▶ 原文链接

She Knows the 250 People Building AI. Here's What They Actually Believe.(中文)

来源: Invest Like the Best | Sarah Guo | Sep 01, 2026 播客: Invest Like the Best 分类: 投资 原文发表: 2026-09-01 纪要生成: 2026-09-02


全集重点


嘉宾/话题简介

Sarah Guo 是 Conviction 创始人兼管理合伙人,也是播客 No Priors 的主持人。她曾在 Greylock 担任 GP,23 岁入行,亲历多轮技术浪潮;本集她与主持人 Patrick O’Shaughnessy 深入讨论了她对 AI 前沿生态、早期投资方法论、计算供应链、开源模型、生物 AI 价值捕获等议题的最新判断。对话围绕一个核心问题展开:在一个被少数超级实验室主导的剧烈竞争时代,像 Conviction 这样的早期机构如何找到真正关键的人与方向。


分节详述

00:39 开场:当前的狂热与混沌感

本节重点

详细精要

💬 精华片段(中文)

“我一直说我想全力踩刹车,但我没有这么做——我正在以 90 英里每小时的速度前进。”

"I keep saying I want to press the brakes as hard as I can but I'm not doing it. I'm going 90 miles an hour."


02:13 核心赌注:少数实验室 vs 生态多元

本节重点

详细精要

💬 精华片段(中文)

“拥有第一、第二或第三个前沿模型的人将吞噬整个经济——我不想要那个未来,而且我并不认为我们会走向那里。”

"The owner of one, two, three frontier models consumes the economy. I do not want that future very clearly, and I don't think we are going to end up there."


04:49 如何在竞争激烈的环境中成为最佳投资者

本节重点

详细精要

💬 精华片段(中文)

“如果你押注于理解技术和社区,并从第一性原理出发,你可能会比别人有更好的接触和决策。”

"If you took the bet on understanding the technology and the community and approach it from first principles, you might have better access and make better decisions than others who are less focused."


08:29 投资决策框架:从技术第一性出发(Harvey 案例)

本节重点

详细精要

💬 精华片段(中文)

“法律是结构化语言。你需要阅读大量文档,我们有检索能力,你需要生成文本,而且有大量先例文本。这看起来真的是一个极好的匹配。”

"Law is structured language. You need to read a lot of documents, and we had retrieval, and you need to generate text. There's a lot of precedent text. That feels like a really good match."


12:09 前沿研究者现在关心什么

本节重点

详细精要

💬 精华片段(中文)

“信念是,随着 AI 研究模型能递归自我改进,我们距离某种指数级智能只有一两年。我认为这个信念对很多研究者来说是在过去 12 个月里新出现的。”

"The belief is, with recursive self-improvement of AI research models that can improve the models themselves, we are a year, two years away from some sort of exponential intelligence. I think that belief is new within the last 12 months for a lot of researchers."


15:47 最担心的障碍:计算、监管与供应链

本节重点

详细精要

💬 精华片段(中文)

“我认为这不是技术、能力或资本主义的问题,而是监管问题和对齐问题。我不是说 AI 对齐,我是说,如果你想在纽约建数据中心,你需要说服纽约人他们应该想要那里有数据中心。”

"I don't think it is a technology or capability or capitalism problem. I think it is a regulatory problem and an alignment problem. I don't mean AI alignment. If you want to build data centers in New York, you need to convince the people of New York they should want data centers there."


19:59 令人惊叹的研究者:Sunday Robotics 与 Tony/Chang

本节重点

详细精要

💬 精华片段(中文)

“从斯坦福地下室的一堆纸板,到我们建成全栈产品、在美国制造、完成数百次硬件迭代——这种速度令人难以置信。”

"From a bunch of cardboard in a Stanford basement to we built the full stack thing, it is manufactured here, we have done hundreds of iterations of hardware... that blows me away."


24:09 投资决策流程:直觉评分与备忘录

本节重点

详细精要

💬 精华片段(中文)

“有些人是不断思考、再思考,然后慢慢爬上坚信;我则差不多是从坚信出发,然后倒推工作。”

"Other people think about it and think about it and then they climb to conviction, versus I kind of start there and then I work backwards."


28:46 时间分配:组合公司优先

本节重点

详细精要

💬 精华片段(中文)

“我大概花三分之二的时间在组合公司事务上——招聘、帮助人们思考事情、试图影响外部生态。然后最大的下一块是看公司,每周大概看四到六家新公司。”

"I spend two-thirds of my time working on portfolio company stuff — recruiting, helping people think through things, trying to influence the outside ecosystem. Then the next largest piece is looking at companies, four to six new companies a week."


31:22 关于融资的经验与信念

本节重点

详细精要

💬 精华片段(中文)

“我建议创业者:你应该理解人们对你所做的事情的异议和问题,但你不应该告诉他们他们想听的话。”

"I advise entrepreneurs: you should understand people's objections to what you are doing and their questions, but you should not tell them what they want to hear."


34:27 父母创业的印记与家庭价值观

本节重点

详细精要

💬 精华片段(中文)

“对父母双方来说,这是一个道德问题:你永远不能担心别人怎么想。”

"For both my parents it was a moral issue. They were like, you cannot ever worry about what other people think."


38:08 灵感来源:反对愤世嫉俗

本节重点

详细精要

💬 精华片段(中文)

“如果你专注于价值、善待他人、与卓越的人共事,并且愿景是值得的,那么它成功的次数会比你想的更多。对‘玩营销和融资游戏’的犬儒主义,我讨厌这一点。”

"If you focus on value and treating people well and you work with extraordinary people and the vision is worthwhile, that works more times than you'd think. There's so much cynicism about playing the game, and I hate that."


40:15 开源模型之争

本节重点

详细精要

💬 精华片段(中文)

“如果你在美国限制开源模型的使用,你基本上只会限制守法的美国企业,拖慢它们,或者把利润转移到别处,而那些真正有对抗性用途的人并不受你的限制影响。”

"If you did restrict use of open-source models in the United States, you'd basically just restrict law-abiding American businesses and slow them down or move profits to different pockets. The actual attackers are not affected by your restrictions."


44:26 会不会没有 abundant intelligence?计算独立性

本节重点

详细精要

💬 精华片段(中文)

“我认为,如果没有自动化,就无法重建美国工业基础。如果我们不进口劳动力,而我们的劳动力很贵、又缺少一些技能,却想生产更多商品——这根本对不上。”

"I posit that there is not a version of the world where we rebuild our industrial base without automation in the United States. If we don't import people and our people are expensive... just doesn't add up."


48:36 计算独立性的短板与投资视角

本节重点

详细精要

💬 精华片段(中文)

“所有这些输入都存在全球供应链,其中一部分像是一个非常窄的筛子,位于一个不一定稳定、也不一定对美国及其盟友可及的地方。”

"There is a global supply chain. There are parts of that supply chain that are like a very thin sieve in a place that is not necessarily stable or accessible to the US and allies."


51:10 Conviction 内部最大辩论

本节重点

详细精要

💬 精华片段(中文)

“我们得出的结论,也是其他人得出的结论,就是今天的市场不同了:有对加速器的集中化规模化需求,甚至有对供应链独立性的需求。”

"We have arrived at the conclusion that others are as well, which is the market is different today. We are seeing consolidated at-scale demand for accelerators, or even supply chain independence."


55:53 生物学投资的翻转时刻

本节重点

详细精要

💬 精华片段(中文)

“这里没有天才。我说,这是一千万美元的合同。另外一部分是,你直接与客户端的科学家、使用工具的人交谈——客户知道它是否有价值。”

"There's no genius here. I'm like, that's a $10 million contract. The other piece is you just talk to the scientist at the customer... the customer knows if it's valuable or not."


58:28 为什么叫 Conviction:信念与风险

本节重点

详细精要

💬 精华片段(中文)

“如果真相被错误定价,而你能够坚持住,你就处于一个很好的位置。你想要信息不对称,以及当别人还没想通时坚持自己意见的信心。”

"If you find the truth and it is wrongly priced and you hold on to that, you're in a good position. You want asymmetric information and the confidence to hold the opinion when other people haven't come around to it yet."


01:00:33 信息优势与对抗噪音

本节重点

详细精要

💬 精华片段(中文)

“人们花了太多投资精力去思考哪一层会赢。我更愿意把精力放在弄清楚:如果我们才走了 1% 的路,那接下来的 99% 扩散是什么。”

"People spend so much of their investing energy thinking about what layer is going to win. I want to spend my energy figuring out, if we're 1% of the way in, what is the next 99% of diffusion?"


01:03:41 一年后的技术世界:杰文斯悖论

本节重点

详细精要

💬 精华片段(中文)

“在每个职能中,随着你学习得更快、做更少的琐碎事务,你会以某种方式重新分配时间。我工作更多。这是 Jensen 的核心智慧:我们都会更充分地就业。”

"In every function, as you learn faster and do less of the mundane, you will repurpose that time somehow. I work more. This is a core wisdom of Jensen's: we're all going to be more employed."


01:06:17 总结与世界上最善意的事

本节重点

详细精要

💬 精华片段(中文)

“硅谷有非常多非常成功的人,一旦跟你聊过天,就几乎不看你的出身。他们投资一段关系或一个人时,主导因素只是他们对想法和这个人的评估。”

"There are so many people who are extraordinarily accomplished in Silicon Valley who care very little for pedigree as soon as they have a conversation with you. The dominant factor in their willingness to invest in a relationship or a person is just their assessment of the idea and the person."


专业术语注释

术语 解释
Recursive self-improvement(递归自我改进) AI 研究模型能够改进自身模型的能力,本集中被认为是通向指数智能的可能路径
Exponential intelligence(指数智能) 指智能能力出现突然、指数级增长的阶段,多位研究者认为距离此阶段仅剩一两年
Great man theory(伟人理论) 嘉宾用来解释少数高行动力个体能改变重大结果的历史观
Monolithic outcome(单极结局) 少数前沿模型拥有者吞并整个经济、缺乏多元生态的未来景象
Pedigree(出身背景) 投资者用学历、机构背景、推荐来源等可读信号代替基本面判断,嘉宾认为这是危险的做法
Harvey 法律 AI 公司,Conviction 的早期投资,被用作“技术第一性出发”的案例
Sunday Robotics 机器人 AI 公司,由 Tony Zhao 和 Changi 等年轻博士生/前行业研究员创立,目标是半人形机器人进家庭
Compute independence(计算独立性) 类似能源独立的概念,指在计算供应链上不过度依赖单一国家或单一环节
Hyperscaler(超大规模云厂商) 拥有海量数据中心与算力的云服务巨头,其基础设施负责人认为 2030 年前没有足够规模可改变局面
PACIA Jacob Hellberg 推动的项目,旨在为计算供应链每一部分投资更多产能并寻找独立路径
TSMC(台积电) 全球领先芯片代工厂,本集中作为供应链“窄筛子”的典型代表
Jevons paradox(杰文斯悖论) 效率提高导致总消耗反而上升的现象;嘉宾希望 AI 在各领域也出现这一效应
Too cheap to meter 出自 Sam Altman 等对智能成本未来极低的描述,指智能将像电力一样普及且难以计量
No Priors Sarah Guo 主持的播客名,也暗含“旧有市场假设不再成立”的思维前提
Chai Discovery 生物 AI 公司,Conviction 的第一张支票投资,与前十大制药公司合作加速研发
Capex intensive(资本开支密集型) 需要大量前期固定资产投入的商业模式,如计算、能源、生物等方向
LP(有限合伙人) 风投基金的出资人,本集中多次讨论 LP 对 GP 的筹资、调查与早期押注
Jevons 与 Jensen 的智慧 指黄仁勋关于 AI 提升效率反而使人们更充分就业的观点
B10 和其他模型服务商 原文提到的服务大量推理工作的平台,代表开源模型经济中的集成方

延伸思考

原文发表:2026-09-01  ·  纪要生成:2026-09-02