风险投资市场风云变幻。人工智能蓬勃发展，而风险投资的其他领域却一片萧条。风险投资作为一种资产类别正在演变和缩减。价格更高的交易越来越少。新的市场类型正在出现。人工智能正以意想不到的方式影响市场。新技术正在进入成熟期。在本期节目中，我将一一介绍！

以下是我在节目中使用的幻灯片，供大家参考。

如果您愿意，可以通过嵌入式播客播放器收听本期节目。

除了上述 YouTube 视频和嵌入式播客播放器，您还可以在iTunes和Spotify 上收听播客。

文字稿

大家好。希望你们这一周过得愉快。两周前，我在圣特罗佩参加了 FJ 实验室的非现场活动。在这次活动中，我基本上准备了一份关于风险投资市场发展的最新报告，并意识到实际上存在着非常深刻的趋势和变化。

从泡沫和人工智能，以及整个风险投资之间的双城记，就像除了人工智能之外做得非常糟糕，这就像一个绝对的泡沫，到新技术的出现，坦率地说，风险投资公司和风险投资基金的性质发生了深刻的变化。当我把所有这些材料放在一起时，我想，好吧，需要分享一下，因为事情正在发生变化。

话不多说，我们开始吧。欢迎收听第 50 集：风险投资市场最新消息。

所以，我想做的第一件事，基本上就是向你们介绍一下风险投资和整个风险投资市场的情况。从宏观层面和全球层面来看，风险投资正在从低迷中复苏。这就是全球风险投资。

你看到的是第三季度到第四季度的大规模增长，以及第四季度到第一季度的大规模增长。但实际上，当你开始深入研究细节时，你会发现这确实是两个城市的故事。在欧洲、拉丁美洲、东南亚和整个亚洲，我们都没有看到投资的增长。

所有融资都在美国，都处于后期阶段，而且基本上都是人工智能。在全球范围内，人工智能占第一季度融资的 53%，在北美占 71%。因此，人工智能完全主导了投资和所有其他类别的投资。现在很难获得融资。而且，如果你看看现在的情况，基本上，融资的公司越来越少，价格越来越高，对吧？

这就是中位数。顺便说一句，中位数更高，因为非凡的交易、疯狂的交易让一切都变得更美好。但种子期交易的中位数是 18，这是前期。A 系列的中位数是 55，B 系列的中位数是 131。但疯狂的交易就像平均值右侧的两个标准差，而且在某种程度上比以往任何时候都多，主要是在人工智能领域，主要是重复创始人或曾在开放人工智能公司工作过的人，等等。

人工智能正在吞噬市场。因此，如果你是一家人工智能公司，在各个垂直领域的估值溢价都是如此，仅仅是人工智能领域的溢价就从 69% 上升到 314%。因此，从这个角度来看，人工智能公司正在蚕食整个世界。现在，最酷的是，FJ 实验室本身实际上一直保持着严谨的态度，我们在人工智能方面一直是逆向思维。

实际上，我在之前的 “玩转独角兽 “一集中已经介绍过我们的人工智能论文，这里就不再赘述了。但一句话总结就是，我们不是在核心法律硕士上竞争，而是以合理的价格投资于人工智能的应用。但是，如果你看一下我们的历史，即使是在21世纪，我们仍然是有纪律的，或者说是中位数，这显然只是美国的情况，因为如果我把其他国家包括进来，结果就会有偏差。但我们的估值中值一直保持在 16 左右，A 系列的种子估值在 40 左右。而我们的 B 轮，[甚至是 20 和，而在 80 年代是数百，我的意思是，样本量较小，所以更容易移动，但我们仍然远远低于美国媒体，部分原因也是因为我们没有把重点放在超级热门的人工智能创意上。

现在，退出市场已经恢复了一些，有更多的 IPO，当然，最近我们看到了 Chime 和 Coinbase 的 IPO，但在大多数情况下，相对于 2020 年和 2021 年的繁荣和大量退出，市场仍然更加接近。A 市场、IPO 市场开放，但在地缘政治紧张局势、关税等因素的影响下，这一切尚未真正发生。

现在，我希望政治动荡能够平息下来，因为他们试图在明年的中期选举中获得连任，也许就不会再做那么多破坏性的事情了。所以，TBD，如果现在发生这种情况，我希望 26 号公路的情况会有所改善。但要知道，TBD 是市场的外生因素，而不是内生因素。因此，我们是市场的接受者，而不是市场的制定者。

因此，尽管如此，情况还是略有好转，我希望随着利率略有下降的环境和流动性条件的改善，情况会继续好转，因为我相信他们会推动财政扩张，也许会减少一点干扰。这只是希望，而非现实，但还需等待，希望能朝着正确的方向发展。

现在更有趣的是，风险投资作为一种资产类别正在经历根本性的转变，我想说这是一种分裂，对吗？排名前 10 的公司基本上筹集了所有资本的 50%，而排名前 30 的公司筹集了所有资本的 75%。

因此，我们感觉市场上存在着这样一种情况：一方面，安德森-霍洛维茨（Andreessen Horowitz）、Thrive、General Catalyst 等巨额资本积累者积累了数十亿美元。我甚至不确定是否还能称他们为风险投资人。他们是整个生命周期的投资者，他们投资的公司，在公司上市后仍然是活跃的投资者，因此我确信，他们将获得的内部收益率更像是 10%/15%，而不是人们在风险投资中一直追求的 20%/25%/30%。另一方面，那些成立时间不到三个月的新兴管理公司获得的资金越来越少，所有这些公司加起来获得的资金仅占基金的 23%，创下了 10 年来的新低。

因此，我们看到越来越多的超大型基金规模超过 10 亿美元，并获得越来越多的资金。此外，还有一些小基金。因此，你需要超级专业化和小型化，或者你需要大型化。中间的每个人都做得不好。因此，我们看到新基金大量减少，风险基金率下降了48%，新基金在三年内下降了68%。

正如我所说，九家公司筹集到了全部风险资本的 50%。自过去三年的高峰期以来，基本上有 2000 家风险投资公司倒闭。我们还看到许多 GP 离开、退休等等。如此大规模的整合，风险投资作为一种资产类别正在大举缩减。

原因当然是内部收益率一直在下降。现在，内部收益率一直在下降的部分原因是，人们在 2020-2021 年，也许是 22 年初进入市场时，价格太高了。在这种估值下，公司并没有增长。因此，回报率在下降，尤其是没有退出市场，对吧，市场，并购市场，因为反垄断在 22/23/24/25 年已经关闭。首次公开募股（IPO）也不多。因此，DPI 非常非常低。因此，LPs，也就是基金的投资者一直没有收到分红。因此，他们没有资本进行再投资。结果，你知道，你没有得到投资私有企业的好处。

过去，当你投资风险资本时，你会获得比公开市场更高的溢价。但在过去 20 年里，纳斯达克的表现其实更好，这显然主要是受到过去几年 “华丽七侠 “和人工智能公司的推动。但你并没有获得流动性溢价。

当然，这是作为平均水平班级的风险投资，而前四分之一班级的情况要好得多。但是，即使前四分之一的学生成绩好很多，他们也要花很长时间。这需要很长时间。我们看到的是，最好的公司在私营企业中的时间更长。因此，从大趋势的角度来看，真正发生的事情是公司说私有化就私有化18 20多年，对吗？

我是创始人基金二期的投资者，该基金投资了 SpaceX。我记得是在 2007 年 12 月，估值大概在 8000 万左右。这已经是 18 年前的事了，很明显 SpaceX 仍然是私人公司，复利的表现令人惊叹，但风险投资基金在市场上是作为 10 年期的投资工具，外加一年和一年的延期。

所以也许是 12 年期基金。是的，现在我们的基金存续期已经到了 10-12 年，但大部分价值仍未分配。所以退出的时间比以前长了很多。因此，前四分位数的基金仍然表现出色，但分配的时间要长得多，最好的公司也要在私募基金中待得更久。

因此，很多LPs认为，我们还没有分销，我们有流动性，我们对这一类别的投资过高，我们正在缩减风险投资这一资产类别。我认为现在是投资风险投资的最佳时机，就像2000年末、2008年、2009年、2010年大衰退期间一样，是投资风险投资这一资产类别的最佳时机。

因为这些公司正在创建，以较低的估值融资，所以竞争较少，所以当它们获胜时，它们就赢得了整个类别，而这正是 Airbnb、Uber、Twilio、WhatsApp、Instagram 和 Slack 的时代。我猜想，在人工智能之外，现在正在融资的非人工智能公司也将成为 2020 年代的决定性公司，因为同样的事情，它们的融资估值较低，竞争较少，而那些赚到钱的人都是为了赢得比赛。

因此，我认为，而且坦率地说，正在发生的事情是，二级市场将成为风险投资基金游戏规则中更大的一部分。到目前为止，73% 的基金从未参与过二次融资或通过二次融资进行销售。因此，他们要么等待企业上市，要么等待企业被收购。仅此而已。

但随着流动性的枯竭，以及这些事情需要越来越长的时间，我认为越来越多的人将会去探索拥有、做二级市场。现在，你有多种方式可以做二级市场。你可以在像 Forge 这样有经纪人的平台上买入或卖出。

问题是，这实际上仅限于 SpaceX、Stripe 和 OpenAI 等顶级风投基金。还有 Anthropic 和 Databricks。这些公司有流动性，但下面的公司其实都没有流动性，而这些有流动性的公司，我甚至不确定你是否想要流动性。对于下面这些公司来说，第二轮融资更有可能发生在正在进行的一轮融资前后。

所以我们现在看到的是 三种类型的第二轮融资的演变。其一，人们在这些平台上进行一轮又一轮的二级市场买卖。其二，有一些基金正在创建，以购买 LP 头寸。因此，我认为现在一个非常有趣的资产类别，就是购买一个已有8、9、10年历史的基金。

所以更接近流动性。你很了解相关资产。你知道或预期未来几年会有流动性，就会购买它们。如果你很了解这些资产的名称，你可能会以资产净值 50% 的折扣价买入 LP 仓位，你知道，这会带来很好的回报。

您还可以以大幅折让的资产净值投资于 GP，因为 GP 急需流动资金，如果只是为了为未来创建的基金调用资金的话。因此，二级市场的时代正在到来。当然，不是每个人都能做二级市场，对吧？如果你是一家风险基金，你是董事会成员，你持有公司 25% 的股份，如果你在二级市场卖出，很可能会毁了公司，因为任何潜在投资者都会说，等等，他们知道什么，我们不知道什么？

因此，他们不愿意投资。但像 FJ 实验室这样的基金，我们之所以做得非常好，部分原因是我们只持有公司几个百分点的股份。因为我们只持有公司几个百分点的股份，所以当我们出售股份时，这并不是一个信号，出售股份只是我们策略的一部分，因为我们是真正的风险投资者，所以我们会回收资金，继续投资新的项目。

如果只剩下 3 个 X，它可能是一家很棒的公司，业绩也很好，但它可能更符合 P. 回报的要求，而不像我们的早期风险投资要求。因此，对我们来说，卖掉它并尽早进行再投资会更好。而且，顺便说一句，创始人往往会请求我们卖掉公司。

他们会说，嘿，格雷洛克、红杉、安德森，他们都想加入，但他们都想要 15%。我不想被稀释 45%，但我希望他们三家都加入，因为我不想让他们资助竞争对手。所以，请你知道，我们为什么不占 25% 或 30% 的股份，你是否介意出售你的部分股份？顺便说一下，我们在这里采取的策略是，我们通常会出售大约 50% 的股份。

因此，当我们认为它们的价格已经完全确定时，我们就会采取反风险投资的策略。如果它上了月球，我们仍有 50% 的权利。太不可思议了如果跌到零，我们卖出的 50% 也能赚 10 倍，我们也很开心。这样也行得通。

无悔哲学。这种情况会越来越普遍。在风险投资领域，很多事情都很容易发生。我们有巨型基金，它们是资产积累者。我们有一些小型基金，它们需要像 FJ 一样，在网络效应和轻资产业务方面美丽而专业。但也有很多专门的小型基金，几亿美元就能做得很好，或者像 Benchmark 这样的基金，会继续做得很好。

LP 正在发生变化。显然，大学考官都是大型 LP。但当然，由于他们现在可能要纳税，我们也不再从联邦政府获得捐赠，这个应用程序的 LP 基础可能正在消失，或至少急剧萎缩。他们甚至在考虑出售部分仓位。

许多公司正在考虑上市，因为它们正在成为资产积累者。因此，通用催化剂公司正在探索首次公开募股（IPO），有传言称 Andreessen 也可能这样做。从某种程度上说，这是在步黑石、KKR、阿波罗等私募股权投资公司的后尘。

现在你看到的也是永久持有工具。像红杉和茁壮这样的公司，只是永远持有资产，对吗？所以就不存在需要创建新基金的想法了。所以总的来说，风险投资正在发生变化。很多基金正在消亡

这就像资产类别的彻底内爆，大型资产积累者之间有明显的细分，这些品牌很有可能成为永久持有工具和上市的组合。一般的催化剂会茁壮成长，世界也会因此而变得更加美丽，几亿美元的小型专业基金也会影响到慢箱集团–FJ、Benchmark，这些都会做得很好，而中间的每个人可能都会死掉，并因此而遭受损失。风险投资正在发生深刻的变化。

接下来我们要介绍的是我们在第一季度进行的一些很酷的投资。当然，这些投资凸显了一些有趣的趋势，尤其是在市场方面。我在一月份已经更新了市场趋势，谈到了 B2B 市场上发生的很多事情，比如像 Palm Street 这样的直播视频购物公司，它在销售稀有植物，但现在我要介绍的是今年上半年的投资，这些投资突出了市场上出现的其他趋势。

因此，我们投资了一家名为 Garage 的公司。车库是一个消防车市场。我们说的是 30K AOV。有趣的是，在车库存在之前，创始人就已经很了不起了。产品也很出色。基本上消防队会把消防车挂在Facebook市场上

他们的流动性非常低，你知道，他们关闭起来会很复杂。因此，他们解锁市场的方式是，他们说，你知道吗，我们要打造端到端的无缝体验。我们来管理运输，我们来管理支付，我们来处理一切。

顺便说一下，这些卡车的平均运输距离是 1200 英里。所以它们被放在平板车上，然后开走。这就是所有的复杂性。因此，在 “市场平台 “之上添加一个服务层，往往能解锁一个以前不存在的类别。因此，消防车在市场上并不存在。

这是我们以前见过的。我在上次介绍市场趋势的播客中提到，Tetra 是一个热泵安装市场，基本上他们解决了更换热泵的所有麻烦，但你不需要找一个总承包商，你只需展示你的设备，他们就会为你找人来安装。或者 Alpagga，这是一个餐饮设备市场，提供安装和运输服务，再次释放了市场的价值。

因此，从根本上说，在市场平台中整合服务层正在释放出新的垂直市场，车库就是一个最新的例子。下一个有趣的例子是，P2P 租赁终于成为或即将成为主流。因此，当 Rent the Runway 创立之初，他们实际上试图成为一个点对点的租赁市场。

当然，从某种程度上说，这样利润会更高，工作量也会更少。但相反的物流方式并不奏效。人们彼此不信任所以他们需要库存他们需要一个仓库。他们需要干洗当然，这是一个资本效率更低的行业，或者说是资本密集型和资本效率更低的行业，利润率要低得多。

现在，像泡菜这样的公司已经完全破解了这一难题。人们已经习惯了这样做。也许这只是因为他们现在已经习惯了，你知道，坐 Ubers 和在 Airbnb 上看别人的房子。因此，人们的行为模式发生了改变，并乐于接受。但很明显，他们在互相租房，而且做得非常好。NPS非常高，你可以得到更好的价格，更好的库存，每个人都对市场的发展非常非常满意。

顺便说一句，这已经是我第 50 次看到这种尝试了。法国有家叫Z Lock的公司旧金山也有一家做电动工具租赁的点对点公司，但都没成功。但在这里，有反复，有INPS，人们喜欢它。

我上次也提到了 B2B 再商务。很明显，再商业在消费领域很重要，但它最终还是走到了前台。我上次提到过一家名为 Ghost 的公司，该公司出售时尚界的过剩库存。历史上，这些库存都是卖给像 21 世纪这样的大型直销店。而通过 Ghost 公司，他们第一次能够以较小的单位进行销售，这样零售店就能以 90% 的折扣购买多余的服装，而很多消费者也能从中受益。

Rebel 是另一个例子，它做的是婴儿用品，并在该类别中创建了一个市场。所以说，B2B 再电商并没有出现，而且做得很有规模，效果也很好。

最后一个大趋势，算是一个大趋势。这是旧的搜索基金的一种进化，现在你会看到很多类似小规模的收购，人们收购这些老牌公司，然后引入技术层或人工智能，使它们变得更有效率。这种情况发生在每一个主要的垂直线下业务中。人们收购洗衣店、理发店、美甲店。

你能想到的每一个主要类别都能将其清晰化，提高运营效率，获得 P.O.S.。在风险投资领域，慢速创业一直是这一战略的主要支持者，而人工智能能让企业提高效率、降低成本。

这绝对是一个刚刚开始的大趋势。

接下来，我想谈谈人工智能市场的影响。正如我所说，我们并没有直接投资于法律硕士公司，但我们所做的一切都与人工智能有关。当人们在人工智能市场公司与我交谈时，他们最关心的问题是，人工智能会颠覆我们吗？

还是说，这些发现将被法律硕士们所掌握，我们将不复存在，我们将成为完全商品化的纯粹执行机器？我想了很久，最终决定事实并非如此，对吗？如果你想一想人们是如何购物的，那么市场上有三种行为。

首先，购物是一种娱乐。你不知道自己在找什么，你要找的是时尚物品。这就是人们在分类网站或 Vinted 上购物的方式。他们基本上都会去那里浏览列表。他们可能根本不需要买什么东西，但因为商品的价格是 30 欧元、30 美元或 40 美元，他们看到喜欢的东西，就会买下来。

法学硕士在这里的作用微乎其微，因为没有任何意义。我们不是为了提高效率。我们只是想看到很酷的东西。我认为，就像去开放人工智能公司告诉他们，嘿，给我创建一个符合我需求的时尚购物界面，这对开放人工智能公司、Anthropic公司或Grok公司来说，甚至都不在1000个优先事项之列。不，绝对不是。

顺便说一句，即使这种漏斗被 LLMs 占据，因为这些都是赢家通吃，并不是后端市场的赢家通吃，大部分价值仍将保留在实际市场中。我既不担心它会扰乱漏斗的顶端，也不担心它会压缩利润空间。

人们在这些市场上的另一种行为是，他们很清楚自己在找什么，所以他们会在搜索结果中输入 “LG C4 65 英寸 OLED”。因此，当人们去亚马逊时，他们通常会输入，你知道，LG C4 65 英寸 OLED，然后他们就会得到那一件商品。实际上，eBay 也是一个搜索市场。

现在，在这里，也是一样。如果您清楚地知道自己在寻找什么，那么法律硕士是否有真正深刻改变这种状况的空间或余地？其实没有。现在，你能改进搜索结果吗？也许可以。你能提出相关项目吗？也许可以。而现实情况是，从某种程度上说，这些搜索也是在谷歌上进行的。

但有趣的是，如果你在谷歌上输入这些项目，比如电子产品，几乎所有的结果都是eBay和亚马逊，因为它们在这两个类别中占有如此大的市场份额。亚马逊和 eBay 在美国电子商务领域的市场份额合计达到 43%。而在某些垂直领域，我们说的是 90% 以上。

那么，我是否可以看到这样一个世界：法律硕士取代谷歌，并在获取漏斗顶部方面获得与谷歌搜索相同的价值？可以。他们能获取更多价值吗？不会。因为同样的，履行、库存管理、运输、保险退货，所有这些都将完成，你将继续拥有你的核心价值。

每个主要的垂直行业都是如此。因此，我并不担心像 LLM 这样的市场会遭到破坏。那些由搜索驱动的市场。事实上，我可以想象，你可以利用它们，利用 LLM 来帮助人们发现产品。因此，我认为，事实上，我稍后会在下一页谈谈我将如何实施，以及我将如何使用 LLMs。

第三个类别现在更开放供讨论。因此，有些类别你并不清楚自己在寻找什么。这是一种经过深思熟虑的购买。你要买车，要买公寓，或者要买高端的滑雪装备，你对这些类别并不了解。因此，通常情况下，你过去会做大量的研究。

你已经上网、与汽车经销商交谈过、试驾过汽车等等。因此，法律硕士绝对可以帮助您反复推敲，并思考出最适合您的方案。事实上，在我考虑从特克斯和凯科斯群岛搬到安提瓜岛或尼维斯岛时，我正是出于这个目的使用了 ChatGPT。

它帮助我思考，首先，这些都是正确的目的地，找到合适的房产，找到经纪人，了解估值等。因此，它是一个了不起的工具。我认为这也是像 curated dot com 这样的公司的部分原因。

实际上，相对于他们的融资额，我卖出的估值相当低，因为我认为存在风险，担心他们会被 LLM 打乱。但即使是在这里，我也不认为这些经过深思熟虑的购买行为中，GPT 就能胜过网站本身的 LLM，对吗？所以，试想一下，Carvana 可能会拥有最好的汽车建议引擎，因为他们拥有所有的数据，能够准确了解人们在寻找什么样的二手车，他们可能会更好地为你提供建议，告诉你买什么车，然后 GPT，也许，也许不是 TBD。

但即便如此，即使是 ChatGPT 进行了推荐，最终的交易还是发生在 Carvana，他们仍然可以获取很多价值。例如，我一直在使用亚马逊的人工智能 Rufus 来征集产品创意，它真的非常非常棒。因此，我可以想象，大多数市场平台要做的事情之一就是增加一个层，如果你想要推荐，它就会出现。

事实上，从长远来看，拥有两个搜索引擎（一个用于搜索结果，另一个用于产品对话）的想法是没有意义的。我认为这将是一个框，而这个框将是灵活的。你可以说出你要找的东西，也可以征求意见，比如长问题，你会得到答案，而搜索引擎会很好地知道哪一个才是答案。

因此，人们一直在问我，好吧，如果我是一个市场，我是否应该在法律硕士中为自己建立索引？我的回答是，如果你在谷歌中索引自己，那么答案就是肯定的。这意味着，鉴于 99% 的人都在谷歌索引自己，答案绝对是肯定的。

你应该做 AEO，也就是他们所说的法学硕士，引擎优化。所以你要回答引擎优化。你绝对应该在那里建立自己的索引，因为目前这是一种免费流量，而且因为很多人都没有这样做，也没有做得很好，所以你将在获取更大份额的免费流量方面占据优势。

现在的情况是，我可以想象你不想这样做。想象一下，你在你的品类中占主导地位。你拥有 99% 的市场份额，当有人需要这个类别的东西时，他们会直接来找你。他们不会去谷歌。这样的话，我就有理由不在谷歌中索引自己，从而不在法律硕士中索引自己。

但是，如果你要在谷歌中建立自己的索引，那就在法律硕士中建立自己的索引吧。因此，换句话说，LLMs 对市场并不构成威胁，但对谷歌的生存却构成了深远的威胁。但我在个人使用中发现，我不再使用谷歌，无论搜索什么，我都会直接询问 ChatGPT。

很显然，我就是其中之一，但有一个答案总比有十亿个答案要好得多，而且还能不断迭代，不断思考，这就是为什么对谷歌来说，像 “双子座 “这样的产品是存在的，但同时也是对其商业模式的存在威胁，对吗？现在，LLM 没有广告，它只是一个特定的产品。

因此，它将谷歌的核心和基本业务模式放在了一起。因此，谷歌可能在人工智能领域占有很大的份额，但如果我是他们，我会把所有精力都放在人工智能上，因为 OpenAI 很有可能被 OpenAI 打乱，而这正是让我彻夜难眠的原因。

与 Meta、苹果、亚马逊、eBay 相比，我认为你可以利用它做一些有趣的事情，但基本上，它不会对你的生存构成威胁。因此，与其担心法学硕士会扰乱你的顶级漏斗，我认为这不会发生，不如这样做。

首先，人们已经真正开始使用人工智能。这是有史以来第一次进行适当的跨境交易。过去，当你在欧洲时，你会为法国推出一个网站。然后分别在德国建立一个完全不同的网站，提供不同的商品，不同的流动性。然后在波兰建立一个不同的网站，在英国建立一个不同的网站。

但现在，有了人工智能，Vinted 首次创建了一个真正的泛欧市场，坦率地说，甚至是全球市场。房源信息自动翻译。买家和卖家之间的对话也会自动翻译。现在，它只在有整合运输和整合支付的情况下工作。

但是，有史以来第一次，通过自动翻译对话、自动翻译列表以及集成的运输和支付，您可以拥有一个真正的全球市场和真正的泛欧市场。因此，很显然，您需要确保这一点，但如果您确保了这一点，您就可以创建。

因此，跨境是一个大趋势。Ovoko 是欧洲一家出色的汽车零部件市场，它在东欧采购汽车零部件，然后主要在法国销售。即使是在 B2B 领域，我们看到 CarOnSale 现在也有 30% 的汽车是通过 B2B 跨境销售的。因此，跨境销售在不断增长，而这主要得益于人工智能的自动翻译和翻译列表。

第二，如果你是市场营销人员，你可以使用人工智能来大幅简化你的列表。以前的列表流程是这样的。你要拍一大堆照片，然后写描述，写标题，选类别，选价格。所有这些都是大量的工作，而且坦率地说，你并不一定最了解这件商品的正确价格。

现在有了人工智能，如果你有足够的数据，你可以拍几张商品的照片，然后人工智能就会说，哦，这是商品，这是类别，这是标题，这是价格，然后 “砰 “的一声，你就完成了。我们的许多市场平台都做到了这一点。我们是 Rebag 的投资者，这是一个手袋市场。他们有一个叫克莱尔的人工智能，可以进行图像识别，让你出售你的手提包。

我们加入了一家名为 Collx 的公司，该公司为交易卡提供这种服务。你只要拍下交易卡的照片，它就能立即识别出价值最高的交易卡，并帮助你出售。第三，提高列表质量。

我们的投资者是一家名为 PhotoRoom 的公司。PhotoRoom可以帮你拍摄商品照片，识别商品的类型和最有可能销售的地方，然后我们会自动更改背景，以提高销售率。因此，在某些情况下，它只是创建白色背景，但在某些情况下，把它放在利基或其他东西前面。

因此，使用人工智能不仅能简化上市流程，还能改善上市情况，这又是一个巨大的趋势。最后但并非最不重要的一点是，每个人都应该使用人工智能来提高内部生产力。因此，现在每个人、所有程序员都在使用 Cursor、GitHub、Copilot，提高程序员的工作效率。

很多公司都在使用人工智能客服来帮助处理退货、回答最基本的问题。坦率地说，这种方式比与真正的代理商交谈更便宜、更快、更好。人们还将其用于销售和营销。其他方面，正如我所说的，我会用它来做一件事，那就是我可能会在每个市场上添加一个商品推荐引擎。

因此，除了核心搜索引擎之外，我还会问 “嘿，什么是我应该购买的最佳商品？”因此，我可能会把这一点添加到我没有放在幻灯片中的 E 中，但它非常有意义。正如我所说的，我会将自己整合起来，并在法律硕士中为自己建立索引。

我们看到的其他很酷的趋势与市场平台无关，但这些趋势最终都会到来。其中之一就是仿人机器人。这是我们投资的一家公司，名叫 Figure。很多人问我，等一下，既然你可以创造专门的机器人，为什么还要人形机器人？

那么，已经投资了数万亿的基础设施。让人类在正确的仓库里工作等等所以你可以为非人类机器人 重建所有这些基础设施 或者你可以直接制造人形机器人图公司创造出了令人惊叹的人工智能和手部灵活性惊人的机器人，它实际上可以取代人类，比如，你知道的，最后一英里的分拣和包装仓库，它们被部署在南卡罗来纳州的宝马生产线上，取代了昂贵得多的机械师的工作，因为机器人每天工作20小时。

因此，我们正处在这场革命的前夜。现在才刚刚开始 但我想他们已经有订单了也许只是一份意向书，我想他们提到了十万个机器人。所以我可以预见在不远的将来 大概四五年后 数以百万计的机器人就会出现在工作中

特别是在像最后一英里，像最后一英里分拣，包装仓库，和在，坦率地说，开始成为在家里。现在它们还是太贵了它们的功能还不完善，无法进入家庭。我认为我们将开始看到第一个版本。这是图二，当图三出现时，也许一年后，我没有确切的时间表，但我猜测，根据这些事情发生的速度。

当图 4 再次面世时，可能会更常见，这只是猜测。我没有这方面的信息，但我觉得三四年后会有。没错所以我能看到五年后的世界，同样，它将遵循典型的形式。最初的几款效果不会很好 会非常昂贵 会被富裕家庭采用 随着成本的降低 最终会变成杰森一家

每个人都将拥有一台，这将大大减少通货紧缩，极大地提高你的生活质量，让你有空闲时间做所有你喜欢做的事情。现在，如果你喜欢做饭，那就做饭，但如果你不喜欢做饭，你可以为自己做饭，去买菜，这将解放人类，让他们能够做自己喜欢的事情。

我们正处于这一趋势的前夜。像Figure这样的公司，顺便说一句，Figure并不是这里唯一的一家。我们有很多中国公司，还有特斯拉和擎天柱，它们都在这场竞赛中，它们将改变这场竞赛。但我们只是想说，机器人即将到来。

虽然为时尚早，但这是一个大趋势。

接下来是加密货币或稳定币的真正用例或杀手级应用。当然，现在加密货币中最大的资产是比特币，它的用例是数字黄金。很明显，如果你生活在新兴市场，你没有一个惊人的储蓄过程，你生活在通货膨胀的政权或没有没收你资产倾向的政权，你知道，想想委内瑞拉，我在想津巴布韦，我在想阿根廷。

用比特币储蓄是很有意义的。事实上，比特币内置的通货紧缩是有道理的，但实际上稳定的比特币才是真正的杀手级应用。在这些国家，你需要一种交换手段 比特币不是一种好的支付方式，因为它会通货紧缩而通过实际的稳定币，你可以用美元或其他货币，但大多数情况下，USDC、USDT都是美元，而且令人惊讶的是，不仅可以储蓄，还可以交易、支付等。

同样，主要的使用案例一直都是新兴市场，但它们正在走向成熟。现在，随着稳定币的推出，随着这个圈子的上市，我可以看到，即使在西方国家，美元兑稳定币作为一种支付手段也变得更容易被接受。现在你可以看到，稳定币的交易量已经达到并超过了 Visa 的交易量。

但稳定币的交易量将超过维萨卡、万事达卡和美国卡。所以稳定币的出现才是大趋势事实上，我自己也创建了一家稳定币公司 叫做迈达斯 它能让你的稳定币产生收益 而且它做得很好

我们已经从零发展到 1.5 亿 TVL，但没有市场营销、没有 KOL、没有积分，只是因为我们拥有令人惊叹的用户界面和出色的产品市场契合度，而且实际上是以完全受监管、合法破产的远程方式进行的。但我们还处于稳定币革命的初期。

未来几年，稳定币将会更加普遍。还有一件事可能值得一提，那就是在乌克兰，战争的未来真的是无人机，看到像Endurall这样的公司，在美国得到了很多媒体的报道，他们做得很好，我们也是投资者，但我认为更有趣的其实是那些来自乌克兰的公司，它们的无人机单次成本和单次杀伤成本都低得多、我希望我们永远不会发生热战，但目前我们显然正处于第二次冷战，一方面是俄罗斯、伊朗、朝鲜和中国，另一方面是西方，由于中国，他们拥有巨大的制造优势。

因此，我认为我们取胜的唯一途径就是超越他们，提高效率和成本效益。现在，至少在乌克兰冲突中，乌克兰正在使用这些廉价的无人机，非常有效。有几个例子可以说明这一点，在那里，机场被轰炸，就像，有一辆卡车，在俄罗斯部署无人机。

无人机飞行并摧毁了所有飞机，其中一架无人机还摧毁了一座大桥，我想是一年前或两年前的事了，一架喷气滑雪式无人机，我想成本大概是 2 万，击沉了一艘价值 1.5 亿美元的战舰。所以，战争的未来肯定是无人机。我认为，我对乌克兰未来的展望是，一旦战争和冲突结束，它也许会成为西方武装台湾的神奇防务制造基地，坦率地说，也会成为美国和西方其他国家的防务制造基地，以备不时之需。因为我认为现在的军队，总体而言，总是在为最后一场战争做准备，所以我们仍在投资过于昂贵的航空母舰、有人驾驶飞机、坦克等等，这些都是完全无效的，也将是完全无效的。

因此，这绝对是战争的未来，你想要最大限度地降低每次杀伤的成本，并保持这些东西的高效率。最后但并非最不重要的一点是，人们谈论自动驾驶已经很久了。十年前，自动驾驶就已经开始起作用了，但由于文化、技术和监管等方面的原因，它并没有凸显出来。

但实际上，它终于来了。截至 11 月 24 日，它在旧金山的市场份额已经超过了优步，我认为到今年年底，它的市场份额有望超过优步，它正在洛杉矶发生，它现在正在奥斯汀与优步合作，但自动驾驶汽车已经来了，它在这里说，所以未来肯定是自动驾驶电动汽车，它现在终于进入了时代，不管纽约出租车和豪华轿车委员会接受它还需要一段时间。成本和价格的下降还需要一段时间，这样才能开始取代更多的司机，等等。所以人们经常担心，”哦，这次技术革命不一样。所有的工作都会消失”

这些事情都是慢慢发生的。这就是最好的例子，对吧？就像人们担心我们会有美国第一大职业类别的卡车司机一样，想想370万卡车司机。取代他们需要 20 年、30 年的时间。不可能一蹴而就。这些人都会找到工作的。

坦率地说，所有其他被人工智能颠覆的领域都是如此。这些事情需要的时间比人们预期的要长很多。但好消息是，它终于来了。顺便说一句，如果你去中国，无人机送货也终于来了。美国在早期做出了非常愚蠢的决定，美国联邦航空局（FAA）说，基本上你需要一个飞行员，你需要视线，你需要购买飞越事物上空的飞行权，所以基本上使得无人机送货在美国基本上是非法和不可能的。在中国，实际上在深圳，这是世界上密度最高的城市之一，你可以用无人机送餐。已经有超过一百万架无人机送餐。因此，无人机送餐不是送到你的公寓窗口，而是送到离你所在位置一两个街区的送餐点，然后你去那里取餐。

但这很常见。如果你能在深圳做到这一点，你就能在任何地方做到这一点。所以这不仅仅是在农村地区可以实现无人机送货。无人机送货在美国还没有普及，但我希望我们能改变空域管理方式，使其合法化，因为无人机有能力大幅降低送货成本，进一步推动这场革命。

对我来说，科技、人工智能、互联网，总是要把事情做得更好、更便宜、更快。我想做到这一点，提高人们的购买力，提高人们的生产力，让他们做更多的事情，有更多的空闲时间，有更好的生活质量，并能真正追求他们喜欢做的事情，无论这些事情是什么。

因此，我迫不及待地想看到这样一个世界：你不再需要拥有一辆汽车，你可以拥有这些自动驾驶汽车，它们带着你，带着你到处跑，每英里的成本比每英里的边际成本低得多，因为它们是电动的。还有自动驾驶汽车，你订购的所有东西都会由自动驾驶汽车和无人机以更低的价格为你送达。

我对未来的发展感到格外兴奋。这就是今年上半年的趋势。我是说，发生了很多事如果有任何问题，请告诉我。实际上，我很惊讶，但我看到相当数量的观众，但这一直是一个为数不多的，我没有看到非常多的问题。

如果你们有任何意见或问题，请告诉我。如果没有，我就收工了，下期再见。我不会像往常一样。下周二，我将为本期节目制作一份文字稿，并在下一篇博文中一并发布，这将是本期节目的总结。

好的。LinkedIn 用户，我有一个问题。我一直在思考，在这个时代，人工智能技术比以往任何时候都更容易制造出超逼真但虚假的图像、深度内容、捏造的评论。IDC Marketplace 正在不断发展维护客户信任的策略，尤其是当真实性、验证、品牌可信度和视觉完整性对购买决策如此重要时。

您认为对确保市场保持可靠来源和数字环境日益充斥被操纵的内容至关重要的系统或创新。

顺便说一下，这在很大程度上是真实的，对吗？不仅在市场上会出现虚假评论，而且在社交媒体和广告中也是如此，比如人们说的深度假话和制造的假新闻，如果你愿意的话。

就像所有这些事情一样 这是一场永不停息的军备竞赛 坏人和好人之间 永不停息的军备竞赛所以，黑客与白骑士。

这种情况正在发生。我们看到很多公司都推出了深度伪造检测技术。因此，越来越多的公司将开始实施这些基本的保障措施，他们将验证评论实际上是由某个人发布的，而这个人的身份已经得到验证，这种内容看起来是真实的。因此，我们现在看到了一种趋势。无数的深度虚假检测公司正在向我们推销。无论是针对媒体，还是针对市场，等等。我相信其中一些公司会崭露头角，变得非常出色。因此，深度伪造变得更好，而攻击它们的公司变得更好，这是一个永无止境的过程。

这场战争会一直打下去。但我猜想，和往常一样，好人会永远领先一步。你一定能想出办法来。我的意思是，现在，尽管质量有了很大提高，但你至少还能感觉到，哦，这些评论，就像，就像，没有用心。

它们不像是由真人撰写的，等等。但是，是的，这将是一个持续的挑战，就像早期一样，持续的挑战是防止有人试图诈骗你，对吗？就像试图向你推销东西一样，早期市场上的骗局层出不穷。

像 Craigslist 可能就是一个比较好的例子，那里有无穷无尽的骗局，但人们找到了应对的方法。

我能对克什米尔和坦桑尼亚感兴趣吗？我不知道你做得怎么样，但我很乐意看一看。只要把甲板发给我，并在 LinkedIn 上给我发一条信息，提供足够的牵引力、工会经济学方面的信息。

交易条款，等等，我们现在来回顾一下。我们在非洲投资不多。我们做了一些，我必须承认，主要是在尼日利亚、肯尼亚和南非这些较大的市场，但实际上我们取得了不错的成功。我们在阿尔及利亚有一个非常大的退出，在中东也有相当多的退出和成功。

所以，我们很乐意看看。

好了，我想本期节目到此结束。感谢您的收看，我们下期再见。暑假愉快如果在那之前见不到你的话

我与 AngeList India 的首席执行官 Dhruv Sharma 就Offline 进行了一次引人入胜的炉边谈话。Offline 是南亚领先企业家的私人会员社区。与 Utsav 和他的非凡独角兽创始人社区聊天非常愉快。印度的科技前景非常光明！

Here is Utsav’s cheat-sheet of golden nuggets that landed hardest with the Offline crew:

🔑 4-point “one-hour” deal filter

1️⃣ 🔥 Team = eloquent and execution-obsessed
2️⃣ 💸 Business = big TAM + pay-back ≤ 6 months
3️⃣ 📄 Terms = fair vs traction, not frothy
4️⃣ 🎯 Thesis fit = lines up with the future you actually want to exist
Miss even one box? → pass. Discipline > FOMO.

🤖 AI ≠ free pass

75 % of Q1 US VC dollars went to AI, yet moats are rare.
“Copilots for X” is a week of code & a crowded party; differentiated data is where the edge lives.
Valuations at 💯× ARR? Careful… the bill for 2021 déjà-vu is coming.

📈 Marketplaces 2.0

Consumer side is largely done; the next trillion-dollar wave is B2B inputs, cross-border & re-commerce.
AI superpowers: auto-translate listings, auto-price, one UX for chat and search.
Answer-engine optimization (AEO) is the new SEO.

💰 Angel math

< 50 bets → odds stacked against you. 50 + → probability tilts.
Stay in your zone of genius, keep cheques small, and sell 50 % of winners on the way up to lock real DPI.

🛠️ Productivity hack-stack

Ruthless outsourcing: VAs for inbox, bills, slides; butlers for offline ops; five rotating nannies on a shared Google Sheet.
Saves cycles for reading 100 books a year, kitesurfing, ✈️ and hosting Jeffersonian salons.

🌏 Macro take

De-risking from China = India’s manufacturing moment. Massive whitespace for SME tooling & export-ready marketplaces.

💡 Biggest reminder: Ambition scales fastest when curiosity, discipline & joy ride together.

Huge thanks to every Offline member who poked, prodded and pushed the convo further.

Transcript

Dhruv Sharma: Good evening everyone. Thank you for joining us Fabrice. It’s great to have you with us. Thank you for making the time. What do you need to know about Fabrice? Most of us know him as the founder of OLX and now FJ Labs. Fabrice is from Nice, France, lives in New York, loves the life of exploration and adventure, and he is built not one, but three extremely successful companies that we don’t always know about.

And the other two being Zingy and Aucland. And just very delighted to have you with us, Fabrice.

Fabrice Grinda: Thank you for having me.

Dhruv Sharma: Before we get into the substance of this conversation, I wanted to ask you about these Jeffersonian dinners you’ve been having all these years.

What, in your mind, Fabrice is a good recipe? Figuratively speaking for these dinners. And what has it meant to you over time?

Fabrice Grinda: So, the reason I have them, is think back to when you were in college. You’re studying everything and anything, and you’re having conversations on the meaning of life and you’re being extraordinarily thoughtful.

This world, this simulation matrix, whatever you wanna call it, reward, specialization. And in fact, specialization is the reason we have the quality of life we have. But we as humans should not purely be defined by the job we do on a daily basis. And actually, I value being able to be polymath and think through a whole bunch of different topics.

And as a result, what I’d like to do is bring together about 10 people. like eight to 10 around a dinner table. The format of a Jeffersonian dinner is that there’s only one conversation at a time. You don’t have side conversations The sum total of the minds at the table or on the issue.

And of course, no one is expected to speak for more than 30 seconds or a minute at a time. I give the topics ahead of time and the topics are very far and varied. It could be like how to reinvent democracy for the 21st century, life in a post singularity world religion in 2100. The ethics and morality of torture, or sometimes the most open-minded ones are very interesting.

It’s like what has been on your mind or something you’ve learned recently that you could share with everyone. This way everyone presents for three, four minutes and then we all discuss it for five, 10 minutes. It’s an amazing way to elevate the debate

The point is to get to the truth. I find that we live in a world where intellectualism is not valued enough. it’s wonderful to be thoughtful about a variety of topics with extraordinary people some of the best friendships and frankly, fascinating conversations upcoming these.

I organize one about every three weeks, changing the guest list. I’m, extraordinarily strict. Your phone cannot ring or you’re not invited again. You cannot be one minute later you’re not invited again. Like I really, and, and by having these rules, people take it seriously.

They come prepared, they read the questions, they know what they’re gonna talk about. and I changed the guest list with like intellectuals at people from academia, policy makers, entrepreneurs, and people from a wide walk of life. it’s super fun. I wrote a blog post on, Dialoguing dinners on my blog.

I’ll put the link in the chat and you could read, sample questions, organization, the email I sent to people about how to get ready,

Dhruv Sharma: That’s amazing. Fabrice, when was the last time you hosted one

Fabrice Grinda: I did one through weeks ago and I’m going to one tomorrow night.

Dhruv Sharma: Oh, that’s great. So you’re not the only host that means you others who host these dinners as well? Similar format.

Fabrice Grinda: people from my circle are hosting the ones tomorrow night. There’s an organization in the US called Dialogue, organized by Peter Thiel and Orin Hoffman, which brings together several hundred really smart people to talk about, problems.

The organization has one event per year for extraordinary people. typically, in California, in April or May. during the year, I use the organization to curate and find people. So this way I outsource, which is something I like to do in life, outsource almost everything I can to focus on things that are in the most fascinating.

They help me find the guest list and think through questions.

Dhruv Sharma: That’s amazing. you also have this thing, Fabrice, that you launched, about a year ago. It’s called Fabrice AI, and it’s a way for the world at large to, in a sense gain a window in your thinking. For those of us who may not know about that, what can you tell us?

Fabrice Grinda: In venture writ large right now, it’s really a tale of two cities. Venture at large has been down like 70%, and yet there’s been a massive recovery, at least in the US in adventure investing because of AI. In Q1 25, 70 3% of all investors in AI. And by the way, a year ago I thought already AI was in a frothy bubble and the bubble has only inflated more since.

And we’ll talk more about that later. But because I started being pitched infinite AI deals, I wanted to really understand what was easy, what was hard. I’m like, you know what? Why don’t I code my own AI to understand, where we are? and what are the complexities? I started by using GPT 3.5 or 3.0 didn’t work actually two years ago.

Didn’t work particularly well. So, I started using Langchain and Pinecone then I started using an open source library to do, voice of text. Nothing worked particularly well, I have to admit. And the complexity was way greater than I thought in the sense that the source files for all the knowledge and data structure, which was basically everything I’d ever written, every podcast I’d been on, every YouTube video I uploaded, et cetera.

The transcription of all of these, the separation of the speakers. So even the data formatting preparation was extraordinarily complex. And then the data retrieval was complicated. I had to use MongoDB ’cause I didn’t, it didn’t work in just my SQL. I put all this work literally a thousand hours of work, and I thought originally it was gonna be like 40.

And eventually GPT released 4.0, and I ripped out my tire backend, connected 4.0 replaced my open source library for voice attacks, put Whisper and everything started working. The point of that is when people started pitching me things like Co-Pilot for X, Y, Z, I realized, okay, when you’re pitching me, I can code in a week.

That is not a defensible, viable idea with a moat. And, and most of these AI companies, I was being pitched that insane valuations. Honestly, were not that differentiated, didn’t have a mote. And so it was extremely useful. It also showed me the risk of both open AI expanding its reach and little by little taking share from other AI companies.

As time went by, I started ripping out all my backend. right now I temporarily, turned off this AI. It is still running, but now I’m using a third party company called Delphi because they were formatting better and their voice interface. So the OpenAI doesn’t have a great voice interface.

So there’s a company called 11 Labs, that could replicate your lab. But I needed a code, the API, to connect to them and train it. I already have all the data sets. I’ll upload to Delphi, use them for now, but I’m currently coding the next version of Fabrice AI, which is combination of my avatar you can have a call with and, pitch me. So I’m currently transcribing all of the calls I’m having with founders, including the deck. the call and my debrief and recommendation, my hope is that ultimately founders can get intelligent feedback. It’s not actually can it surface of deals, it’s just, right now we get 300 deals inbound a week.

We, FJ Labs talked to 30 to 50 of ’em and I Fabrice talked to like five of them. And so there’s. 270 founders a week that don’t speak to anyone at FJ Labs and 295 that don’t speak to me. so I’m like, can I create an AI good enough that as a service to humanity, if you’re a founder, you’re gonna get intelligent feedback.

Now your deck, your presentation’s, not scripts. The total addressable market size is not, gr is too low. The unit economics don’t make sense. there’s a lot of competition. can I get there? I don’t know. I only have 150 transcripts right now. I think I need at least 500 to get there.

But I’m playing with it. And maybe it’ll even surface fee deals. I probably a year away from having something that I’m satisfied with. it’s fun, it’s an intellectual exercise And it’s a public service to founders Fabrice AI, already works pretty well.

You can’t pitch it, but if you wanna have a conversation on anything I’ve ever written about. you can have a meaningful conversation.

It works really well for that. It just, you can’t pitch it ideas yet and have intelligent feedback.

Dhruv Sharma: I do something that I asked it, which I’m gonna maybe quote later just to check if that tallies up with your current thinking in the topic. But we’ll get to that. I think this is a great moment.

I mean, you said 300 deals a week. You said a week, right?

Fabrice Grinda: Yes.

Dhruv Sharma: That’s a lot. And I believe you’ve had this four step investing framework that you’ve stayed true to all of these years. Can you tell us about that?

Fabrice Grinda: Yeah, so the 300 deals inbound. most VCs are structured very differently from us, right?

Most VCs have these teams of analysts and these poor analysts are basically cold calling and scoring LinkedIn and going to tech conferences to try to find deals. We’re in a very privileged position, right? I never intended to be VC. I describe myself as an accidental VC. It just so happened that when I started my very first company back in 1998, which is an eBay of Europe, I was a very visible consumer facing internet, CEO.

So I was being approached by other founders and at that time what it meant to be a tech founder was not very clear. even the basic stuff like what tech stack to use, was not obvious. There was no open source.

it cost millions to turn the lights on. I had to build my own data center So I thought long and hard, should I be an angel investor in parallel to being a tech CEO, because it is a distraction from your core mandate. And so I decided, If I can articulate lessons learned to others, it makes me a better founder.

I’m running this horizontal multi-category site as I did for OLX. If I can meet vertical founders and keep my fingers on the pulse of the market, it makes me a better founder. So I took key to be an angel as long as I don’t waste too much time. back in 98, I created these four selection criteria to decide in one hour if I would invest in a startup or not.

So I’d meet the founder and evaluate four things. One, do I like the team? Now, of course, every VC in the world will tell you I only invest in extraordinary founders. The thing is, extraordinary founder cannot be like porn. It can’t be, I know it when I see it. And so we’ve actually defined what that means to me, that someone who’s extremely eloquent, who has a way with words, who can weave a super compelling story and someone who’s very eloquent and salesy can raise money at a higher valuation, can attract a better team, can do better BD deals, can get more PR, but that’s a necessary but insufficient condition.

You need someone who also knows how to execute. and the way I tease that in a one hour call, they know how to execute, is number two. Do I like the business? And so what is a total addressable market size? What are the unit economics? We mostly are seed investors, so post-launch, post product, early revenue.

Even when we do invest pre-launch, we expect people to thought through it. Have they done landing page analysis. Based on that, what are the density keywords in Google? What is the average CPC What is the conversion rate to people signing up? What do you think the customer acquisition cost is based on the average conversion rate from visit a purchase in the category?

What is the average order value in the industry? They better be in line with the average order value in the industry, not expect to be 10 x above it. What’s the margin structure? Do they really understand their cost structure? Do they recoup their fully loaded costs on a contribution margin two basis after six months?

And do they three x it an 18 month? And if not, will the unit economics improve with scale without needing all the stars of the multiverse line? Number three, the deal trips. Nothing’s cheap in tech, but is it fair? Is it fair in light of, the traction, the opportunity, the team, et cetera. And number four, is it in line with their thesis of where the world is going?

And I have a clear perspective on the future of food, the future of climate, the future of mobility, the future of real estate. And I want them something that is both useful to the world and meaningful in line with where I think the world’s going. I will invest if all four criteria are simultaneously met.

it’s not three out of four, it’s literally four out of four. An amazing team is not enough because if an amazing team meets a bad business model typically it’s the reputation of the business that wins over the team. some founders or VCs will say, the team will figure it out.

The thing is that to me leads to a too high failure rate. So far we’ve invested. In 1200 companies, we’ve had 355 exits. We’ve actually made money on 45% of our exits, which for a pre-seed seed investor is an extraordinary hit rate. because in 95% of startups after five years, that’s that have raises seed fail.

So we’re doing very, very well because we’re disciplined evaluation because we’re disciplined unit economics. the reason we get all the deal flow is because I have a brand. People know me as the asset like, network effect marketplace investor. And as a result it’s a very privileged position where this deals just come in.

So we don’t actually do any outbound. The deals come in, the other VCs share deals with us, the founders, that we’ve backed so far. So 1200 companies is like 2000 founders. They send us to their friends, they come back to the next company, they send us to their employees, and then of course cold inbound.

And we get over 150 cold inbound deals for a week.

[00:13:24] Dhruv Sharma: That’s very impressive. Fabrice, so you started this journey of angel investing and it worked out really well for you. In fact, you say it made you a better founder for any of the founders here attending who have that question in their mind, should they start the angel investing journey? and if they already have, should they scale it?

What would you say to them? And based just not on your experience, but maybe also based on the experience of your friends, which might have been different from yours?

Fabrice Grinda: I think it depends on you and who you are, what you want, what your advantage is, Like I am, intellectually curious polymath, right?

The reason I like angel investing is there’s infinite problems in the world and boundaries basically see these problems. And by the way, something like climate change or inequality of opportunity is not one problem. It’s a thousand sub problems. the way I see founders, they see problems and they wanna go solve them.

And I found it fascinating to be able to meet these founders that are brilliant people who want to solve these problems, to hear the stories, and then share best practices with ’em. And the number of founders who’ve told them the interactions they’ve had with me have been meaningful, I made me realize that I can get fundamental leverage through doing it.

That said, in order to not be distracted, I only did dealt with inbound deal flow, and I only focus on marketplaces because I knew that’s what I knew innately, and I had this mechanism by which I would invest it in one hour meeting. So should you do it well? Do you have free time? Do you have free cash?

Do you like it? Is it interesting to you? And by the way, you’re only gonna make money if you have a diversified portfolio. Meaning on average, if you invest in less than 50 startups, you’re gonna lose money. If you invest in more than 50, you’re gonna make money. The more companies you invest in, the better you end up doing.

And so you really want a diversified portfolio. So should you scale it, not necessarily. So reflection of your personality, if you like doing it, if you have an amazing deal flow and if you have free cash and time, why not I didn’t intend, to become a venture capitalist post selling, OLXI. I created a family office, which was investing companies, creating companies, but I tried a whole bunch of other things.

I try to buy Craigslist, I try to buy eBay classifieds. I try to convince the Castros to let me run a free trade zone in Cuba. Actually a special administrative zone, copying what, Deng Xiaoping had done in Guang Dong. I thought of becoming a public intellectual. It’s just these things required someone else to you to say yes.

And I threw this spaghetti of the wall, this spaghetti didn’t stick. And writing people checks actually kind of worked. then all of a sudden I started getting approached by third parties saying, I wanna expose you to what you do, that’s why I became a VC. But I’m a non-traditional VC because I’m a VC who behaves like an angel.

Every year. I invest in 150 deals. I write small checks. I don’t lead, I don’t price, I don’t take board seats. I try to be this value added friendly guy. Only do it if you think it’s fun and interesting and it’s a reflection of your personality and you have a process and deal flow.

If you don’t have deal flow, getting it is hard, because you guys are visible and successful, you do have deal flow. but it’s not obvious that you should do it. Many of my friends would rather use the time they have post exit, to do something else.

And that’s totally fine. and that could be anything, right? Like from investing more time in family and friends and relationships, to hosting intellectual salons, to entering politics, to doing different forms of philanthropy. Now, my take of philanthropy is a bit contrarian because I actually find that what I do through tech is a form of philanthropy,

As I said earlier, it’s a boundaries see problems and they try to solve ’em. So then I try to solve problems that I think are meaningful. And the reason I’d like the for-profit mechanism is because it creates something that’s scalable and repeatable and doesn’t need to go fundraising every year.

Dhruv Sharma: Fabrice, before I ask you the next question, this is for everyone in the audience. If you have something you’d like to ask Fabrice right away, let’s do it. Maybe just leave your question, or just tease your question out in the chat window, and I’m more than happy to invite you to come, ask you a question.

Fabrice before we go ahead, reaching out of the Castro family to set up, a trade zone. We kind of let you proceed before you tell us a little more about that.

Fabrice Grinda: So, by formation, my background, I’m an economist. I studied economics at Princeton and one of the things I love, was market design, which is why I was well suited for creating marketplaces, in terms of like measuring elasticity of supply and demand, figuring out what the correct, key creator business model was, et cetera.

But studied economic performing in China, and I was a huge fan of Deng Xiaoping, who basically single-handedly took a billion people outta poverty. And I went to live in China in 1994. I studied Mandarin at Beijing Normal University.as I was thinking through after our lack of giving back and thinking through what could I do to make the world a better place, I’m like, you know, these free trade zones and more importantly, special ac administrative zones have that profound impact in changing the livelihood of people around the world.

You see what Dubai has become, despite not having the same natural resources than the other, a Arab states. I love the Caribbean. I like splitting my time between, the US and New York, which is a haven of an intellectual, social artistic endeavors with a place where I can be more reflective and read, write kite, surf, do yoga, meditate.

now I’m working there, I play tennis, I play paddle, and, and that’s my work life outs. I started looking at the time, so we’re talking 2013 at Cuba. Cuba is, foreign direct investment was about a hundred million a year. The average salary is about $20 a month. And it’s essentially a failed state.

They have a highly qualified, labor force with super low GDP, nothing working, and they have large swath of the country that, are underutilized and completely empty. So I’m like, look, give me, a couple percentage points of the country to build in my special economic zone where I’m essentially 70 miles from Florida.

So creating a small agricultural and manufacturing hub, obviously using capitalism where you can export to the US makes infinite sense. they originally said yes, the reason they ultimately said no is I didn’t just want a free trade zone. It needed to have a different constitution, it needed to have a different currency.

it needed to be its own little city state because I needed respect for property rights and there couldn’t be arbitrary confiscation of all these assets. the currency didn’t need to be the dollar. It could be the yen, the Euro. it needed to be convertible.

It couldn’t be the Cuban peso. and they thought that was a step too far. ultimately, I did think, it probably is a good thing it didn’t happen because the issue is, imagine it had been very successful. they could have just confiscated it, right? they have a military, so I’m not sure I could have kept it, even if it worked extremely well, difference between the state itself and, doing it like Deng Xiaoping did, and being a third party foreigner, trying to implement it.

But it was a great idea. I think it would’ve worked because, bringing incentives, structures and systems there, and I had actually a couple hundred million, I think I had 900 million lined up, which tripled the FDI per year, of the country to go and build this. So, didn’t happen, but it’s okay.

Dhruv Sharma: A remarkable story to tell. are there other economic reformers you look up to, Fabrice or if any of us are interested that you would encourage us to read more about their work?

Fabrice Grinda: Well, my role models and heroes through history, probably, Octavian or Augustus because he single-handedly created the Roman Empire and he inherited from his uncle Julie Caesar at 17.

 It was not given that he was gonna be the winner with Mark Anthony and the other pretenders to throne, even Castus breach was gonna pretended. And he put an end to the Civil War, created the Pax Romana created a, basically a period of relative peace and stability that lasted for a hundred years and a quality of life that was unheard of at the time.

The other one is Alexander Hamilton. The debates between Jefferson and Alexander Hamilton in terms of what the US should become. we are all collectively glad that Alexander Hamilton won that because Jefferson had more of a Russo-Esque agrarian society. We don’t repay the debt, we don’t have a modern financial system, et cetera.

Where we ended it was extraordinary. But then of course, some economists have, played a fundamental role in creating the world we knew today, like, Keynes, right? and I don’t mean Canadian thinking, I actually think is actual impact of creating the IMF, creating the bread and wood systems, to help rebuild post-World War II.

Was profoundly, impactful in creating the post-World order that we had that led to peace, stability, and prosperity for 70 years post-war. which is sadly, coming to end right now. And even the thoughts of like Freedman and others, now of late, right?

It’s hard to have people that you look up to, in democracies by design. It’s very hard to make changes. And, and so it’s not a, the conflicts in democracies, it’s not a bug, it’s a feature of democracy. The founding fathers, and thus through most democracies, want check and balances between the Senate, to the judiciary and the legislative and the executive.

And that’s why almost nothing gets done. But that’s okay because the fundamental systems are pretty good and people, are worried about the conflicts they see in today’s democracies. But the reality is they’re not worse than they were before.

People forget, we had a full civil war in the US in the 19th century. There were like race riots in the 1960s. There were anti-war protests that were violent in the 1970s. what do you have today? Like, is nothing relative to that. And I think ultimately. Even though it ebbs and flows the system, does lead to the right outcomes, even though as Churchill said, the US could be counted on making the right decision.

After all, other options have been exhausted; which of course is the second most, popular quote of Churchill after that democracy is the word of also, except for all others. So, yeah. Are there great leaders, in time that I also, admire, et cetera? You know, Allah Churchill, absolutely. But, but in terms of economics, no.

I think, Deng is probably the one that has had the most impact, but what’s interesting, of course, is now India, is growing extremely rapidly and is helping get hundreds of millions people out of poverty as well. So it’s, accelerated. The leaders of Singapore have been extraordinary, so that’s probably another thing. The issue with these dictatorships though, is enlightened dictatorships are only as good as your dictator. sometimes you have enlightened dictators, but then you have incompetent, fools like Mao or Xi Jinping, that set these countries back dramatically,

Like Zimbabwe used to be the shining star of Africa, and then Mugabe destroyed it, these, aspiring autocrats, like Erdogan in Turkey, he’s betraying the legacy of Atatürk.

Dhruv Sharma: Fabrice, when OLX first enter India, which year was it?

Fabrice Grinda: Very early in our journey. I had been friends with, Suvir, and Avi from Bazi, who of course had had then sold to eBay, became eBay India, and they, and s Suvir had created Nexus, a venture capital firm.

And, OLX, originally I would’ve preferred it to work in the West, but because they were incumbents, marketplaces with liquidity didn’t really take off. And so we opened in a bunch of countries, and it really took off in places like Portugal and Brazil,

Suvir is like, look, I wanna fund you, but you have to enter India. I think the market is amazing here. in fact, I’ll help you, find your country manager. that was the first employee, at eBay, India, which is Amar, who’s on this call.

I don’t actually remember the year, but Amar maybe does,

Amar Jt: It was 2008. I joined in the end, but I think OLX was already live So it must have been 2007. The site must have been live 2007. we cleaned up a few things after that, but, the real work started 2000, I mean, when we started scaling up for 2011 when we did our first big campaign, but during that time we were doing interesting, things about consolidating and stuff like that.

Fabrice Grinda: Yeah. And, and load the groundwork. But yeah, we basically launch very early, but like, yeah, the real effort and cleaning up the side, making sure we had focus, having all the languages, and making sure we’re in all the different languages. And, started, when Amar joined in 2008 the company was created in 2006, and we went live probably in early 2007, so very early in the journey of OLX.

Dhruv Sharma: Yeah. So, 2008 Fabrice, 17 years. And, clearly there’s been an economic transformation in those 17 years. Now we’ve witnessed it firsthand, you maybe have a different perspective because you’ve always had one leg in and one leg out. So what does this look like to you?

Fabrice Grinda: I’d say the main transformation. what led to the real explosion in India? I would say was 3G, right? Like once smartphones started, coming of age, and you started having these super cheap Android phones with 3G, you started having the entire population having access to, I mean, imagine today your poor farmer In India, you have the sum total of humanities knowledge in your pocket.

You have free global video communications, and you have a super smart advisor in the form of GPT in your pocket. So basically you have more access to information and more access to communications that the president of the United States did 25 years ago. And that’s as a lower class farmer in Africa or India.

It’s mind-boggling. people take for granted the revolution we’ve had, but it is so profoundly transformational and actually egalitarian, right? It gives, people access to these tools in a way that they weren’t before. And, and we lived it because we were early, because the company I built before OLX was a mobile content company.

We were early in the journey and discovering that we needed to go mobile. and so Ol X from the beginning, we had a mobile website. We were very early with apps. So we lived the Indian, internet revolution through mobile from the very get go. and, and, and what we did right, was realizing if you wanna reach people that are offline, the best way to do, it’s through TV.

And that’s why in 2011, we started doing these TV ads especially around cricket, which helped bring a lot of people online and of course helped consolidate OLX the leading classified buying sell site.

Dhruv Sharma: There’s a couple of questions in the chat window. Why don’t you ask a question, please.

Utsav Somani: Yeah. Building on that Cuba thing, what’s the one crazy thing that you’ve attempted in the last decade that’s not worked out, but you wish it did?

Fabrice Grinda: in 2006, when I left, my prior company Zingy, I went to see Craig, of Craigslist and I’m like, Hey, you are creating, you have already have network effects. You’re doing an amazing job.

you have a 1995 interface, and more importantly, you’re not moderating any of the content on your site. Let me run it for free. Not ’cause I wanna make money just because I wanna make the world a better place. I wanna provide a public service to humanity. And we could always talk later if you’re happy about some form of conversation.

Of course, even though I was credible, ’cause I’d sold my last company had grown it at 200 million revenues profitable. he didn’t take me seriously. So, then I went back in 2013, when I left OLX, I went to Craig and I’m like, Hey. Look, I build a site bigger than Craigslist, 300 million units, 10,000 employees, 30 countries.

It’s amazing. And we’ve solved all the problems you have. Let me do this again for free. and he said no. I think he and the CEO were like so high on drugs. they were fading out of consciousness. I don’t think they even remember we met. They didn’t remember we’d met before. Every time I met them, they forgot we’d met before. It’s like mindboggling. They just didn’t care. I tried so many times over the years to take it over. I tried to buy it. They said, no, I tried to run it. They said no, et cetera. So that one, because I said earlier, these things, they require someone else to say yes.

I try to buy eBay classified twice. So, eBay owned very big classified sites in, Western Europe, Canada, and Australia, especially in Germany. I tried to buy it in 2015. I had, lined up private equity. It was like a six or $7 billion deal. They said yes, signed an LOI. But then they divested PayPal, when they divested PayPal, they had enough money that they didn’t feel they needed to anymore.

So they changed their minds. a year worth of work for nothing. I tried again in 2019 and they ultimately sold, I thought I’d won, but they sold it to my biggest competitor at OLX, which was at Adevinta. So sadly I didn’t win buying eBay cost. in hindsight, I probably would’ve hated it because eBay is a ginormous company with like, they had like, o well actually is kind of built for me by me and it’s kind of, I would like to believe it. Enlightened dictatorship with centralized, everything. each country had local product and local et cetera. Like we had one tax stack and I could say jump and, you know, people would say, ah, hi.

eBay was like different silos, different techs in every country, overlapping fighting fiefdoms. It would not have been easy to execute on. So, in hindsight, probably, something good, but I try to build, in the Caribbean and to Dominican Republic, a massive kind of off-grid commune, not a commune in the communist sense of the word. More of a place where tech founders could come and build tech startups and innovate, where spiritual leaders can host events where, where, where artists could come and create. and, and the idea was to foster. Something not too dissimilar to my Jeffersonian dinners by bringing smart people around to go and create things.

And I bought several hundred acres of land. There was a mile of beach front. the problem is, the Dominican Republic was more of a banana republic than I expected it to be. that was my second attempt. The first attempt with a massive piece of land in Belize. definitely a banana republic and failed.

And so I’m like, oh, Dominican Republic, it’ll be fine. Democracy, et cetera. But like everyone wanted bribes from the local mafia to the mayor to the vice president of environment. I didn’t wanna bribe anyone. Maybe that was a mistake. And so after like seven years of things going from back to worse with, tropical diseases, burglaries poisoning my dog, we, we even got full on attack by gun men with shotguns.

So there was a Call of Duty shootout in my garden. The time had come to move on. I left after seven years of essentially failure, lost like $8 million in the process, and I moved to a much smaller, but much more beautiful and safe place in, in Turks and Caicos where I don’t have the incubator and all that stuff, but I do host three tech conferences every year.

I brought the CO of eBay. I bring unicorn founders to brainstorm about the future of humanity, the state of tech. I host a crypto conference, et cetera. I own seven acres of land it’s much smaller. But I’m toying with the idea of moving to either Antigua or Nevus and getting 30, 50 acres and restarting that.

But again, it’s more for fun as much as anything else. It’s like intellectually stimulating it’s less scalable, meaningful than what I do on the professional side, some of the companies in a way that I’ve helped build, like Vinted, you know, Vinted now is like 10 billion in GMV, a billion in net revenues, hundreds of millions of users.

And it’s the first cross pan-European cross border transactional site. My right hand man at OLX is now the CEO there, the leverage I have through my tech investing in terms of impacting millions, if not billions of people, is way greater than these things.

But these things are fun and also an amazing place for me to bring my entire family and extended family together.

Dhruv Sharma: Fabrice, before we switch our focus to marketplaces and AI, I’m seeing there’s two or three questions, we can go through Amar Jt, would you like to ask your question? Then maybe Dheeraj can go on and we’ll take it from that.

Amar Jt: I think there are so many questions for Fabrice all the time, I think he’s doing so much. I think it’s fascinating. with the intensity and the speed at which you work, Fabrice, you always astonished me. You work on fifth gear and I think I sometimes I feel I’m not even on the first. really always so amazing to hear you.

I always remember, whenever we used to chat about it, I think India was one. I mean, you are already investing in India a little bit. I handle emerging markets now for Spotify, when you used to build companies, you know, and invest in them, you used to look at the world. When we started OLX, we were looking at, okay, how do we take great models from us and take it to the world?

Of course, you have moved way ahead of that now. Now what is your thought process about emerging markets, there is not enough happening there, but also enough happening what is your current, understanding of this market and how you’re investing? What do you think about the future? Love to have your latest view on this.

Fabrice Grinda: Maybe later I’ll explain how I can do as much as I do, because I think there are many trucks in terms of what you can outsource in life and how to be, productive in order to focus on the things that you love doing and not be distracted by all the things that take time that are actually non-value added and not in line with your own ethos.

So, FJ labs throughout history, we’ve really been reasonably constant, which means we’re about 50%, US and Canada, about 20% Western Europe, about 15% in Latin America. 5-10% in India and the rest of the rest of the world. So, emerging markets for us really is we prefer larger market countries like Brazil, like India, because there are large enough that you ha you can have companies that can scale two B unicorns just in that one country.

You end up having actually viable, venture capital ecosystems. You end up having exit markets, et cetera. So I’d like larger markets. But market large markets like Nigeria and Kenya, which are 200 and plus million. A hundred plus million, or actually still too far down the curve. So we have a few investments there, but we’re not a focus.

because of the geopolitical, macro trend right now of the new Cold War between the Westford writ large and China, well, I guess it’s China, Russia, Iran, and North Korea, and one side, and the west on the other side. everyone’s trying to move their supply chains out of China. And the only place that I think is scalable and interesting and the most scalable and interesting is India.

But because India is manufacturing base is mostly SMEs and as opposed to large manufacturers, there’s a unique opportunity to invest in marketplaces that help small business manufacturers in India export, and work with, Western companies. That’s why we’re investors in Zyod, Which is helping apparel manufacturers in India sell to brands in the west because they don’t know how to answer RFQs. They don’t know how to do prototyping. They don’t know how to do customs and export, et cetera. And we’re doing this kind of vertical by vertical. In fact, Utsav, one of this thesis right now is like scaling manufacturing in India and investing in manufacturing companies, which is fully aligned with this general macro trend of like moving manufacturing out of China, especially in India and scaling it and building brands locally, et cetera. So, that said, if I’m building a company, I would still be focusing on the US market right now because it’s where you get the most value.

You have 350 million rich people that are early adopters of your products. you can sell them at pretty high prices. it’s easy to raise capital, it’s easy to get exits. Even if you failed, you get acqui-hires and you can still salvage from the exit. I still prefer building and investing in the US. That said, especially a few markets like Brazil and India – are fascinating of course, Warren.

And if I was in India and I have a comparative edge there, I’d be very happy to focus and invest and be in India. it’s harder to build, are there interesting things happening everywhere globally now? Yes. the cost of launching a startup has basically gone to zero.

Even a non-programmer could use Lovable or Cursor and start creating a startup for almost nothing. So between, Cursor or GitHub with CoPilot and AWS and the cost of launching a startup and you have an explosion in creativity and company creation from non-traditional backgrounds.

You no longer need to be a Stanford computer science grad. we’re seeing interesting companies emerge from everywhere. But because the world is still capital constrained and you need people to buy your products, I still try to focus on the larger, more interesting markets. We’re seeing amazing things happening everywhere, and we do invest a little bit in Vietnam and the Philippines

It’s just that all that falls in the 5%.

Amar Jt: Thanks, Fabrice. And for everyone else, I want to really tell about him. He is, one of the most positive people I’ve met. and, I think he’s very large hearted, very transparent.

there’s nobody in the world I’ve met who will share everything about himself openly. I have found that amazing because I feel that people operate in two worlds of wonder that they show and watch they live. I think he lives the world so openly. And I think since I worked with him long time back, I feel that he’s relentless.

I wish you more power, for your amazing. You’re unstoppable and I think keep going and love to catch up with you soon.

Fabrice Grinda: No, thank you for the kind words. Look, I’m built ambitious, right? Where does that come from? Who knows, right? I never built companies to make money.

I built companies. I like creating something I nothing. I’d like having an impact and try to make the world a better place and living what I find is a purposeful life. Now, to Amar’s point, most people I think don’t, or not the true, authentic self, right? Like they have a facade when they interact with the rest of the world.

But when you are authentic, it comes through. And by the way, it’s so much easier to be yourself. Now that said, if I lived at, I don’t know, Mexico and there was risk of kidnapping of my kids or me or whatever, I probably wouldn’t be as transparent and open as I am. I feel that I can be in a position to share and be my truth with no downside and risk.

it resonates better with people to be authentic. I’d like to answer Dheeraj’s question because I think it’s very important, given the current bubble in AI. So Dheeraj, if you wanna ask it.

Dheeraj Jain: Yeah. you’re right. But I think there was two aspects. One is, is it more complex than the previous generation of startups, in the previous decade you could foresee where the market’s gonna evolve, what the product and the solution.

the uncertainty was very limited. But now some companies have scaled up so quickly, they like 10 million ARR, which used to take four, five years now happening in one year. And then the future rounds become very expensive. You know, you don’t probably able to do pro-rata and all that kind of stuff.

So that’s one side of the question. The second is also these companies, as you said, 75% of the companies in the last one year, are they also, gonna fail faster because the newer ones will come eventually. So they’re kind of two different things.

Fabrice Grinda: So look, is AI a megatrend?

Absolutely. will it transform the world in ways we cannot even begin to imagine? Absolutely. I also think though, like in every other tech revolution, it’ll take longer than people think, right? Like in the nineties, we had pets.com, we had Webvan. the ideas were good because ultimately we ended up having Chewy Instacart and Uber.

the tech and the market weren’t there right now, the valuations in which people were investing, implied that AI will. Take over the world, lead a massive productivity revolution immediately. And we on this call, or the early adopters, a hundred percent of our startups are using AI to improve programmer productivity, to change customer service, to improve sales, to actually change the funnels of buying and selling.

But when I look at these larger companies, or even SMBs, they are so far removed from implementing this. I suspect this revolution is gonna take a lot longer than people think. there’s gonna be this moment of disillusionment that is coming. And as an investor, I like to be contrarian.

in February 21, I wrote this blog post entitled, welcome to the Everything Bubble, where I said, if the valuations are so insane in every asset class, if it’s not anchored to the ground, sell it. I suspect something similar is happening in AI right now. people missed the boat, they should have invested in GPT or Open AI five, six years ago, and they missed the boat.

So they’re investing in everything else. The problem I’m seeing is multifold and it harks back to the problem of 2021. For each AI category, I see five or six or 10 well-funded competitors. So whatever category it is you have, you have a team from Harvard and a team from MIT and a team from Princeton and a team from Stanford all with 20 million of funding all going after it.

What worries me is that the competition between them is gonna bring the economics down to zero. They’re fighting for talent, they’re fighting, they’re increasing their customer acquisition cost by spending money and marketing or sales, and they’re decreasing pricing in order to gain share. So it’s leading to non-economic behavior.

One big problem that’s super competitive with a lot of companies is hard to tell of these amazing companies, which one is gonna win Problem number two, valuation you’re investing in is very high, such that if you don’t grow to a hundred million in ARR in record time, it’s hard to imagine how you’re gonna make money problem.

Number three, you may think you’ve won and then all of a sudden Open AI offers what you’re doing for free. you go from zero to 4 billion evaluation like, Olive AI, or you’re like Lensa went from zero to six 60 million a month, or 30 million a month back to zero, three months later.

When people realized they had used it and were done with it. I see these companies, even the current winners, right, like the two winners from an A RR perspective right now are Cursor and Lovable. If you’re at a hundred million ARR in record time. The thing is, can I imagine a world where GitHub copilot just gives it away for free?

And this is a race to the bottom in revenues in the future. Yes. Can I imagine a world where for whatever reason, opening I decide this is something they want to do and they offered away for free. Two years ago, I used Midjourney for images. Now I just used Dall-E. Two years ago when I wanted to do a video, I used Runway.

Now I just used Sora. this is kind of scary. You can have, category expansion my Open AI. Lots of competitors, new orthogonal people arriving. deep seek ended up not being an existential threat to Open AI, but it could have been. these deep seek moments could happen in any major vertical.

it’s not to say I don’t invest in AI, but I’m very careful. The AI invested in is vertical applications, based on differentiated data sets where the data you have access to is profoundly different at reasonable valuations that are not far from the normal valuations. Right now our pre-seed deals are like five or six but in AI they typically are 25.

I’m not gonna do more than 8. Our pre-seed deals in the US are like 10 to 12 pre, and it either could be 50. I’m not gonna more than like 15 or 10 to 12, I’m not gonna do more than 15. Or A deals right now are like 7 to 18 pre, eight at 20 pre, maybe 10 at 30 pre.

And in AI I’ve seen that like a hundred million. the median is way lower than the mean because of like second time founders AI, et cetera. I try to stay within the right valuation range and I am seeing deals. At the right price with differentiated data sets in verticals where there’s a willingness to pay,

The issue is there are many categories where there’s no willingness to pay. And giving me metrics like, users per day and DAUs, is not something I value. I value unit economics CAC retention and, contribution margin, we are seeing interesting deals.

We’re like Voca, which is helping doctor’s offices in Europe and France specifically, to digitize and connecting their electric medical records. We’re in a company that’s helping cities collect parking tickets. there’s a whole bunch of stuff in these vertical AI companies that we’re doing, but we invested in the right valuation and the right teams, et cetera.

As an angel, I would be very careful because you might be better off investing in the current round of Open AI at 300 billion valuation I think they’ve won. they’ll continue winning and there’ll be a multi-trillion dollar company in the future.

That’s probably a safer bet despite the price than investing at a hundred pre in a seed or A company in AI. Got it. Thank you.

Dhruv Sharma: I think this is a great moment to ask you a question about your anti VC strategy, right? Selling winners on the way up, especially against this backdrop of AI and forever private companies and permanent capital vehicles. another structural question around venture. We’d love your thoughts.

Fabrice Grinda: Venture is seeing a profound shift, right? Because there’s been no exits in 22, 23, 24, there’s been no DPI. So venture is in the midst of a transformation and recession. 2000 VCs have closed their doors in the last, few years in the US and we’re seeing a schism between. The winners that are in massive funds that, or I would not even describe them as venture funds anymore.

I would describe them as asset accumulators. The Andreessen Horowitz, the General Catalyst, they basically now billions. And they’re becoming evergreen. They’re saying once the company’s going public, their eyes, they’re holding the sock, et cetera, they’re no longer going to be giving three X returns.

They’re gonna be compounding at 10- 15%. which pension funds think are okay, but it’s no longer venture. And then on the other end, you’re gonna have specialized, like we are in a way, the marketplace- seed- asset light funds that are like three, four, 500 million that are specialized in different categories that because their expertise can do rather well.

And everyone in the middle I think is going to die. What’s interesting is many of these funds, the reason they’re dying is because the LPs feel overexposed to venture. They’ve had better returns by investing in Nasdaq in the S&P 500, and there’s no illiquidity premium. And our strategy, of selling or winner is on the way up, is actually, especially for an early stage fund, proven to be correct because the beauty of not leading, not having 25% of a company is twofold.

You don’t need to do your pro-rata. So if a company is doing okay and you own 25% of it, you have to do your pro-rata even if you don’t want to, because if you don’t do it, the company dies. It doesn’t get funded in the next round. VCs end up writing a lot of follow on checks in companies they would rather not but they have to.

The second thing is. I don’t believe in this. the winners do keep on winning, but you don’t invest at any price. I will evaluate every single round as though I’m not an existing investor, knowing what I know now of the team, of the traction of the category, of the valuation. Would I invest in this round?

And what happens is the winners often, I feel, are priced to perfection. They’re like at a hundred X, It’s gonna take years for them to grow in valuation. but they are the winners. there is demand for it, right? You can’t sell the dogs, right?

Like, so I don’t wanna mess on the dogs, but no one will buy them for me. But if the company is crushing it and I find is extremely. Priced over where it should be. We often sell 50% of the way up because there’s an opportunity as the VCs want more ownership. the founders don’t dislike it because they’re like, Greylock, Andreessen, and Sequoia all want in.

I want them, they all want 15%. I don’t want ’em to find a competitor, but I don’t want 45% dilution, so I’ll take 30% dilution. I need to find secondary, and you happen to own 2-3% as a favor, would you mind selling them? I’m like, look, I love you, but I’m happy to help out. And so we typically sell 50% when we feel the company’s doing well.

it is the reason we’ve added 30% realized IRR we have had exits in 22 23, 24 and 25. Most of our exits have come through secondaries, either through platforms like Forge, which are marketplaces for secondaries or on the way up, but you can only sell the winners.

We don’t always sell the winners because if the winners are priced correctly, we keep investing. It’s based on what I know now, the company at the traction of the terms do I think it’s overpriced. And if it’s a bit overpriced, we probably hold.

But if it’s massively overpriced, then we sell, in 21, whatever. Tiger and SoftBank were coming in, we basically had heuristic. If they’re coming in, we sell. it was a very good heuristic to have. Right now there’s actually a lot of stuff that we feel is priced correctly and we keep investing

The last strand was at a 5 billion valuation. if we had had more capital, we would’ve backed up the truck and wrote it like 10 or 20 or 50 million Now, I don’t have enough capital to do that. My fund is only, 290 million last fund. This fund is gonna be 300 million. We just first close.

But yes, selling on the way up, taking liquidity, you’re not gonna regret it. Selling 50% is, kind of the magic number because if it goes to zero, you’ve made 10 x, you’re happy. If it goes to the moon, you still have 50%, you’re happy. on average we sell 50%. Now, if it’s outrageously priced like a hundred X ARR, maybe we’ll sell 75%.

And if we’re on the fence, but if we sell 25%, we get three X and we’re still happy to run it, maybe we’ll sell 25%. But most of the cases when we do sell, we sell 50%. we typically don’t hold in the public markets, So we either sell into the IPO or we sell after the lockup. the reason we don’t is we’re not public market investors.

Right now I have proprietary access as CEO, I can talk to the CEO and, it gives me perspective. I could be helpful. We can have meaningful conversations.

We have propriety data, The minute it’s a public company, we’re a tiny investor amongst fidelity and whomever. We no longer have proprietor information. I’ve been compounding at 30% IRR in the private side. I think there’s zero way I’m gonna compound in 30% IR in the public side.

So I’m gonna keep reinvesting in the private side, so take liquidity and reinvest. I have a venture fund slash angel investing machine so I can do it. If you don’t have that, you probably shouldn’t do that.

Dhruv Sharma: Fabrice, Do we still have some time with you?

Fabrice Grinda: Yeah. I’m free for the next 25 minutes.

Dhruv Sharma: Alright. we can’t have you and not ask questions around marketplaces. Fabrice, you’ve solved so many different problems within marketplaces, global marketplaces, but, what do you think still remains unsolved or under solved at scale in the context of marketplaces?

Fabrice Grinda: So if you think of your consumer life, Marketplaces have basically transformed them for the better already, right? If you’re in the US you have DoorDash, you could get food in like 15 minutes. You have Instacart and you can, your grocery delivered an hour or two. You can go to Amazon, which is mostly a marketplace.

And you could get whatever you want in a day or two. you have Airbnb, you have booking. your life is pretty extraordinary. despite that, there’s still innovation in marketplaces. we’re starting to see peer-to-peer rental marketplaces like pickle come to the fore.

We’re starting to see live shopping marketplaces like PalmStreet for rare plants but if you go to the B2B world, we’re at day zero. We’re in the dark ages, if you want to buy inputs like petrochemicals. There’s not even a catalog of what’s available. There’s no connectivity to the factory to understand manufacturing delays and capacity. There’s no online ordering. There’s no online payment, there’s no tracking. There’s no financing. This needs to happen for every input geography and category. So whether you’re buying minerals or gravel or steel, but it’s also for machine parts and heavy machinery and you name it.

So all input categories. SMBs need to be digitized. And that’s not only true in places like India where all the little bodegas, et cetera, are all pen and paper. it’s also true in the US. We’ve been investing in digitization software for dry cleaners, people running, yoga classes and spas.

We’ve been running it for barbershops, for pizzerias, for Chinese restaurants, you name it. All of these are like literally sub 5%, typically sub 1% penetration. That’s true of, Creating the entire infrastructure to support these marketplaces. payment mechanism, shipping mechanism. Last month, picking and packing, automation companies were investors, a company called Formic, which basically is a marketplace to help companies, bring robots into their assembly lines.

So, recommerce is now coming to B2B, so we’re starting to see recommerce marketplaces for apparel, a company called Ghost, or for baby stuff, company called Rebel. we’re at the very beginning of marketplaces making a dent in B2B. And B2B of course, is way more of GDP in a way than the consumer world.

we’re talking about trillions Petrochemicals, 4 trillion, steel, 1 trillion, We’re at day zero. So everything needs to be done. It’s a massive mega trend that will last for the next two decades, basically. And I’m happy to ride on the coattails of it.

Dhruv Sharma: A related question to the role of AI in marketplaces of Fabrice.

This one’s a specific question. Amongst the smartest founders, you know, how are you seeing them figure out how to get better at answer engine optimization in this post-LLM era?

Fabrice Grinda: What’s interesting is many of the marketplaces were worried, that oh the top of funnel is going to be captured by the LLMs and they’re going to be disintermediating me etc.

And I do not worry about that at all. because if you think about. The way people behave and buy on marketplaces is threefold. Either they don’t know what they’re looking for and they like to browse and shop. it’s like shopping as entertainment, which is like OLX or Vinted, in which case you don’t want, an LLM to give you a better funnel,

Because the purpose, is actually seeing a feed. you typically look at low value items and find something you like, you buy it. The conversion rate of visit, it buys low. So, there’s no role for an LLM there. Number two, search. So aside like eBay or frankly Amazon, you know exactly what you’re looking for.

You put in search engine, boom, you’re done. There’s no reason to go to an LLM just as an extra click. And number three though, consider purchases. So something where you don’t know exactly what you’re looking for, or it’s something where you are gonna take your time because it’s a high value item.

So buying real estate, buying a car, or maybe buying specialty equipment an LLM can play a role, but even then it’s not obvious that the winning LLM is GPT versus actually a specialized site like Carvana that has an LLM to help you make decisions on what to buy. Like on Amazon, there’s Rufus, which is pretty good at giving you advice on what to buy.

So, I wouldn’t worry about the LLMs disrupting the top of funnel. So I would index myself. I would do AEO. And there are companies like Graphite HQ that are very good at AEO. And by the way, the way I think about it is if you’re willing to index yourself in Google. You should be willing to index yourself in the LLMs.

And if you have a 99% market share in your category, then don’t index yourself in Google because you don’t want this. The behavior people go there, in order to search. But even then, like if you have a massive dominant share, even if people search in Google or the LLMs, the results will come from your site, so you’re still winning because you’re doing the fulfillment, you’re doing the payment.

And these guys will never do that. That’s not, it’s not at the top hundred things that GPT wants to do, and then in the next decade. They’re not thinking of that. so even if they capture top of funnel because you have liquidity, because you have fulfillment payment, et cetera, you’re still gonna capture all the value.

You know, if you go to Google today in the US and you type A whatever, LGC 4.0 lead TV 65 inch. All the results are basically Amazon and eBay, they have like 80% market share, so they’re getting all the value, even if there’s a little bit of revenue there. So I would do AEO, there’s a company called Graphite HQ that’s very good at helping with AEO.

Now there are things beyond AEO that I would do, you know, get the free traffic. But beyond that, that you could use AI for the following things. One, if you want to be cross-border, with ai you can actually translate the listing. So within India, for instance, because there are different languages I could imagine translating a hundred percent of the listings across all the languages.

you translate the listing and then you can auto translate the conversations between the users. So right now in Vinted in Europe before Vinted Europe was not Europe. It was a French site, a German site, a British site, a Polish site, a Lithuanian site. Vinted actually translate the listings, between all the languages.

The conversations with consumers are also translated, and they have a great embedded payment and shipping. Now, it only works because they have embedded payment shipping, but as a result, they’ve now become a true pan-European company. So I can imagine having a true, true Penn Indian company with automatic, translated all the listings, conversations, et cetera, as long as you figured out a way to do payment and shipping.

cross border is becoming a thing because of AI. We have a company called Ovoko doing it for car parts, CarOnSale, doing for B2B cars. Number two, I would use AI to massively improve your cell funnel. The old way of listing things on marketplaces is you take 10 photos, you pick a category, you write a title, you write a description, you select a price.

But the reality is you as the marketplace, you have the data. You should be able to recognize what the item is. You should be able to pick the category, write the description, write the title, even consider the condition and suggest the price. And so you should make it so easy, that your visit to listing rate should improve dramatically.

I would improve the search engine. in the long term, the way Amazon has done it, which is two search engines, Rufus to ask questions about products and one search engine for search results doesn’t make sense. You’re gonna have only one search engine where you can ask anything, whether it’s like, oh, I wanna buy whatever the LG C 4.0 LED tv, or, what TV should I buy?

it could be one search box for both LLM type answers. and product type answers. I would use AI to improve program and productivity. Everyone should be using GitHub or Copilot or Cursor. AI for customer care to decree to do all the basic stuff like whatever returns, payment issues.

I mean, AI could be so profoundly good and everyone should be using AI and improve your sales teams. There are tools out there that could bring your average salesperson productivity to 90% of your best salesperson productivity. So, I would use it for that. And then on top of that, I would do AEO, in order, in order to get access to free traffic.

Dhruv Sharma: what my final question for you, Fabrice. How are you able to do so much? And, you said you tell us how you outsource ruthlessly, and which side quests are you most excited about right now?

Fabrice Grinda: So, I guess two different questions. I like to be intellectually simulated and I outsource all the things that I find repetitive and not interesting. This way I like to be the creative director if you want. I like to play with product. I wanted to be doing the Figma and writing the user stories, et cetera.

Here at FJ Labs, I want to talk to the most fascinating founders. both the new ones that I would like to invest in and the ones that I’ve invested in to help them. And then daydream about what are the trends, what should be happening, what’s happening in AI, write blog posts, et cetera.

The reason I have FJ Labs, the reason I have a VC fund is the team allows me to, to only focus on things I’d like to do. So FJ Labs, we have. Nine investors total who filter the 300 deals. I can talk to five. that’s why I have a VC fund, by the way. But things I do, so I have virtual assistants in the Philippines.

I have two of them. And I outsource ruthlessly, everything from like thinking through, I. Paying all my bills, coming up with ideas for days, new interesting shows to go see. managing, of course, all my agenda. You can outsource a lot more than you think you can. I have an offline butler that takes care of all the offline things.

Now, of course, a butler is a bit more expensive, though, probably not in India, but a virtual in the Philippines full time is like 1500 a month. I like creating for my parents a photo album of all my adventures of the year before. I have a photo album creator that I hired on Upwork, who’s in Bangladesh.

I’m actually a pretty good video editor myself, but I’ve someone to help me and video editing that I found, also an Upwork to, to help with the video editing. Even the childcare, I have two kids. a 4-year-old and a 1-year-old. I found on care.com five nannies that I hire.

I pay them hourly and create an online Google Docs where I write the coverage I want, and they agree between themselves who takes what shifts. when I travel, one, takes one week, the other takes one week, et cetera. they fully self-manage and regulate the shifts they take because of course the French nannies only wanna work 20 hours a week max, so that’s why I need five.

But you can out all the things you don’t like to do. Like I don’t like to cook, so I have a chef You can outsource everything in life to focus on the things you’d love to do. Now, the side quests, in my case and the things I love to do, as I said, it’s a combination of like meeting these sort founders, writing, thinking.

But frankly, it’s like spending time with my kids and I love to play. And I play tennis. I play paddle, I kite surf, I go back country skiing. Every year I go in a crazy off-grid adventure where, you know, I walked to the South Pole, two years ago where I was pulling my a hundred pound flood in negative 50 weather, which is kind of like an active Vipassana retreat because you’re alone with your thoughts walking eight hours a day, being off grid in this hyperconnected world is such a true privilege, where you can be reflective and thoughtful.

So yeah, continue to explore, what means for me to be spiritual, to be a good person I read an hour, an hour and a half every day before bed. So I read like a hundred books every year. Currently I’m rereading Meditations by Marcus Aurelius, which I highly recommend, to everyone out there.

I have a lot of side quests hosting my salons, and it’s fun. The purpose of life for me is to live a rich life. to find things that bring you purpose, meaning, to communicate, to be social and to be entertained. even work for me is entertaining.

[01:03:05] Dhruv Sharma: Fabrice, this has been great. Thank you again so much for your time. This is, This is really fuel, I think for all of us who attended. This is Fuel to go by for the next few months. Such a pleasure. Have a great rest of the day.

Fabrice Grinda: Thank you. I have a few more minutes, so if Snehil wants to ask this question, I see his hands raised.

Right. That’s fine.

Snehil Khanor: No questions. Just wanted to thank you. I don’t know if you remember, we have been connected on Facebook for more than 12 years now.

Utsav Somani: Oh, wow.

Snehil Khanor: I think since 2012 when I was just starting entrepreneurship, there was hardly any resource or network to learn from other founders.

You have always been kind. I think we have interacted three, four times on Messenger, over the last decade. And, you have always been kind to reply and guide, so thanks a lot for that.

Fabrice Grinda: I mean, it is words like these that are the reason I’d like to share, that I like to write that I’d like to, put out there, what I’ve learned.

I think the reason I have my blog every AI is, sharing what I wish I knew when I was starting as a 23-year-old entrepreneur, and founder that I now know.

Snehil Khanor: Thanks. I’m from that, 22 year, 23-year-old also.

Fabrice Grinda: Thank you everyone. This was super fun. feel free to follow up and I look forward, continuing the conversation.

Clutch的发展历程令人难以置信。作为 “加拿大的 Carvana”，它在 2021 年一飞冲天。2023 年，它与死亡擦肩而过，并实现了不可思议的逆转。现在，他们再次一飞冲天。Clutch 创始人兼首席执行官丹-帕克（Dan Park）与我们分享了一路走来的经验教训。

如果您愿意，可以通过嵌入式播客播放器收听本期节目。

除了上述 YouTube 视频和嵌入式播客播放器，您还可以在iTunes和Spotify 上收听播客。

文字稿

法布里斯-格林达：大家好。我希望你们度过了愉快的一周。我非常激动，因为本周我们将讨论 Clutch 这个令人难以置信的故事，有点像加拿大的 Carvana。我曾与死神擦肩而过，然后我有一个非凡的复出，他们现在再次泪流满面。

在不同的情况下该怎么做，我们都可以从中学到很多经验，我很期待这样做。好了，废话不多说，我们开始吧。欢迎收看第 49 集。离合器和不可思议的复出故事

谢谢你的参与。

丹-帕克谢谢你邀请我。这太棒了。

法布里斯-格林达：那么，我们为什么不从你的背景说起呢？

丹-帕克所以我想，也许像大多数创业者一样，他们会把汽车销售作为尝试做大事的媒介。

但我最初的职业生涯是从金融开始的。实际上，我在风险投资或阶段性投资方面做了几年，我很喜欢这些工作。我觉得在金融的不同领域中，风险投资可能是最接近拉动杠杆的行业，但你并不在其中。当然，日复一日，我在 2016 年得到了加入 Uber 的机会。当时，Uber 正在尝试汽车以外的一些不同的商业理念。他们创建了一个名为 “Uber，一切 “的小组，具有讽刺意味的是。

不是所有的东西，而是所有不属于合乘的东西。所以可能什么都不是。但当时特拉维斯还在那里，特拉维斯问了一个问题，如果你能在五分钟内把车开到别人那里，那你能在五分钟内给别人带来什么？于是我就做了，这是第一次，第一次围绕食物做实验。

我们开始看到一些需求信号，人们需要按需提供食物。这有点像食品外卖的早期。你知道，DoorDash 当时的势头很猛。在欧洲，世界各地还有其他的外卖模式，而 Uber Eats 则刚刚起步，实际上在全球范围内，它起步于多伦多，也就是我所在的地方。

因此，我很高兴能参与其中。从很早开始，我们的第一次迭代甚至不是一个市场。只是一群人坐上 Uber，或者不好意思，去买吃的，在停车场碰面，往 Uber 里塞 25 个三明治。那时候还没有为 “吃 “这个产品单独设计一个应用程序。

这只是优步的应用程序。在某一天，有汉堡，有泰式炒饭，有三明治，你可以选择你想要的，然后送货上门。但库存计划是一场灾难，因为一个司机的 25 个三明治会在一小时内用完，而另一个司机四小时后仍会带着 50 个泰式炒饭到处跑。

但这是按需送餐的早期信号，我们从那时起就开始收购餐馆。我们建立了现在这个版本的优步美食应用。这是在当地的一间会议室里完成的，历时三个月。超级原始的 MVP，但我们把它做出来了。

法布里斯-格林达：那段时间你在多伦多工作？

丹-帕克总部设在多伦多，Uber Eats 产品在多伦多开始全球业务。所以我们是启动城市。

法布里斯-格林达：哇哦。太神奇了

丹-帕克我认为，如果 Uber 把事情搞砸了，就会有人说，哦，这是加拿大做的事情之一，你知道的，把它扫到地毯下面去。

这是一个错误。但它是成功的。我的损益表在三年内从零变成了 30 亿。至少在 GMV 方面是这样。你知道，我们学到了很多关于市场的知识。三方市场，超级复杂的驱动因素 我们称之为食客、司机和餐馆

这很难，因为在这种市场环境下 很难让任何人满意。因为司机想要更多的报酬，餐馆想要更少的报酬，而顾客真的不愿意为外卖支付超过5到6美元的费用。因此，在市场平衡、定价和外卖价格方面，穿针引线的工作极具挑战性。

有一次，我想我们一年烧掉了4亿左右。特拉维斯来了，说你们是什么东西？我想我们在纽约一趟要损失17美元 你知道，基本上就像这是谁的生意，这生意能行吗？必须有人去证明单位经济效益

因此，在短短八周内，亚特兰大市场和我们自己。多伦多市场基本上也是如此，因为我们是最早的两个市场，必须想办法在很短的时间内实现单位经济效益。我们做到了，然后又进入了增长模式。

法布里斯-格林达：太棒了。

丹-帕克是的。然后你知道，是的，我在那里待了大约三年多，Clutch 公司的一位董事会成员找到我说，嘿，你看，有一家处于种子期的公司。他们在网上卖车，在网上送车。

如果你在网上送餐，你应该这样做。我对加拿大二手车市场的空白感到非常兴奋。我认为，这是最后一个尚未被电子商务完全颠覆的真正的电子商务品类，或者说，是尚未被电子商务完全颠覆的零售品类。在加拿大，不像美国有一些较大的品牌，比如 CarMax，在美国，加拿大没有真正的二手车零售品牌。

这是一项复杂的交易，不透明的交易，缺乏透明度的交易。我们觉得有更好的办法。因此，我们现在已经在加拿大建立了一个全面管理的二手车市场。

法布里斯-格林达：那么，你是什么时候，什么时候在种子队之前加入 Clutch 的？

丹-帕克是的，那是 2019 年。我认为，如果你回顾过去，我不知道还有没有比这更糟糕的五年了。考虑到大流行病，考虑到现在的关税，考虑到 2022 年和 2023 年成长型股权资本的蒸发，早期阶段的二手车业务都是如此。

但我们挺过来了。就像你在节目一开始说的，我们从中学到了很多。

法布里斯-格林达：那么在 2019 年，Carvana 的状态如何？

丹-帕克是的。他们起步较早，有一些独特的动力。我认为这是一个证明点，证明这种模式可能行得通。但你知道，与此同时，你也很了解这个故事，我是说，过去也有 BP，我是说，人们，这不是第一次尝试。所以我认为

人们可能更倾向于在网上交易。自 bp 时代以来，电子商务显然有了长足的发展，人们更愿意在网上购买大件商品。因此，我认为消费者已经发生了变化，但在 Carvana 耗资数十亿美元的时候，商业模式肯定还没有得到验证。因此，这项业务仍然存在很多问号，但我们知道这一点，而且我们坚信，一定有更好的方法让人们享受到更简单的二手车交易体验。

法布里斯-格林达：你为什么不带我们了解一下离合器式二手车购买体验，以及为什么它比线下体验要好得多。

丹-帕克是啊。你知道，传统的购车体验是，你走进一家汽车经销商，花上四五个小时与销售人员谈判。他们试图把一堆金融服务产品塞进你的喉咙。要知道，在某些市场。如果你进入一个汽车广告网站，他们甚至不会告诉你汽车的价格。

他们只会说联系我们了解价格，这就意味着当你来拜访我时，我会根据你的谈判能力或我认为你是个笨蛋的程度来决定能收你多少钱，对吧。所以，你知道，这是非常可怕的。我认为，定价的透明度非常低，而通过技术，我们可以谈谈我们如何在车辆定价方面应用相当深入的技术。

肯定有办法提供更透明的体验。我们就是这样做的。因为我们直接从消费者手中买车。因此，我们在市场上推出了即时现金报价，客户来找我们，提供一点关于他们车辆的信息，我们就会给他们的车一个即时现金报价。

我们每天购买价值约 200 万美元的汽车。然后，我们将这些汽车送入一个完全垂直整合的翻新流程，这就是管理市场要素发挥作用的地方，对吗？我们承担库存风险，提供更好、更高质量的产品。

因此，我们对这些车进行翻新，把那些出于某种原因别人不想要的二手车翻新成一辆非常受欢迎的车。我们把它放到网站上，提供一种无缝、简便的购车方式。几乎就像，你知道，我们试图让买车就像订披萨一样简单。

这有点像我送外卖的日子。但我们想让它变得超级简单，我们也围绕这一点建立了物流，因为现在还没有真正高质量的三家 PL 可以，你知道，我们可以把一辆车，你知道，直接运送给消费者。没有任何东西可以做到这一点。

因此，我们必须建立自己的原木物流，以及与之相关的全套基础设施，我认为这为我们创造了很多优势和竞争优势，因为在这一点上，仅凭交付产品所需的基础设施的物理特性，真的很难颠覆。

法布里斯-格林达：让我总结一下我听到的。如果你要卖一辆车，你拍几张照片，做几句描述，马上就有人出价，然后 “砰 “的一声，车就卖出去了。车就卖出去了。所以这超级简单！

丹-帕克我们甚至都不拍照，我们甚至都不要求拍照，因为我们是按比例拍摄的。

法布里斯-格林达：好的，所以只需要提供车辆识别码或其他信息。然后，买车的人有一个固定的价格，他们点击购买，然后车就会直接送到他们家。

丹-帕克或者在我们的某个地点领取。

法布里斯-格林达：好的。好极了好的你在使用库存，你在买车，你在翻新它们，等等。

你们是否考虑过一种模式，即只有在售出后才有可能进行翻新？这样你就不需要库存，或者你觉得库存是必需的？

丹-帕克我们早期尝试过寄售模式，但寄售模式给消费者带来的价值很小。

你知道，你必须说服别人，我要把你的车开走一段时间，对吧？如果你还不知道车卖出去了没有，那你就没办法了，但是，你知道，如果车真的卖出去了，我们会做一些工作的。但请相信我们，让我们玩玩这个等待游戏，看看会发生什么，你知道，对我们来说，现在，因为我们可以提供的报价，我们可以提供即时的确定性，你知道，如果有人要出国，这是一个非常大的用例，你要出国或你要卖掉你的车，你会把车开到我们这里，在去机场的路上把车放下，对不对？

它允许那种确定性 而如果你是那种通过。或尝试在 eBay 或 Craigslist 或其他网站上挂牌，则需要多人过来谈判。有些人是认真的，有些人不是。

所以很麻烦。所以现在个人卖车的传统方式超级复杂。所以我认为我们真的解决了这个问题。现在，我们手机产品的 NPS 大概是 85。如果不盘点库存，就很难创造出这样的体验。

我认为，我们可以继续讨论其中的一些问题。但是，早在 2020 年，早在 2019 年，仍有两种观点围绕着 “盘点 “还是 “不盘点”？当时的答案还不是很明确。而现在，我认为答案更清晰了，因为在过去五年里，能够存活下来的都是那些建立了库存基础设施的公司，这样他们就可以，一是提供更高质量的产品，二是控制整个供应链或基本上整个流程。

我不一定要对埃隆发表政治评论，但在他的书中有个概念叫白痴指数，对吧？我不知道你是否读过这本书，但白痴指数是指提供的商品的价格与投入的成本之比。

如果这个比例超高。要么是流程中存在缺陷，要么是有些事情可以做得更有效率。在汽车翻新方面，你把一辆车开到某个车行去修理，成本会非常昂贵。换个空气滤清器也就 60 美元，对吧？

更换保险杠、油漆、塑料和金属，再加上一点人工费，根据汽车的不同，费用高达 2,000 至 4,000 美元。我们所做的就是大规模翻新，这样我们就能以相对较低的成本在汽车上做大量的工作。

这对消费者来说具有非常非常高的价值，使我们能够在购买车辆时提供更高的价格，然后以相同或更低的价格向消费者提供更好的汽车。

法布里斯-格林达：你会接受任何一辆车吗，因为很明显，如果你的存货在贬值，而你又卖不出去，那就有问题了。

丹-帕克是的所以，我们几乎每辆车都要卖，一半是零售，一半是批发。因此，一半的业务是不符合我们零售标准的汽车，我们不会对其进行翻新，而是进行批发，由其他经销商或其他拍卖行购买。

法布里斯-格林达：我们有几个问题。我们正在做这件事。你们是在加拿大任何地方送货上门，不论地点，还是只在某些地方送货上门。

丹-帕克是的。今天，我们在地理位置上进行了限制，最终我们将向所有地方送货。但基本上会是区域化的库存集群，因为到了一定程度，在全国范围内移动库存就没有意义了。

如果你的选择达到一定程度，你知道，如果你的库存中有 50 辆本田思域，你是否需要从不同的市场获得额外的 20 辆。人们会怎么做，这是一个优化问题，他们会选择可能便宜20美元的本田思域，但它在全国的另一端。

而且，也许差别不大。因此，根据地域限制库存也能提高物流能力。

法布里斯-格林达：有道理，还有一个问题。你们是什么时候筹集资金的？是一开始就筹集资金，还是在意识到发展需要资金时才筹集？或者说是怎么做到的？

丹-帕克是的，我们遇到了一个有趣的鸡生蛋、蛋生鸡的问题。购买存货的成本高得惊人。每次我们购买一项资产，一件存货就是 2 万到 3 万美元不等，我们利用债务融资来解决这个问题，但在早期，你没有太多的历史记录。你在烧钱。所以贷款人不愿意贷款给你。然后是股权供应商会问，谁会为你的库存提供资金？在你找到可以为你的存货提供资金的人之前，他们是不会提供资金的。所以在早期，这是一个真正的冷启动问题，因为你需要库存来扩大市场规模。

因此，我的意思是，早期我们的债务融资利率非常高。我为库存提供了个人担保作为创始人，我和史蒂夫都为存货提供了个人担保。所以我们很早就不得不筹集资金 来支撑资产负债表。

这是一个资本密集度极高的行业，我不知道这是好事还是坏事，我们可以讨论一下，但对我们来说，我们必须尽早行动。因此，我们目前已经经历了多轮融资，今天我们筹集了大约 2 亿加元。

法布里斯-格林达：对于需要资金的公司来说，你所描述的是一个相当标准的传送带。你先从超级昂贵的家族办公室开始，以每年 18% 或 20% 的价格获得对冲基金的资金。这是非常昂贵的，但你别无选择。

然后渐渐地，你会得到更好的条件，你会得到像二线银行一样的条件。最后，你就能获得花旗银行或汇丰银行的 A 级信用额度了。现在，大概是 78%。

丹-帕克有了安全保障，你就能得到一个黄金分割点、

法布里斯-格林达：好的。哦，甚至更好。Oh, even better.

丹-帕克是的我们一开始的利率是十几美元。对于担保融资，我们最终降低到更合理的水平，也就是现在的最优惠利率。

法布里斯-格林达：那我们为什么不谈谈高飞的岁月呢？比如说，在某些时候，你筹集到了一轮惊人的资金，一切看起来都进入了正轨。

所以，我们要讨论的是，当时发生了什么，如何出错的，然后你是否还会再犯。

丹-帕克我不知道你会不会提，也不知道我会不会提，但我的意思是，我仍然清楚地记得 2020 年的那次谈话，关于 “在扩大规模之前先钉钉子”。

法布里斯-格林达：是的。

丹-帕克其实我经常想这个问题。我在想，我们是否应该更好地遵循这些建议，或者，我们是否仍然做出了正确的决定。我觉得回首往事很难，我觉得我们所做的很多事情都是当时的环境使然。

当时的资本真的很便宜，也很容易获得。我认为这种模式在很长一段时间内都行之有效。Uber 就是一个很好的例子，因为有了廉价的资金，它才得以存在。因此，你知道，在当时，我们有一个竞争对手，这在很大程度上是一种赢家通吃，甚至是赢家通吃的心态。

这对我们的很多决策都起到了推动作用。我认为，当我们还在优步的时候，我们输给了竞争对手很多地面战，因为我们不够早，当我们最终推出这些城市的时候，我们仍然需要花费大量的时间和金钱，甚至，你知道，在他们已经建立起来的市场份额上，我们还得有所作为。因此，当时我的很多想法都是这样的，尤其是面对咄咄逼人的竞争对手。我当时的想法是，我过去也做过这样的事情，你先扩大规模，然后再考虑单位经济效益，因为我们知道汽车销售业务利润丰厚。因此，一旦达到一定规模，就没有理由不能实现相同甚至更好的单位经济效益。现在，卡蒙特公司在公开市场上已经实现了这一点。他们的财务状况比 CarMax 等公司更具吸引力。但事实并非总是如此。因此

法布里斯-格林达：所以，你觉得这是在抢占市场，而在抢占市场的过程中，你希望花钱抢占市场和市场份额。有趣的是，这种心态实际上扼杀了 Beepy。

丹-帕克是的

法布里斯-格林达：但有些时候，如果你走得不够快，赢家通吃或占据大部分市场，你就会死。

因此，博弈论的决策树通常是你花钱，因为他们认为，如果你不花钱，他们花钱，他们就赢了。如果你们都花钱，在某种程度上你们都会输，但这是纳什均衡。

丹-帕克是啊。如果你不花钱，你就不会发展。如果不发展，就得不到资本。如果你得不到资本，你就不会发展。因此，我认为，只有在低利率、低成本的环境下，这种做法才会奏效。低利率，像廉价的资本环境。你知道，在我们开始之前，你问我，你知道，你会接受那一轮一亿美元的投资吗？我的意思是，我认为从开始， 像第一次介绍 到关闭不到四个星期。

超级简单。

法布里斯-格林达：就是那些日子。

丹-帕克那还不是最高估值。我们的估值高出 30%，我还记得当时在电话里和投资者交谈，说服他为什么估值太高，对吧？就好像是反过来一样。出于种种原因，我们接受了一个较低的估值，事后看来，这也许是个正确的决定。

但是，它就在那里，很容易。有人出价一亿美元让你去建公司 你很难拒绝而且它的估值很高 所以我们都很兴奋

法布里斯-格林达：那么我们来具体说说。是在 9 月、10 月还是 11 月 21 日？

丹-帕克那是2021年底是的，没错。

法布里斯-格林达：好的，那么你们筹集了多少资金？

丹-帕克对一些早期投资者（而不是员工）来说，75%是一级市场，25%是二级市场。职位是 5-75 个。

法布里斯-格林达：好的。之后呢？

丹-帕克因为在第一次董事会会议上，我就开始行动了。我拿到了一本叫《七种力量》的书，这是一本很棒的书。这本书讲的基本上是竞争护城河和可防御性。我们开始建设，对吧。我们开始建设的时候，可能远远超过了我们的增长速度，但这就是我们需要做的。

预计，你知道，基本上是什么 像200％的增长率。那是第一次董事会，然后我说，嘿，看，这是你的，你有一百人，好像还有一百人。就这样，你知道，走，走规模。第二次董事会，哇，哇，哇。

法布里斯-格林达：嘿，顺便说一下，首先，你们的规模从多少扩大到了多少，就像使用资本一样。

丹-帕克我们在一年内从 8700 万增长到 2 亿左右。

法布里斯-格林达：是的，我是说 2 亿。这已经是相当大的规模了。

丹-帕克是的，下一年预计会有五六亿。所以，就像是纯粹的加速，但在下一次董事会会议上，也就是 2023 年初，裂缝开始显现。公开市场和所有的电子商务股票 如果你还记得的话 下跌了95%人们都在挖战壕。我想每个风险投资人都在那个夏天逃到了阿斯彭或者意大利。那个夏天死气沉沉没人愿意接我的电话我们需要资金 我们以为我们能筹集到资金 因为情况很糟 但其实没那么糟

到了秋天，情况开始变得非常糟糕。而且，你知道，基本上，如果没有足够的，

法布里斯-格林达：你一个月烧掉多少钱？

丹-帕克我们每个月要烧掉 500 万美元。

法布里斯-格林达：每月 500 万。好的

丹-帕克所以，我们建立了所有这些基础设施，以扩大规模。

因此，我们需要继续这样做，这有点像仓鼠轮。奇迹出现了，我们在 2023 年 10 月拿到了 9500 万美元的 C 轮融资合同，这让我们继续沿着这条轨迹前进。但在我们的内部，情况是怎样的？四五周后，在我们本该完成交易之前，领投人打电话给我们说，嘿，看来我们对这笔交易没感觉了。我们退出所以那是最后一根稻草，我认为那让我们走上了一条真正健康的道路。我的意思是，当时我在。我是在一个家庭度假 并没有地方接听电话， 所以我在浴室里。

我记得很清楚当时我们在浴室里，我接到电话说：”嘿，听着，很抱歉，我们不会再这样做了。我立刻和我的执行团队通了电话 我们得想想办法现在纯粹是生存模式

所以我们去了，那天我记得很清楚 是2023年1月4日我们完全进入了生存模式就这样，我猜是在13天内。1月17日，我们进行了一次大裁员，裁掉了65%的员工，非常艰难。

法布里斯-格林达：那么，从多少人到多少人？

丹-帕克从巅峰时期的 350 降至 87。

法布里斯-格林达：哇。几周后基本上是这样

丹-帕克一天之内

法布里斯-格林达：一天之内。

丹-帕克不好意思，我们在 2023 年做过一轮，当时大家都在做 2023。那是健康的一轮。那是一轮减脂。然后我们又做了一轮然后，我们做了一个真正的深，深切割。我们只裁了两个人

法布里斯-格林达：你一共说了多少，65%还是80%以上？

丹-帕克第二段是 65。从 350 到 87。

法布里斯-格林达：好的。

丹-帕克削减了三个市场。所以，我们当时有五个市场，你知道，削减到了两个核心市场，然后，当时最重要的就是盈利能力。生存和盈利。围绕我们的利润率，制定了一些非常明确的目标。

我们为此付出了巨大的努力，在接下来的 12 个月里，我猜是 16 个月，我们的净收入达到了正数，这真正改变了我们的游戏规则。

法布里斯-格林达：让我们谈两件事。一是收入下降了多少？

丹-帕克是的第二年，我们只减少到 1 75 家。是的，所以我们，我们没有，我们，我们削减了，我的意思是，在我们早期的市场上，有不成比例的燃烧量。对

法布里斯-格林达：好的。

丹-帕克所以我们刚刚启动了这些市场。它们还没有机会扩大规模。它们还没来得及成熟。因此，这些市场的火热程度不成比例，幸运的是，这再次让我们的决策变得更加容易。但我们只是，我们只是 启动这些市场。但我们只是启动了这些市场。

法布里斯-格林达：当你解雇了 75% 的员工，却只缩减了 10% 的规模时，你会觉得非常疯狂。

丹-帕克是啊。我们还没开始进入某些市场，就被人打断了腿。

法布里斯-格林达：是的，值得一谈的是你是如何生存下来的，因为你需要资金，你需要现金。你没有现金了所以你通常会解雇所有人，但你也没有现金了。所以我们必须彻底调整公司的资本结构这就彻底改变了上限结构。

那就跟我们说说吧。

丹-帕克所以我们在 2023 年夏天做了这件事。所以我们做了一轮惩罚性的下调，基本上是把资金投入到业务中，清理上限表。在这一点上，有九家机构投资者。我认为，这九家投资者中的七家都会比在首轮融资前处于更有利的位置。至于那两家不同意的，我想他们是理解的，也觉得自己受到了公平对待。我和他们仍然保持着很好的关系，每个季度都会通电话。但对很多人来说，这确实是一个重要的时刻。

法布里斯-格林达：是的。所以，如果，如果我没弄错的话，如果你愿意分享的话，我想，基本上是早期投资者。站了出来。所以就像，黄金和FJ实验室

丹-帕克卡南也大显身手。

法布里斯-格林达：没错。我们爱这家公司，我们爱丹。我们希望它能生存下去。因为资金越多越好，但我们正试图让它实现盈利，需要多少钱呢？我想如果我没记错的话，我们做了 500 万的投前估值。

Dan Park:15、

法布里斯-格林达：我们从 5 75 个创始人那里掠夺了资金，并筹集了 1 500 万或 2 000 万，我不记得了，这是获得盈利所需的金额。

丹-帕克是的我们做到了。这就是资本的目标。我们实现了净收入为正。之后很快就实现了。这项业务的魅力在于，一旦你建立了基础设施，我们的扩展能力，我不想说它很容易，但我的意思是，有一个玩法。

对

法布里斯-格林达：你们在定价方面有什么改变吗？比如你买车的价格低了吗？有没有提高售价，增加利润？

丹-帕克不，我们所做的最根本的事情是改变了我们的供应采购渠道，所以我们大部分都是从拍卖会上批发采购，如果你仔细想想，这里面有几层成本，还有一些人把手伸进口袋里拿走利润。因此，我们所谓的前端利润，也就是我们在汽车上的实际利润，就变得非常微薄。因此，前端利润和直接面向消费者的利润。

并获得他们的汽车。其实你可以出更高的价格，还可以省去很多这些费用。这也是一辆更好的车。这样就能以更好的价格买到更好的车。因此，我们就能获得这些汽车，并在此基础上获得利润。然后，我们还能提供更好的后端项目产品，实际上我们销售的产品，比如我们没有花很多时间销售金融服务产品。比如向消费者提供保险和保修服务，因为那时候我们只关注规模和速度。

法布里斯-格林达：好的。还有融资

丹-帕克表面上，我们是汽车销售商和零售商。在背后，我想说我们有一半的业务是金融服务。

法布里斯-格林达：好的，你是什么时候实现盈利的？之后的发展又是怎样的？

丹-帕克2024 年 7 月。

法布里斯-格林达：是的。对我来说，我职业生涯中最难忘的时刻其实不是我卖掉 OLX 或 Zingy 或其他东西的那一天。

实际上，我的第二家初创公司就是在那天实现现金流的。在我连续四个月没发工资后，我欠下了十万美元的债务，等等。我当时想，你是自己命运的主宰。现在的未来，是的，你很安全，你还活着。

丹-帕克我是说，这是一种神奇的转变。

你知道，作为创始人和首席执行官，世界的重量，有点像，我不会说解除，但肯定是减轻了，因为。要知道，当你每月烧掉五六百万美元时，每天就是 20 万美元。你知道，你一觉醒来，你会说，告诉你自己，这个企业今天会在某个地方增加20万美元的价值吗？

而当时间站在你这边时，当你开始赚钱时，这种感觉就会很好。

法布里斯-格林达：那么请告诉我们现在的情况。

丹-帕克现在，我是说我们已经证明了盈利能力。我们打算继续在收支平衡的情况下开展业务。

你知道，我们在大量烧钱方面有很多疤痕组织。因此，我们非常清醒地认识到，我们的增长目标要略微超前，而不是超前两三年。我认为，如果你真的想一想，我们当年可能犯的一个错误就是，我认为我们把目标定得太远了，比我们的需求扩展得太快了。

因此，现在我们在增长方面更加深思熟虑，更加有条不紊，我认为这是健康的，而且我们的资本效率更高。我们启动市场的方式更加节约资金，我们处理和翻新汽车的方式也更加高效。

我们所做的每一件事都能提高资本效率，因为我们不得不这样做。我们有一个强制功能，迫使我们这样做。但今天我们坐在这里，我认为我们在扩大规模方面的处境要健康得多。在这一点上，你知道，即使把那些下滑的年份考虑在内，我们今天所处的位置，从五年前算起，我们的业务年复合增长率达到了 100%。

因此，增长依然存在，但现在的盈利能力更强了。

Fabrice Grinda：你提到或分享了你现在的规模？

丹-帕克是的，我们现在的收入已经接近 5 亿美元。

法布里斯-格林达：5 亿。好吧，你已经过了高峰期？

丹-帕克是的。

远远超过顶峰

法布里斯-格林达：好的。太不可思议了。我想筹资环境可能正在改善，那么，你认为这个市场会有多大，比如加拿大的二手车市场有多大？这是一家十亿美元的公司吗？

比如这个能有多大？

丹-帕克我认为加拿大市场的独特之处在于，你可以获得更高的相对市场份额，因为它的竞争不如美国激烈。它更像是一个 “赢家通吃 “的市场，甚至可以说是一个 “赢家通吃 “的市场。也许我太天真了，但我认为这里是天空的极限。

你知道，我认为如果你把它代入 Carvana。你知道，10% 的加拿大市场与美国市场相比，我们的终端退出价值将达到 50 亿美元。但是，因为我们可以在以后的购车过程中，特别是在金融服务方面，有更多其他的内容，所以我们的野心很大。

我坚信，这可能是我工作的唯一一家公司，也可能是我工作的最后一家公司。

法布里斯-格林达：听着，如果我不需要出售 OLX，我就不会出售。就像我不想卖一样。一旦你在一个平台上有了规模，你就可以不断扩大，做得越来越多。

你已经有了一定的实力和地位，当你有了规模之后，进行多元测试就容易多了。当你启动的时候，风险投资公司就像AB测试一样，N等于2，AB测试并不意味着什么。这就像是随机的。当你有数百万和数百万的收入，那么它实际上真的是可扩展的，可持续的，它是一个惊人的平台，不断扩大。

所以，我完全可以这样做。我同意，如果你有一个了不起的平台，就继续坚持下去。

丹-帕克是啊。Yeah.这就是我们的目标。我们会看到未来会发生什么。我已经试着停止预测事情，因为在过去的几年里，我已经抛出了无数次曲线球。

所以，但我认为，今天的基础是强大的，团队是优秀的，业务是可行的。

法布里斯-格林达：我们有一个问题，在这一轮下跌中，作为创始人，你是如何避免被完全消灭的？你如何看待稀释问题？

丹-帕克我们做了一件非常有趣的事情。因为它是资本密集型的，所以需要有这样一个数字，这样我们才会有动力，你也需要有从事这项业务的人有动力去建设。

因此，我们通过谈判，从那一轮融资中拿出了相当可观的管理层分红。但我们真正做的是，史蒂夫和我并没有真正改变所有权。但我们真正改变了其他人的所有权我们把公司缩减到了87人 有很多优秀的人在那里工作

标准之一是你对这项事业有多大的决心和热情？因此，我们的核心团队大部分成员至今仍在这里。我得说，也许有两三个人已经离开了这个团队。

但是，我们的核心团队还是存在的，对大多数人来说，我们至少让他们的股权增加了两倍或三倍。

法布里斯-格林达：太神奇了。听着，这真是一次不可思议的过山车之旅，对吗？

就像起起落落，到回升，再到现在。你知道，尽管有一些我猜是宏观、地缘政治和三位一体的关税等因素，但情况看起来还是很乐观的。

有几个问题，你现在知道了什么，你会采取不同的做法吗？

丹-帕克我的意思是，你看，如果我有充分的后见之明，我会把那 100 美元或 75 美元藏起来，你知道，不去碰它，渡过难关，不那么快扩大规模。我想，这就是我告诉自己的，让自己晚上感觉好一些。

法布里斯-格林达：你的竞争对手是如何融资的？这是否也是一种选择？当然，如果竞争对手筹集了资金并进行了消费，那么如果他们没有筹集资金，就很难做到这一点。这样做比较容易。

丹-帕克他们在我们之前刚刚筹到资金，所以，是的。

法布里斯-格林达：好的。

丹-帕克所以我们同时也在融资，你知道，他们在媒体上公布了一个亿的数字。我们不知道这完全是债务还是股权，所以我们想，我们需要去匹配这个数字。这其中充满了动力我的意思是，这是一个激烈的， 竞争的那种动态。

法布里斯-格林达：对了，他们怎么了，他们刚刚死了。

丹-帕克拯救我们的是，我认为我们的行动和反应非常迅速。2023 年初，他们仍有 750 人。因此，他们的规模并不比我们大，但我们只有他们的三分之一。所以他们在2023年3月就宣布破产了

法布里斯-格林达：好的。不，我是说，这可能也会有很大帮助。是啊。竞争压力也会随之降低，然后你就能处于获胜的位置。

丹-帕克是的。成为品类领导者有很多飞轮效益。在我们并驾齐驱的时候，我并没有充分认识到这一点。

法布里斯-格林达：所以这个类别有很多像，你知道的。这些年来，尤其是在全球范围内，比如你看英国，有一家公司上市了，我想是七八十亿英镑。就像，所以像100亿美元 被称为Cazoo。很多公司都失败了

那些失败的人怎么了？你们有什么不同？有哪些人正在做着有趣的事情？

丹-帕克我认为最关键的是，很多人在广播开始时都提到了这一点。有这样一个阵营，做完全的垂直整合或轻资产，对吗？

这就是典型的风险投资，对吧？比如，我喜欢你做的事 但你能不能做得更轻资产一点？对不对？现在的赢家都是那些拥有全面垂直整合能力的公司。Carvana 就是一个赢家。他们现在的息税折旧摊销前利润率为 10%，市值回到了 500 亿至 600 亿美元。股价从3美元涨到60美分，现在又回到了300美元

因此，我的意思是，这对他们来说也是一次规模更大的反弹。但在美国，弗鲁姆和转变我们更多关注的不是基础设施，而是更轻的资产。因为情况是一样的，他们花得太多了。他们在营销和品牌上花费巨大，这很有帮助，但你也需要产品。

产品的核心是车辆的质量。我认为澳大利亚人做得非常好。他们在垂直整合方面也采用了类似的思维方式。墨西哥的卡瓦克（Kavak）公司，他们在市场方面有一些细微差别，但我认为他们现在的发展轨迹也很好。

但这只是核心。能够控制翻新和翻新成本，对这项业务至关重要。能够为客户提供质量更好的车辆，这才是真正为客户创造价值的地方。

法布里斯-格林达：你有没有看到类似邻近类别的有趣发展？

比如，把我的拖车开到美国，比如收藏级汽车，甚至是 eBay 上的 Caramel 整合，试图在关闭后完成。你对这些有什么看法吗？

丹-帕克是的，我认为有一些很好的解决方案能让消费者体验更好。我认为我们的做法是，是的，我想，我们想拥有一切。

我们希望拥有它的端到端，使其天衣无缝。要想做到这一点，实现真正的无缝连接，就必须拥有端到端的所有权。否则，你就会有这些互动点。举几个融资甚至保险方面的例子。我们不是保险公司。

这意味着什么？好吧，在加拿大，一旦你经历了我们给你带来的超棒体验，超级无缝，全数字化，你就会被踢出去，然后你会被告知去和你的保险经纪人谈谈。你得去注册这辆新车。这可能需要一天，也可能需要三天。对所以，这就造成了我们无法控制的流程摩擦点。你知道，最终我们希望能够控制它。

法布里斯-格林达：即使你不是负责融资、保险等业务的人，比如是的，它也应该是一个 API。

丹-帕克Where they integrated.对

法布里斯-格林达：是的，没错。是的，是一体化的。他们获得融资、保险、注册，一切都搞定了。你只需交付汽车，他们就可以去开了。

丹-帕克是啊我们不会很快承保汽车保险，但你知道，我们应该承保。

无缝集成到我们的流程中，因此您不会被踢出这个流程。

法布里斯-格林达：有通过这种情况发生的途径吗？

丹-帕克是的。

法布里斯-格林达：是 12 个月还是 24 个月？是36个月吗？

丹-帕克大概是 12 到 18 个。

法布里斯-格林达：有人这样做了吗？我指的不是加拿大，而是全球。

丹-帕克是的。当然也有类似嵌入式保险的提供商。但挑战在于

法布里斯-格林达：是的。你搭档的假期

丹-帕克嗯，不仅如此，如果你想，如果只是一个报价，那么客户就会想，你知道，如果我不喜欢这个报价，他们就会责怪你。对不对？对吧

是啊所以，你想提供多种报价，这样就会变得像更多一点的经纪公司。这里面有一些细微的差别，但是，是的，肯定会有这样一条路。

法布里斯-格林达：所以，你认为这将会公之于众，你认为会有什么途径？

丹-帕克未来三到四年，我们的目标是在三到四年内上市。因此，我们是否上市取决于资本市场和私人市场的情况。如果首次公开募股市场开放呢？但我认为，在一个战略收购者有限的世界里，你想为我们创造另一种选择，那就是上市。

因此，我们的心态是。如果我们有这个选择，或者如果我们想这样做，而且这是一个正确的决定，那么在未来三到四年内，我们就可以上市。

法布里斯-格林达：关税对你们有什么影响吗？

因为，我的意思是，这些都是二手车。它们是在加拿大准备好的。它们不是从美国运来的。

丹-帕克它们将产生下游影响。我对关税比较乐观。我确实认为，我们最终会找到一个地方，让消费者不再为此付出高昂的成本，因为在过去的 60 年里，我们已经围绕汽车建立了一个相当高效的全球供应链。

要在一年、两年甚至四年的时间里完成这项工作并不容易。我认为，原始设备制造商需要进行大量的规划和决策，才能决定在哪里建厂。印第安纳波利斯、印第安纳州或其他任何地方。因此，我乐观地认为，情况不会像现在这么糟糕。

从长远和短期来看，都会产生一些下游影响。如果新车价格上涨，那么需求就会开始转向二手车，从而对二手车造成价格压力。因此，我们可能会看到一些这样的情况，但目前还没有看到。

法布里斯-格林达：二手车价格是否有漏斗式变化？我的意思是，很明显，它们在此期间大幅上涨。是否完全缩减了。它们是比以前贬值得更快，还是一如既往？

丹-帕克是的，我们仍然比大流行前的水平高出 10%到 15%，但汽车价格，就像你记得的那样，我是说，美国和加拿大的情况不同，但都上涨了 20% 到 40%。

然后就降价了。我认为，一个有趣的现象是，很多几年前买车的人，现在他们的车已经出现了很多负资产，因为价格下降的幅度比一般情况下要大。因此，很多消费者的车辆都出现了大量的负资产。

法布里斯-格林达：我明白了，电动汽车的影响有多大，比如在你们的平台上，现在销售的二手车中有多大比例是电动汽车，从定价或其他角度来看，它们的表现与汽油车有什么不同吗？

丹-帕克我是说，现在特斯拉的价格走势很有意思。

主要是因为埃隆一直在做的一些事情，但我们大部分的电动车库存，我想说大约 80% 都是特斯拉的。好吧，有点反映市场。我们大约有 10%、12% 的汽车是电动汽车，这也反映了当今市场的情况。加拿大的电动车基础设施建设可能不如加利福尼亚，而且这里天气寒冷，电动车不太受欢迎。

但我认为这种趋势会继续下去，技术会改进，范围会扩大。采用率也会慢慢上升。

法布里斯-格林达：好的，你有没有考虑过长远的问题？比如让你彻夜难眠的事情是，我们是否能实现一个Waymo的世界，你知道，他们是自动驾驶汽车，随处都能按需使用，你不再需要拥有一辆汽车。

比如说，这是你担心的事情，还是说这就像是未来 20 年的科幻小说，因此不用太担心？

丹-帕克我在旧金山的 Waymo 车里。太不可思议了简直难以置信。

法布里斯-格林达：是的，很棒。

丹-帕克这是一款令人难以置信的产品。我认为汽车所有权可能会发生变化。当技术越来越好、成本越来越低时，一般会发生的情况是，消费者不再是大众消费者，而是自己想要拥有它，对吗？比如，我就想要一辆Waymo。

到时候，消费者就会觉得价格合理。每个人都只想在自己的车库里放一辆 Waymo 汽车。当然，可能会有一个网络流传，就像Uber流传一样。但我认为，汽车保有量可能会下降，但拥有一辆汽车的基本需求，我的意思是，你有孩子，要把汽车座椅搬进搬出，还要担心后座上的麦片。

就像，我让我的孩子们去打曲棍球，我认为消费者会希望继续享受便利，但又不必开车。因此，我认为汽车也许会改变。但所有权不一定会改变。而且我认为，我们也已经有了很多基础设施，可以帮助建立一个庞大的汽车网络。

因此，我不知道这是否会让我们彻夜难眠，但如果你不断进化，在生态系统的不同部分都有足够多的 “熨斗”，你就能够适应。

法布里斯-格林达：我很震惊。这很有趣，就像我投资 Uber 的时候，我曾担心，哦，自动驾驶会减少模式等等。

好像是 10 年前吧？十多年前好像是2012年还是13年是啊 Yeah.我很震惊现在的自驾车已经很有意义了但现在，比如你去旧金山、奥斯汀或洛杉矶 它就很突出、很普遍，而且还在加速发展。

它会越来越多这就是原因是啊，我想过这个问题，但就你的观点而言，你知道，现在有三个孩子，我希望有第四个，而不仅仅是未来的狗，汽车座椅，玩具，尿布。这就像，这就是为什么我需要双向的，就像，我需要一辆巨大的车。

现在市场上甚至没有一款车适合我们。我有一辆从轻酸到重酸的车。

丹-帕克我的意思是，也许你会因为有 Waymo 的车在身边而少开它，但你还是会喜欢它，还是会拥有它。对吧？

法布里斯-格林达：是的，有道理。是啊

有什么我们没有涉及但应该涉及的内容吗？

你有什么想分享或提及的东西吗？

丹-帕克是的，我想，我是说，我在过去五年中最欣赏的事情可能就是团队这一块，以及与其他人一起经历困难的事情。当我在风险投资公司工作时，我觉得我花在团队上的时间不够多。

你知道，你看数字，你看业务，但我的意思是，归根结底，这都是关于执行，对吗？从商业模式的角度来看，我们现在做的事情并不新鲜，也不前沿，就像以前有人尝试过这样做一样。

我认为团队非常重要，你必须长期与人建立信任，这有多难，有多少内讧、政治动态和关系管理必须发生。我认为随着时间的推移，这些都会变得更好。所以，我的建议或者说我的收获就是，这是一个永恒的话题，永远不会消失。

而现在我明白了，拥有一支相互信任的优秀团队是多么重要，以及团队的战斗力考验。我们的团队经历了现在的一切，就像我一样，你知道，我们有很多信心。

我们有很多信心，就像，是的，当然，我们可能会遇到一些难题，但我们会解决的。

法布里斯-格林达：所以基本上，风险投资公司不应该只盯着创始人，他们不应该是一个庞大的团队。其次，我同意，如果你在战壕里战斗，你有濒临死亡的经历，那些幸存下来的人，就像你的纽带，你是你的战友，你知道，永远的纽带。

丹-帕克我的意思是，你会喜欢，你知道，2010年，不管是什么年，喜欢，你会坐下来，你会抢啤酒，你会，你会告诉那些战争故事。像那些人，我的意思是， 这就像一半的乐趣给我。对这就像能够 告诉那些故事。 因为这就像， 那种我们为什么这样做。

对

法布里斯-格林达：完美。太棒了

感谢您的参与。谢谢大家收看我们的节目，我们下期再见。谢谢大家

丹-帕克真棒。

听着，我没想到自己会喜欢这本书。我在周末闲暇时拿起《地牢爬行者卡尔》，以为这又是一本可有可无的文学RPG游戏。然而，我得到的却是一部野蛮、聪明、奇特而又感人至深的奥德赛，它远远超出了它的重量级。

故事的前提非常荒诞–地球被改造成了一个虐待狂的星际地牢爬行真人秀节目，而我们的主人公卡尔（和他那只搞笑的猫咪 “甜甜圈公主”）就是其中一名倒霉的参赛者。但关键在于：在血腥、混乱和挥舞着电锯的地精小丑（是的，确实如此）的背后，是世界构建的无情效率。它的巧妙之处就像初创公司一样–游戏化系统严丝合缝，激励措施清晰明了，而且循环往复，令人上瘾。

卡尔是我所能理解的那种不情愿的创始人：被推上领导岗位，不断适应，从失败中学习，实时建立联盟。他并非一开始就是英雄。他通过快速迭代和拒绝放弃成为了英雄。故事叙述精炼、幽默，充满了让我关心的时刻。我发现自己在想，如果埃隆-马斯克（Elon Musk）、亨特-汤普森（Hunter S. Thompson）和特里-普拉切特（Terry Pratchett）在一场死藤水疗养之后合写了一部地牢爬行游戏，那会是什么样子。

是文学作品吗？ 不 它让人上瘾、聪明绝顶，是我今年读过的最有趣的书之一吗？ 当然可以！