they've got to be working with NVIDIA for three or four generations in advance. We had this with Celestica. Celestica was a contract manufacturer. This has been a disaster industry since 1999. It went all offshore to China. It was commodity, but they hung on. Celestica's heritage was IBM supercomputing, and they kept all that talent and skill. And then we noticed they were the sole supplier of the Google TPU server three years ago The stock was trading at eight times earnings And then they also had this whole business of selling Ethernet white box which is code word for commodity white box Ethernet switches into the clouds. It turns out these are excellent businesses. Not only do they have tremendous growth, but to do an AI server, it's liquid cooled. It's running so much hotter. it's two or three hundred thousand dollar piece of machinery whereas an old server was five thousand dollars if it breaks you just throw it away if this thing breaks the whole thing goes down so you become a commodity like supplier to like selling a critical part on a plane you'll never get swapped out it turned out they were quite good at liquid cooling a lot of other people tried to do it and fail and so they've retained that position then it also turned out that the ethernet market. In the old days, you would go from 100 gig to 400 to 800. It would be a seven-year cycle to upgrade. Now they're upgrading every year. That's really hard to do. Then there's a whole software layer the open Sonic layer The guys at Celestica invented were some of the people that wrote that open software They work very closely with Broadcom What we thought was just a great growth driver turned out to be great competitive advantages And they have like 50, 60 percent share of the cloud Ethernet switch market, which is a crucial market for AI because AI is incredibly network intensive. And then even something like the printed circuit board, a regular server, you need 10 layers. these AI servers, you need a 40 layer. And there's very few PCB suppliers that can make this. There's all kinds of complexities in there. Then we also own Elite Materials, which makes the leading ingredient, which is copper clad laminate, which goes into these boards. The PCB units are growing. The layer counts are rising. So you've got like a 50 to 60% CAGR just in the units. And And then the gross profits are rising and your visibility, which used to be, hey, we'll call you next week if we need you to like, hey, we need you for the next four years to be like designing this roadmap with us. You've gone from a 5% grow or low margin to a 35%, 40, 50 top line CAGR for the next four years with roadblocks. [50:00] margins. On top of that, there's shortages of everything. So even if it is a commodity, it's going to be a great cycle. So we see that up and down the supply chain. You find these companies like, I mean, Corning, they make the fiber. They've got some ridiculously high share of the fiber. I was reading this Microsoft data center they just built. There's enough fiber to circle the world four and a half times in that one thing. And their fiber is thinner and more bendable and can be specially manufactured to the exact specs. And it's higher margin, and it's the fastest growing part of their business. In networking, there's scale out, which is connecting all the server racks together. Then there's scale across, which is connecting the data centers together. And when you want to build one of these huge clusters and you can't get all the power in one place for training. You want to wire them together, but the wires that you need like 10 X, the wire has to be so much thicker. So that's creating huge growth And where the real kicker comes in is when you do scale up that connecting every GPU in the rack to the other ones that done over copper Eventually that be done over fiber When that happens that two to three Xs cornings opportunity So you just have at every layer of the rack. Everyone's overwhelmed. Everyone's overwhelmed, but in the power supplies, every NVIDIA chip or rack uses 50 to 125 percent more power That drives the ASPs of Delta and advanced energy. I can't believe these stories when I hear, I'm like, wait, so your ASPs are going to like go up 40% for the next four years in a row? And it's higher margin? the broader picture is the AI demand, if we're right with this L curve, we're already short the DRAM market, the NAND market, the PCB market. We're already 30% short all these things as we are now. The measure of percent AI, percent market share. Do you care more about the absolute or the rate of change of those metrics? We did this presentation in 2024 where we listed everybody's market share And then I asked Claude to plot it to a thing and it actually didn get it right because what it didn get is the rate of change So the rate of change is important And that incredible too because you go from 10 percent to 30 percent and your growth rate accelerates and your margins accelerate So rate of change is very important. Why don't more people get this right in public markets? If your whole framework is S-curve, competitive advantage, underappreciated earnings power, it feels like the movie's been played out a lot over the last 25, 30 years. My mom said, why do you tell everyone your secret? It's like, why does the casino teach people how to play blackjack? It's really hard to do. You have to be comfortable investing. I've been doing tech for 20 years at Whale Rock. We've got a team that's been doing this, covered many cycles. No one's paid attention to hardware and chips at all. So you've got all these newbies coming into it. You and Gavin, that's it. Yeah. And Gavin's done a great job. People weren't comfortable with it. It's harder to do than it seems. And a lot of these companies, their charts are up. So it's scary. Can I buy? And then you also have to have the holistic view, because if you don't have conviction every time with NVIDIA over the last four years, it's, oh they had a great year Oh my God it got to be a bubble And then they had another great year and it like six months of marking time It got to be a bubble This is like getting out of hand This is pretty scary The Bayer cases are not totally without merit, but if you can see the whole picture and understand how these things are unfolding and gain conviction in that, frankly, if you're just a semi-analyst, So many semi-analysts missed it because they didn't see what was really happening at the foundational model layer and how this broader picture. So it helps to have the big picture. It helps to have decades and scores of S-curves that you're looking at and where it plays in different things. What makes you the most concerned or uncertain? Is it just the rate at which all this stuff changes? And what keeps you worried amidst what seems like pretty extreme bullishness? One thing that bothers me is there's a lot of negativity in the general population about AI. And there's a lot of negativity in some aspects of the government. You know, I think Maine just banned data centers and only 20% of the people are optimistic about AI and potential for negative regulation. But I do think kind of the genie is out of the bottle. Another risk is that if AI [55:00] slows down in its improvements. I think there's a whole lot of AI adoption to happen, even if the models didn't improve. But Jensen said this years ago when he was talking about his graphics chips, if good enough is good enough, I won't have a business. Every year, he made the graphics a little bit better, and people always wanted the best. In AI, if Anthropix sort of hits a wall and stops improving our open AI, then the open source models will catch up. It might be a race to the bottom and it won't be good for the stocks, probably. It could be good for the chip companies. The chip companies don't care who wins. So that's another positive. And they'll benefit. Jensen really wants open source to take off. It's all he kept on mentioning at his last GTC. So that could be a risk. Another thing is if one or two of the players falters and loses its position and can't compete, that could be like a lot of compute that they don't need in the future. Now, if AI is so big, somebody else will suck that up. And we saw that with Oracle canceled a big deal and then Meta went right in But let just say Meta decided not to be involved with AI Hey we can keep up It just going to be a waste of our resources. We watch that very carefully. In general, we see more companies truly going after this, and even Microsoft trying to build their own. Those are some of the key risks. Seems like you really have done very little in the application layer of AI. Historically, the apps ended up being most of the market cap, not the infrastructure. And there wasn't really a model layer in the past. I guess you could say it was the clouds or something. Why focus so much on the bottom layers of Jensen's five-layer cake versus things in the application layer that are actually getting used by consumers? Well, A, it always comes later. So the first three or four years of the iPhone and then the applications really took time. So maybe it's just starting. To date, we found that area to be pretty risky because where does the foundational model end and where does the application begin? Can the applications build enough of a moat where they can fend off and build businesses in that? We thought we would see it in some of the incumbents of CRM and they starting in maybe just a matter of time but we really haven seen it in the enterprise world There are some very good startup application companies out there but the ecosystem not clear. When we started, the ecosystem in chips was clear. When we started, the foundational model ecosystem wasn't clear. Now it's clearer to us. And at the application layer, it's still kind of unclear and a little bit dangerous. But there will be great application companies built. We really were watching Brett Taylor at Sierra. Brett was CEO of CRM. He wrote Google Maps. He was CIO of Facebook. He's building this fantastic company called Sierra. We're not involved. But that's where the rubber hits the road. Will he be able to turn this into a huge company? And he's doing quite well. And we'll see. It's a matter of timing when these things really start to come into their own and prove they're sustainable. It usually doesn't start in the first three or four years. It comes a little bit later. At your office, you have this giant award wall for the best research job or project of the year given to an analyst. And I think you won it. You gave yourself a word in the early days when you were by yourself. But you got this now long 20 history of every year one or more people put their name on this wall for having done the best job on a research project that year I so curious about the nature of that research and how it's changing as a result of all of this. Say, you know, the person that's going to win the award this year and the sort of work that that requires a human to do when so much of the work that probably would have won you the award in 2009 or something could probably be fully automated or done in an hour with CloudCode or something today. How is the nature of research and what gets you on that WhaleRock Award wall changing in real time? I would like to say that we're so advanced in our AI systems that it's a huge change. But so far, it's helping us get up to speed and we have a handful of great apps, but it's not supplanting the job of the analyst. And so much of what we're doing is we're meeting with as many companies as humanly possible. We're developing relationships with the management teams that we cover. We're talking to the competitors. The system we use is right out of common stocks and uncommon profits, which was written by Philip Fisher in the 1950s. And it's the scuttlebutt approach. It's growth investing. It's get out there and talk to suppliers, customers, competitors, looking for the key characteristics of these [60:00] companies and really developing conviction in them. Now, if it's a new complicated area like ABF substrates or PCBs, we're able to get up to speed on those things quickly, but it can't pick stocks for you in any kind of a way. I will say that if you're an analyst who's good at the blocking and tackling, there's a role for that, but you need to have obviously the insight on top. So we're now like using AI to write notes or review the quarter. And those notes are much better, but there better be a really good paragraph on top, which is the wisdom. What does this mean? How does this deal with our thesis? What changed? Don't just be a reporter. So the AI can be a great reporter. It can't pick into the future. The job that the guys did on Apple 11 two years ago, I think we got two of the best ad tech guys. I knew ad tech. I started actually nearby here in New York at an internet advertising startup. And after I did banking, I knew internet advertising and ad tech, which is historically a terrible industry. But Michael and Sam really figured out the Apple oven story before anybody And they followed it when it was private They know all the competitors They know all the intricacies of terminology And Sam went to the Las Vegas app advertising conference, and we went to Khan, and we talked to scores and scores of people. So did the work on the model and developed a great relationship with Adam Ferrogi. He's one of the best managers out there. I don't see AI doing that. What role does talking to other investors outside of your firm play in your life? One of the great things is just the friendships I've built with so many smart investors. And frankly, Philip Fisher said part of his process was get to know a good 10 or 15 like-minded people around the country and share ideas. They're great friends to make. A lot of them have been on your podcasts. You develop good friendships and then you share ideas, talk ideas. It's important that it's a two-way street. I call it the tripod when I like something and then my analyst likes it and then somebody who I really respect also likes it. That three legs of the stool can really help the conviction What have you learned about shaping the products that you offer your investors across the history of the firm It not just one monolithic structure anymore There several things that if I'm an investor and I want to give you money, there's a couple of ways I can do that. How did you arrive at those things? And how can you turn that experience into advice for other investors that are trying to provide their LPs with the right set of options? So the first 15 years, it was a long, short fund, and we want to be focused. And if you defocus, that can be hard. So we grew that, and we got that to the scale that we wanted to. We're 20 years old. Maybe 10 years in, people started to ask for a long-only product. Sometime in maybe 2015, we formalized that we might be doing privates, and so we gave investors the option to opt in or opt out. And you could do 15% or 25%. So we didn't break the seal in the privates until 2020. We just think there's a huge structural underweight of the largest tech companies in the world. We also realized that a lot of our performance over the years was from some of the largest companies whether it be Apple or Amazon or Tesla And so a lot of our largest pools of capital endowments or what have you they realize they have been massively underweight the largest tech companies in the world because they have a lot of privates They don't have a ton of public. And then maybe half the public is international. And then of their public bucket, there's a belief that there's no alpha in large cap. So they underweight large cap and they have a lot of small and mid managers that are stock pickers because it's intuitive that large cap can't have alpha. And then in their hedge fund portfolio, even if it's long bias, they're not going to have 15 percent in NVIDIA and all these other things. We realize that people are worried that there's these big companies. This is just a product of the digital economy in that in tech, the leader usually grows bigger and wins and develops very high market share quickly. And there's great competitive advantages. and they're also selling around the globe. So this is gonna lead to massive profit pools and massive market caps. And it's just gonna happen in the future. Most endowments are betting against this because they're completely underweight this. I think there's tremendous alpha in the largest cap because if you think about it, a small cap, it just takes one person to figure out it's good and move it up. But it takes 100 people, 100 donors, [65:00] diversified PMs to realize Google's not a loser, it's a winner. And can we figure that out before 95% of those generalist PMs? And we've been able to do it. You like your odds in that. Yeah, we like your odds in that. And so there is alpha to be had there. And then as an asset category, it's great because these companies by definition have wonderful moats. And maybe they're not the super S-curve, but sometimes they are. I mean, NVIDIA sure is. And TSM is really levered to it. And Hynix is extremely levered to it. And ASML is levered to it. We're four months into that one. It sort of sounds like really what you've built is a research machine to understand the world through the lens of companies. The thing you're constantly trying to improve is that research machine. And then the way that you then express that through products is multiplied. But if I was to try to understand WhaleRock, it would be to investigate the research machine first and foremost. We call it the WhaleRock Learning Machine, and it's a group of 10 highly experienced individuals. Warren Buffett reads books, and we read books, and we read blogs. In tech you got to go out and talk to people So we do 2 3 face meetings with management teams Munger and Buffett talk about compounding knowledge We been compounding that knowledge for 20 years. There's changes to the team, but broadly, there's a lot of consistency to it. Andrew and Michael have been with me for 18 years, and the average experience level on the team is 10 or so years, and that includes some of the newer people. That research engine can support all these products. And it's the same people that do publics and the private. So we're not going to scour the world and turn over every A, B. But when we see something that fits into our system, we're able to act on it. It's so much fun to do this with you. When I do this, I ask the same traditional closing question of everybody. What is the kindest thing that anyone's ever done for you? It's definitely my father. I was super lucky. My father graduated Cornell, a double E electrical engineering, pivoted to Wall Street, and had a great career at Goldman Sachs. He ran corporate finance in the 80s and then ran private equity as chairman in the 90s. He was just whip smart but he had such humility and was such a great gentleman When I started Whale Rock friends and family he was the first call but he said you know I been at Goldman for 41 years How about I come and join you? I'll be the gray hair. I'll be the oversight. I'll be the chairman. You do what you do. You build the firm in Boston, build the team, run the money. I'll help raise some money. And we got to work together for six years until he passed away in 2011. But I just feel so lucky to have worked with him. It's not easy running a fund. We never raised our voice. And he was just an amazing mentor to so many people. And when he passed away, I got so many letters from people who said, your father was just such an influence on me. He was such a gentleman. He was such a great mentor to me. I just feel so lucky to have worked with him. And if I could be half the person that he is, I'd be completely winning. How did he do that? What was his method? Why do so many people say that? He was modest. He was whip smart. He was wise. He was also known as a great investor, which isn't the most common thing at a lot of investment banks. He also was on their commitments committee and kept them out of a lot of tougher situations And he was very warm and people could go into his office with problems and he handled it with grace whether it a personal problem or a work issue or what have you. And he just had this soft way and he also had a great sense of humor. I'm so lucky. Alex, thanks so much for your time. Thanks so much. If you enjoyed this episode, visit Colossus.com. You'll find every episode of this podcast, complete with hand edited transcripts. You can also subscribe to Colossus, our quarterly print, digital and private audio publication featuring in-depth profiles of the founders, investors and companies that we admire most. Learn more at Colossus.com slash subscribe. Thank you. [70:00] 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 data, understanding your process and producing real outputs. Check them out at rogo.ai slash invest.