← 全部单集 All-In 播客中文版
嘉宾访谈·中文精读

All-In 最佳投资点子大赛：四位投资人现场亮出顶级交易

All-In's Best Ideas Pitch Competition: 4 Investors Present Their Top Trades Live

四位基金经理分别推介 MGM 度假村、Talen Energy、Aktis Oncology 和 GEODNET，随后由主持人复盘并评出优胜者。

杰森、查马斯、弗里德伯格，嘉宾 Aaron Cowen、Oleg Nodelman、Kyle Samani、Gavin Baker··时长 1 小时 8 分·官方文稿·精读约 1.3 万字 · 阅读 29 分钟
官方音频00:00 / --:--

本期章节点时间可跳到音频对应位置

  1. 0:00查马斯解释“最佳想法”赛制
  2. 2:31Suvretta Capital Management 的亚伦·考恩推介美高梅度假村
  3. 13:07Bornite Capital 的丹·德雷福斯推介塔伦能源
  4. 27:19EcoR1 Capital 的奥列格·诺德尔曼推介 Aktis Oncology
  5. 40:20Multicoin Capital 的凯尔·萨马尼推介 GEODNET
  6. 54:50老友们复盘路演并宣布获胜者
中文精读由 AI 基于原始节目改写，非官方翻译，可能存在偏差或错误；观点归属与表述以原音频与官方文稿为准。说明

四位投资人，每人六分钟，站在同一个台上，讲自己眼下最看好的一只标的。台下坐着几百号人，加上杰森、查马斯、弗里德伯格三位主持人，一人一票。

最后的结果挺有意思：观众把第一名投给了 Talen Energy，拿到 50% 的票；三位主持人在后台一合计，冠军却是没人看好的 MGM。同一场路演，两拨人给出了完全相反的答案。

这就是 All-In 这期特别节目「Best Ideas Pitch Competition」的全部看点。它把华尔街最私密的一件事搬到了聚光灯下：那些平时只对 LP 汇报、在 CNBC 上永远看不见的基金经理，第一次当众亮出自己的底牌，并且接受同行的当面盘问。

查马斯为什么要办这场比赛

开场杰森问查马斯：你是怎么挑出这四位讲者的？

查马斯的回答得从十年前说起。有一个叫 Ira 的投资人，年纪轻轻死于癌症，家人成立了 Sone Foundation，办一场年度活动，地点在纽约的林肯中心。2015 年，当时还是年轻风投的查马斯收到邀请函，上了台。

我说亚马逊会变成一家万亿美元的公司，然后被全场笑出了门。我的老朋友 David Einhorn，他人很好，但这次完全错了，他说我认识万亿市值的公司长什么样，这家不是。结果我错了他的错。（查马斯）

那笔赌注后来大赚。2016 年他讲了特斯拉，选的是可转债。2017 年他觉得这是自己的巅峰之作，喊出「AI 是未来」，然后挑了 Box。

要是我当时挑的是英伟达，我早就可以退休了。（查马斯）

所以这次他想复刻 Ira 的那场活动，把那些真正在一线赚钱、却拿不到曝光度的经理人请上台。查马斯的原话是，这些人帮 LP 赚了大钱，却没有分发渠道。杰森补了一句：你没必要非在 CNBC 上见他们。查马斯接：你会在这里越来越常看到他们。

MGM：一个赌场故事，但主角不在拉斯维加斯

第一位是 Aaron Cowen，纽约一家管着 40 亿美元的基金 Suvretta Capital Management 的掌门人。他在这行干了 29 年，之前替索罗斯管股票业务，后来给 Steve Cohen 当 CIO。他一上台就自嘲，说自己绝对算同龄人里偏老的那一档。

他管的是全行业基金，手里一堆科技股。但他很清楚台下的听众是谁，所以这次要讲的跟科技没关系。他说既然这会的另一个主题是扑克，那就讲个博彩公司。

大部分人对 MGM 的印象是「拉斯维加斯公司」。它在拉斯维加斯有 13 处物业，和 Caesars 并列是当地两大赌场资产持有者。前不久 Caesars 刚被收购，Cowen 判断拉斯维加斯整体在回暖。但这只票的关键根本不在这里。

真正让 Cowen 兴奋的，是 Barry Diller 最近在疯狂扫货。这位传媒大佬目前已经持有公司 26% 的股份，而且这个仓位占了他个人净资产的 80%。

更巧的是，Cowen 两周前做完这套演示时，股价还在 37 美元。昨天的功夫，Diller 直接对公司发起了收购要约，报价 48 美元，股价也顺势冲到 40 多。

我不会把股票卖给他，一秒都不会。（Aaron Cowen）

他给的理由不止是 Diller 在买。这家公司自己也在回购，六年时间买回了将近一半的流通盘。一个传奇人物在扫货，公司同时在回购，再加上 80% 的 NAV 押注，Cowen 说这背后一定有什么东西大家还没看见。

他找到了两块「隐藏资产」。

第一块在日本。日本前几年搞了全国公投，各地知事投票表决要不要开赌场，最后只有大阪同意。项目预计 2030 年开业。日本本来就有不小的博彩市场，柏青哥加上赛马大约 400 亿美元。拿它跟澳门比，澳门是 300 亿；跟拉斯维加斯比，后者「才」100 亿。Cowen 估算这个 Osaka 项目能做出约 20 亿美元的 EBITDA，MGM 持有 40%，还能额外收管理费。

位置也是卖点。大阪飞上海比澳门和新加坡都近，飞北京跟澳门差不多、比新加坡近得多。Cowen 的说法很直白：日本人爱赌，中国人真爱赌。周末想赌一把，住上海或北京，大阪是首选，何况人家还是发达国家。

澳门有它自己的问题，是个低倍数生意。但这是日本，是第一世界国家。（Aaron Cowen）

第二块在迪拜。MGM 正在那里建一个大型综合体，里面有 Aria、有 MGM、有 Bellagio。现在迪拜赌博还是违法的，但他们在楼里偷偷留了 30 万平方英尺的空间。哪天迪拜决定放开博彩，赌场往哪儿放？就在那儿。两年后，Wynn 会在离迪拜 45 分钟车程的 Al Marjan 开一家赌场。Cowen 押的是，等 Wynn 开业、加上迪拜想重塑自己的地位，本地开赌场的可能性就来了。

把三块加总：拉斯维加斯资产约值 60，日本约 50，迪拜一旦成真再添 40 到 50。

我们认为这只股票能翻三倍。记住，Barry 在收购这家公司，他图的是发财，不做战略整合。所以这家公司现在进入了博弈状态。股价轻松能到 100 以上，甚至 150。（Aaron Cowen）

弗里德伯格追问了一个角度：抛开赌博，这些物业的娱乐价值变现有多少空间？他听说 Diller 花了很多时间想在娱乐上做文章，认为这块严重变现不足。

Cowen 承认如果 Diller 能把娱乐做得更好会加分，但他强调，博彩只是表面，这只票真正的看点落在亚洲，公司自己的演示材料里对日本几乎只字未提。被问到催化剂路径时，他讲了段亲身经历：自己曾在 SAC 工作，最关注的就是催化剂。而这次的催化剂，就是 Diller 直接亮出了底牌。

查马斯抛来一个尖锐问题：Caesars 也曾在迪拜等牌照，为什么 MGM 会不一样？

Cowen 的回答分两半。关于大阪，他复盘了自己 29 年的经验：当年 Wynn 从拉斯维加斯起家、后开澳门，市场是在开业前大约三年才开始定价这部分的，而现在 MGM 正好快进入这个时间窗。关于迪拜，他说这本质上是一个期权：项目是别人替他们建的、在替他们运营，当初就聪明地留下了 30 万平方英尺的空置空间备用。就算这个期权作废，钱也已经能翻一倍多；一旦兑现，就是三倍。

Talen Energy：把数据中心看成炼油厂

第二位是 Dan Dreyfus，来自 Bornite Capital。他先从电力周期的「解剖学」讲起。

正常年份，电力需求跟 GDP 同步增长。真正的变量是技术突破，因为很多技术突破极其耗电，会让需求瞬间飙升；等大家都用上了，需求又回落到跟 GDP 挂钩的曲线上，接着进入节能阶段，循环重新开始。

历史上掀起第一轮的是家电和空调。七八九十年代需求重新平稳，两千年代的主题变成节能，LED 照明、智能暖通、隔热玻璃、智能电器集体上马。同一时期，美国还把大量高耗能基础设施拆掉、搬到中国，比如电解铝厂。于是整整二十年，电力需求基本零增长。

现在美国正走出一段空窗，重新进入技术周期，电力需求要从 2% 出头这个水平开始爆发。

有一点极其重要：我们不需要 AI 需求，电力市场在未来二十年也会极度紧张。AI 需求只是火上浇油，它制造短缺，仅此而已。（Dan Dreyfus）

他引用了一段职业生涯早期的记忆。当年跟 Sam Zell 同台，主题是蒙古的机会，他在看一座铜矿，Zell 在看房地产。Zell 说了一句跟了他一辈子的话：能以一个硬资产的「重置成本」以下的价钱买到它，而这个资产未来会被需要、需要有人去新建产能，那就买下来拿着，等市场醒过来，再以一个大幅高于重置成本的价钱卖掉。Zell 就是这么对待 Equity Office Properties 的，在市场顶部卖出，但买的时候是重置成本以下的价钱。

Talen Energy 是一家发电商。它有 2 吉瓦核电、6 吉瓦天然气基荷电。今天在股市里，可以用 250 亿美元的企业价值买到这家公司。而重置成本是 450 亿美元。因为有债务，光是让股权价值涨到重置成本，股价就得翻一倍以上。如果按 Zell 的剧本走，这一轮周期结束时，它最终会以一个远高于重置成本的价格成交。

接着他抛出一个更宏观的判断。美国和中国在这件事上拉开了差距。这一轮周期开始时，美国的发电量是中国的两倍。到今天，中国的发电装机容量是美国的 三倍。他给出三个「要么有、要么没有」的推论：如果相信 AI 会带来科学突破，你手里得有电；如果相信 AI 会驱动机器人，你手里得有电；如果相信 AI 关乎国家安全和军事，那要么有电，要么就完了。这场建设必须完成，否则就会被甩在后面。

他给了一个特别好用的类比。数据中心本质上就是炼油厂。传统炼油厂把原油炼成航空燃油或汽油；数据中心把电「炼」成光子、token 或者叫智能。同样是大资本密集资产，每吉瓦大约 50 亿美元。电力之于数据中心，就像原油之于炼油厂，是那个输入。

推特上流传的英伟达说法是，需要比现在多 1000 倍的电。Dreyfus 说，哪怕这句话只有一半是真的，也需要能想到的每一种电源：几百吉瓦的核电、太阳能、轨道计算，全都得上。真正的瓶颈在供应链。一座数据中心和卫星发射、轨道数据中心抢的是同一批关键矿物，发电厂需要的镍基高温合金和造火箭用的是同一种，光伏电池要的银也一样。结果一定是各种东西都缺、到处都在延期。

他举了一个具体区域。美国有个叫 PJM 的电网（宾州、新泽西、马里兰一带），电网运营商的预测是，未来十年这一小块地方就需要新增 106 吉瓦电力。十年在地质尺度上差不多是明天早上，但基建偏偏就是按地质尺度走的，美国已经习惯了互联网时间，点个键外卖送到、车两秒来接，可建基础设施不是这个节奏。

他还点名了微软那桩著名的交易。他本想夸微软是家绿色公司，但微软硬是说服了 Constellation Energy 重启三里岛那座核反应堆，那个曾经熔毁、让核电背了三十年骂名的反应堆。为了让对方「有动力」动手，微软承诺电价保底 50 美元每兆瓦时，愿意按每兆瓦时 100 美元的钱连付二十年。他的原话是：三里岛，由微软 Azure 荣誉出品。

这类交易现在越来越难做了，因为监管会问：你把电网上的电都拿去做数据中心，我们怎么给居民供暖？他给这个时刻起了个名字，叫「紧要关头」。

回到数字。Talen 现在的股价在 300 多美元的高位。如果管理层什么都不干，就坐着经营，让跟亚马逊的数据中心合同逐步涨价，这家公司每年能产生每股 50 美元的自由现金流。假设股价 350 上下，那就是约 7 倍自由现金流。美国优质基础设施资产一般交易在 15 倍，所以光是「躺着不动」就能翻一倍。

如果继续以溢价跟数据中心签合同，或者电价上行，能到每股 70 美元的年经常性自由现金流，给个 15 倍，股价就是 1050 美元。再进一步，如果他们自己建电厂，现在监管甚至把发电商和数据中心叫到一间屋里，撮合双方达成一笔都有合理回报的交易。只要建 4 吉瓦（在所需的 100 吉瓦里只占零头），每股自由现金流能超过 100 美元。

所以，去买这只股票。这笔投机值得下注。（Dan Dreyfus）

杰森立刻补了一句免责声明：不构成投资建议。

Gavin Baker 问了监管风险：没人喜欢电价上涨，AI 越来越政治化。Dreyfus 用高速公路来回答。凌晨四点开上高速，你会觉得修这么多车道纯属浪费；但赶上高峰，你又会嫌车道不够。电力也一样，一天里只有几个小时真正吃紧。他的解法是，跟数据中心签 PPA 时，强制它们配大量电池和调峰机组，扛过最紧绷的那几个小时，作为建更多电厂的过渡手段。他相信人类的创造力会赢，数据中心的电会有，居民电价会涨，但会相对可控。

面对「这需要电站在电表后面还是前面」的提问，他明确说这是后者。他还给出了三档回报的对应关系：每股 50 美元收益是「什么都不发生」，翻倍；加上电表前后更多合同，到每股 70 美元；再开始建新产能，到每股 100 美元以上。

被问到终端倍数时，他解释自己用的 15 倍是个混合数：合同部分像债券现金流，理应给高倍数（国债 5% 对应 20 倍出头，再看增长），现货敞口的部分更波动、能见度低，给低倍数。合同越多倍数越高，这是他的经验法则。

Aktis Oncology：把癌症治疗比作一场战争

第三位是 Oleg Nodelman，旧金山生物科技基金 EcoR1 Capital 的创始人，13 年前起步时只有 1300 万美元。

一般来说，投资生物科技公司是个糟糕透顶的主意，夹在电影、酒庄和 SPAC 之间。（Oleg Nodelman）

他说这个行业很多时候更像赌场而非真正的金融市场，大部分来玩的「游客」在拉老虎机。而在 EcoR1，他们自认是打扑克的。在一个几乎人人都是押注科学的动量投资者的行业里，他们盯着的是安全边际。他们刻意不用博士或医学博士来管钱，因为不想对科学「上头」，只想对风险回报比率「上头」。用他的话说，是要「把别人家小孩的科学作业变现」。

这是他投资生物科技的第 25 年。基金从 2013 年的 1300 万起步，至今给投资人做出了 10 倍回报、年化 20%，如今管理约 25 亿美元。而且四年来第一次重新开放募资。

他讲了一个用军事类比串起来的行业史。早在 1973 年尼克松签署《国家癌症法案》时，人们就用军事术语描述治疗手段。这个类比很贴切，因为两边要做的事一模一样：找到敌人，选出最好的武器杀掉它，同时把误伤降到最低。

早期的手术和放疗像中世纪围城：把整座城堡夷平，烧掉周边村庄，指望敌人在废墟里。化疗源自一战时一次偶然观察，芥子气会杀死快速分裂的组织。肿瘤细胞分裂快，所以医生往病人体内灌化疗药，赌它先杀敌、后杀己。问题是头发、皮肤、肠道、骨髓细胞分裂也快，毒药根本不分敌我。第一代靶向药像 GPS 制导武器，不地毯式轰炸每一个分裂细胞，而是找准敌人的指挥中心点掉。可任何武器都有个问题，敌人会适应、会躲藏，在癌症里这叫耐药突变。十年前出现的免疫疗法，是不派自己的兵，而是招募本地盟友（T 细胞）替你打，效果惊人时极其出色，却高度依赖「地形」，也就是肿瘤微环境。

然后他讲到今天的正题：现代放射性药物。它像一群微型无人机，小到能在血流里穿行，靠分子识别找到目标，然后引爆一枚精确弹头，爆炸半径只有 100 微米，约等于一个细胞的直径。一次自主完成的定点暗杀，威力如同钻地弹，附带损伤却最小。

这家公司就是 Aktis Oncology，代码 AKTS。市值 10 亿美元，企业价值 5 亿，账上现金够烧三年以上，足以撑过明年的关键里程碑。Aktis 五年前成立，最近刚上市，3 亿美元的 IPO 超额认购 18 倍，礼来还下了 1 亿美元的基石订单，就是那家做减肥药的礼来。

公司设计的平台能搭载任何放射性载荷，足够复杂去打击多种靶点，又足够小，能被身体清除、副作用轻微。这个路线最妙的地方在于，医生在早期临床试验里就能用成像验证药物是否真的到了肿瘤，这大幅降低了临床开发风险。另一个降风险的做法是，前几个项目专挑已知有效的靶点，比如 nectin-4 和 B7-H3。nectin-4 在膀胱癌里至关重要，第二个针对 B7-H3 的项目野心更大，这个靶点在几乎所有主要实体瘤上都有表达，包括前列腺癌、结直肠癌和肺癌这「三大」。

Aktis 去年启动临床试验，对外指引是这两个领先项目在 2027 年出初步临床数据，其中 nectin-4 最早可能在第一季度。

哪怕只有一个项目跑出信号，公司拿到的价值就不止这两个项目，而是整个微型蛋白平台。这是生物科技里的圣杯：仅凭「未来可能」就获得估值。（Oleg Nodelman）

更打动他的是大药厂对放疗的浓厚兴趣。Bristol、Novartis、Bayer，以及给 Aktis 上市做基石的礼来，都在搭建放疗能力，急需管线资产。过去几年这个领域的并购和交易规模有 150 亿美元，他认为现在还在早期。这个模式还特别难被复制：仿制药一般不碰放射性药物，而且因为它涉及放射性同位素，中国也进不来。在生物科技这个行业里，能挡得住复制的护城河本来就罕见，这里算一个。

弗里德伯格追问为什么同位素对中国是禁区。Nodelman 解释，Aktis 用的载荷是锕，而锕是从镭-233 制造出来的，后者是 1950、1960 年代美国核计划的产物，本质是废料。中国当年走的是完全不同的路线（浓缩铀和钚），所以连这种原料都没有。

弗里德伯格接着问，当年 Amgen、Sanofi 那场最高法院判决之后，生物科技的「中国复制」风险一直存在，改一个氨基酸就能绕过专利。Nodelman 承认这个逻辑对很多生物药成立，但放射性药物不一样：同位素供应必须本地化，在美国本土解决，所以他没看到来自中国的竞争。

被问及为何上市后市场如此低估，他给了个很「生物科技」的解释：这只票上市以来横盘，因为这个行业的投资者极度短视，哪怕离数据只剩八九个月，对他们也还是太久。他预计资金会在下半年开始建仓，提前埋伏明年一季度的数据。

Gavin Baker 抛出一个「自私」的问题。他早年管过一只生物医药基金，就在人类基因组测序完成后不久，当时大家以为测序会引爆疗法和个人化医疗的爆发，但过去 25 年，整体进展并没有当年预期的那么大。他的假设是，基因组这个问题空间对人类大脑、或人类写的软件来说太大了，而 AI 会解锁一批革命性疗法。所以他问：在座所有人有生之年，发达国家的人均寿命会不会大幅超过 100 岁，到 125、150？

我愿意押「会」。原因之一是，我们已经有最好的长寿药了，只是很多人没意识到，就是 GLP-1 和那些减肥药。真正被数据证明能延长寿命的事情之一就是热量限制，而所有减肥药做的正是这件事。（Oleg Nodelman）

他说这只是个开始，因为它教会了人们可以靠吃药获得健康。查马斯递上来两个观众问题，一个是微重力能否作为治疗变量，Nodelman 认为跟本案关系不大；另一个是 AI 在药物开发和大规模癌前筛查上会不会颠覆精准放疗，他提到 Aktis 内部有个小小的「臭鼬工厂」式的 AI 项目，一群数据科学家在试着利用公司自己那套专有数据集。

节目最后，查马斯调侃了一句：刚才有人在 Oleg 还在台上时直接冲进了这只股票，现在涨了 6%。他说：别在我们都想买的时候这么干，行吗。

GEODNET：给 AI 造「轨道」的加密网络

最后一位是 Kyle Samani，Multicoin Capital 的创始人。他八年前创立这家基金，几个月前刚刚卸任。他在业内最出名的事，是在 Solana 上线前的三轮融资里全部领投。他还特意感谢了 David Sacks，说 Sacks 当年给 Multicoin 投了种子轮。

他要讲的是一个鲜为人知的加密资产，GEODNET。理解它的第一步是先看 GPS。台下几乎每个人都遇到过那种尴尬：手机把你定位在错误的位置、朝向错误的方向，屏幕上那个小人看起来正对着墙。GEODNET 用的是一种叫 RTK（实时动态定位）的技术，能把定位精度压到大约 两厘米。作为对比，普通 GPS 的精度大约是两米，也就是说精度提升了约 100 倍。任何形式的机器人都用得上，无人机是最明显的例子。

今天 GEODNET 是全球最大的 RTK 网络，也是增长最快的。左边那三家公司 Trimble、Hexagon、Topcon，都建了二三十年的 RTK 网络，加起来在全球大约有 1.2 万个基站。GEODNET 2021 年成立、2022 年开始铺网，到现在规模约等于这三家加起来的两倍。网络覆盖 150 个国家、超过 1.1 万个城市，覆盖了全球约 80% 的人口（不含部分受制裁国家）。

关键问题是它怎么铺得这么快。答案藏在那个去中心化的加密模型里。照片里是某个民宅屋顶上的 GEODNET 基站。全球那 2.2 万个节点背后没有 AT&T 或 Verizon 这种运营商。装它们的是业余爱好者、专业玩家或小企业主，为了赚点外快，自己掏钱把设备装上屋顶。今天上 GeoNodes 官网，花几百美元就能买一个基站，装在自己家或小店的屋顶上，它发射无线电波，你就赚钱，而且是用 GEO 代币支付。这个激励系统就是它冷启动的秘密。

他给了一串使用场景。农业上，美国农业部几年前推出项目，鼓励农场主采用精准农业技术，其中就包括 RTK 网络，现在农业部实际上在补贴全国很多农场主采用高精度农业，而大部分底层就是 GEODNET。一个具体例子是公司 Burro 做的「机器人骡子」，图里它正驮着葡萄，能驮任何东西，几乎适用于任何农场。随着计算机视觉、芯片、电池这些 AI 要素的进步，这类产品卖得飞快，而它们都会用 GEODNET 或类似网络。另一个例子是约翰迪尔最近推出的全球无人喷洒系统，车自己跑、给植物喷农药。Samani 说当天早上刚确认，纳帕这边就有酒庄在用约翰迪尔的这套设备，所以你今晚喝的那杯酒，可能就是这个系统、也就是 GEODNET 驱动的。

自动驾驶是另一个显而易见的应用。TomTom 是 GEODNET 的客户，它几乎是全球所有自动驾驶项目的供应商，用 GEODNET 的数据更新地图，让地图更精确，因为要覆盖地球的每一个角落。他还特别看好消费机器人的下一波，最典型的是割草机器人，没人喜欢割草，而这类机器人今年预计卖出 100 万台，来自 Yarbo、Sunseeker 等厂商，全都由 GEODNET 驱动。无人机这一块，全球最大的无人机制造商大疆是客户，虽然并非所有机型都用了，但很多机型在用。

他喜欢 GEODNET 的核心原因是它是个网络效应极其明显的生意，看起来就像一家天然电信公司。基站遍布全球，你得覆盖整个地球，而电信在历史上天然形成垄断，他认为这里大概率也一样。今天 GEODNET 是最大、增长最快、同时成本结构最低的网络，因为去中心化，人们把设备往自家房顶一放就行。

看经营数据。业务几天前刚刚跨过 1100 万美元年化收入，同比增长超过三倍，他预计未来 12 个月还会再翻三倍以上。它资本效率很高，还在把资本返还给代币持有者：1100 万里大约 80%（也就是约 880 万美元/年）被用来在公开市场上买入 GEO 代币，这一切都在 Solana 链上完成，地址公开、可实时验证。剩下的 20% 覆盖研发和销售团队扩张。客户网络很集中，约翰迪尔的人认识大疆的人、认识 TomTom 的人，所以它在核心客户群里病毒式传播。而且客户一旦用上，用量会随时间上涨：第一年通常每年花约 6 万美元，到第二年通常约 17 万美元，也就是第二年在 GEODNET 上的支出大约翻三倍。过去两年它的客户基数翻了五倍。

收尾时他给了几个硬数字。市值按完全稀释算约 1.5 亿美元。代币在 Solana 上交易，现价 24.07 美元，代码 GEOD。

问答环节，杰森的问题很直接，先问市值。Samani 说，去 CoinGecko 或 CoinMarketCap 这类网站看，可能显示 6000 万到 7000 万，因为不是所有代币都流通，但完全稀释后约 1.5 亿。杰森又问，背后有没有一家正经公司，还是那种开在开曼、巴拿马、董事会没人认识的「项目」？Samani 说团队是美国公司，在旧金山，CEO 叫 Mike Horton，是个很好的人，一直做 IoT 智能设备。杰森追问公司实体和代币的关系、该持有哪个。Samani 说应该持有代币，因为他自己就持有大量代币、不持有任何股权。两者的关系是：GEODNET 这家公司面对约翰迪尔、大疆这些客户，和 GEODNET 基金会之间有合同约定，把收入的 80% 拿去公开市场买代币。被问到公司是否拿过风投，他承认自己之前的基金 Multicoin 领投过 GEODNET 一轮。

查马斯转来一串观众问题。第一个：你像喜欢 GEODNET 一样喜欢 Helium 吗，那个「5G 版的 GEODNET」？Samani 说他六七年前就领投了 Multicoin 对 Helium 的投资，至今是长期信徒。第二个：有一长串已失败的 DePIN 项目，因为人们根本不认可代币奖励，这次有何不同？他的回答还是那几个数：880 万美元返还给代币持有者，1.5 亿估值，今年增长三倍，这是极其便宜的资产，只是赶上加密熊市、没人注意。

弗里德伯格想把生意本质问清楚：这算不算对客户现金流的一种证券化权益？Samani 说是，本质上就是收入分成。查马斯顺势确认了机制：约翰迪尔付给 GEODNET 的钱越多，回购就越多，代币价格理应被推高。

弗里德伯格接着从自己熟悉的领域发问。他管过一家叫 Precision Planting 的农业公司，知道约翰迪尔自己也做 RTK 系统。既然如此，约翰迪尔这些公司为什么要用 GEODNET？Samani 的回答是资本开支换运营开支：这些网络已经遍布全球、成本极低，GEODNET 大概只有自建方案价格的三分之一到四分之一，而且哪里都能用，还能缩短销售周期，客户可以直接说「买拖拉机就行，导航的事有人管」。

弗里德伯格又提出一个挑战：现在有资金雄厚的公司在推微卫星，号称能提供亚厘米级精度，用由 SpaceX 发射的网状网络，实际上同时替代 GPS 和 RTK，这样不就把地面基站的需求抹平了吗？Samani 说不存在在成本上竞争的可能，因为往太空发射卫星太贵，而 GEODNET 的基站只要几百美元。至于能不能完全替代 GPS 本身，他认为 GPS 是另一回事，因为 SLA 不同，要替代 GPS 必须做到无处不在、要有足够多的卫星。他抛出一个折中判断：最后可能是混合方案，一批 GEO 和 LEO 卫星，加一批遍布各地的地面基站。弗里德伯格则更极端，认为只要 LEO 卫星够多，单靠 LEO 就能替代整个 GEO。

Samani 还提到一个容易被忽略的点：能耗。往太空送信号消耗的能源远超地面基站。拖拉机也许无所谓，但无人机或任何对电池敏感的应用，地面永远是更优解。查马斯还问他为什么在这个赛道里专挑 GEODNET，而排除那些做「分布式训练」的代币。Samani 说他见过 GEODNET 创始人很多年、一路跟踪；分布式训练那批人他挺怀疑，认为大概率都跑不通；分布式推理可行，但没达到预期，他几年前投过一些，能用，但不是 A+。

主持人的复盘：分歧在哪

投票之前，三位主持人各给了一段点评，这一段其实是全场信息量最密的部分。

查马斯说他信奉「德鲁肯米勒式的投资法」：没有切身利益，你就不会真在乎。他喜欢这四个想法，区别在于仓位大小。他判断这四个标的呈现的是截然不同的不对称 alpha、截然不同的下行风险，还有流动性问题。

我很喜欢 Kyle 的想法，问题是我没法建到足够的仓位，我甚至不觉得今天能买进 100 万美元。一进场就会推动市场。我可能得从 10 到 20 万、再到 30 万慢慢买进去。Talen 你可以吸收几千万，别人眼皮都不抬一下。生物科技的问题是，像 Friedberg 说的，存在这种不连续的、可能归零的流动性风险，但又有 10 倍的上行，礼来会来收购它。而 MGM，我认为 MGM 和 Talen 才是你能下重注的。另外两个，你只能拿一小块，因为它们像彩票。（查马斯）

Gavin Baker 先把形式和内容分开看。

重要的是要区分「一场精彩的表演」和「一个真正好的风险回报」。Oleg 和 Kyle 的演讲都很棒，但我既不是医疗投资者，也不是加密投资者。纯从风险回报看，我认为 MGM 是最好的。（Gavin Baker）

他给 MGM 的理由是下行被 Barry Diller 的收购要约「封住」了，同时日本和迪拜是很有价值的未来价值来源。他同样觉得 Talen 的风险回报很有吸引力，唯一的问题是 AI 相关的监管风险在上升，而他不知道怎么把这个变量量化。他点破了 Talen 的关键：它的最大负面外部性跟 Talen 自己毫无关系，而是可能来自政府限价之类的「天降横祸」，或者是换一届政府、换一届国会、通过某些让本地算力变难的法律，从而改变电力供需。

另外两个，他给的评价也不低。他认为 Aktis 很吸引人，因为这个行业真正能带来千亿美元级结果的机会，前提是你能找到一种可广泛适用的平台机制，而这种机会极其罕见。至于 GEODNET，他承认整场讨论非常精彩。被杰森问到为什么不碰加密，他打了个比方：他对加密的感觉就像对单板滑雪一样。他这辈子都在学双板滑雪，滑得还行，有 25 年的经验、教训、伤疤都来自公开市场股票投资，让他现在换到单板上，就像要从零开始。

想滑单板的人尽管去滑，想玩加密的人尽管去玩，都很好。只要别横着滑下山坡、毁掉雪道就行。（Gavin Baker）

弗里德伯格按「上行、下行、时间」三个维度给判断。他给 MGM 大约三倍，还补充了一个他从拉斯维加斯朋友那儿听来的数字：威尼斯人酒店里的 Sphere 每有一场演出，每天就能带来 100 万美元的增量 EBITDA。一旦有了娱乐引流，博彩收入就会飞起来。他还单独听说，Diller 确实花了很多时间想重塑这些物业的娱乐体验。所以他觉得，哪怕把新地点的上行全部打折，MGM 也还有不少功课可做：下有要约托底，就算这笔收购不了了之、公司继续独立经营，两年内大概也有三倍空间。

对 Talen，他给的是三到五倍，但指出这是八年后的事，而且更受利率影响，因为自由现金流很大程度来自购电协议，估值倍数会随利率走。如果利率飙升，他设想的 15 倍就会遭遇「利差压缩」，这是他给 Talen 的下行情景。

对 Aktis，他的担心来自两个方向。他自己就投了一家做深层蛋白偶联物的公司，在实体瘤上疗效很强，他认为正在讨论的一些新治疗模态可能会给 Aktis 带来风险。他也认为中国风险是真实的，因为他在整个生物科技领域见过太多「被抄走、然后大家去中国」的案例。但如果读出一个好数据，礼来也可能六个月内就来收购，所以上行确实有，而下行是：一旦读出坏数据、或中国、或新模态出现，可能跌掉 75%。

对 GEODNET，他的判断最不留情面：他认为未来十年太空方案很可能会替代所有 RTK、甚至所有 GPS，而且这是「顺势搭上已经在升空的系统」，几乎不可避免。

我的排序大概是 MGM、Talen、Aktis，然后 GEODNET。（David Friedberg）

杰森把自己的思路分成两桶。AKTS 和 GEOD 属于彩票，可能有疯狂回报，但如果跑不通，归零的概率很大。MGM 和 Talen 明显有下行保护，人总会赌、灯总会亮着，所以这两个他都喜欢。他说自己在现场每个投了 20 万美元。

查马斯接了一句，说自己跟着买了他一半。弗里德伯格则开玩笑说自己还在等「三个订单」，还特意说自己没有 Robinhood 账户，得打电话给办公室。杰森透露自己的组合当时涨了 7%。最后他给的排序是 MGM、Talen、GEOD、AKTS。

两个不同的冠军

观众投票，共 150 票。

第四名，Kyle Samani，5% 的票。第三名，Oleg Nodelman，21%。第二名，Aaron Cowen 的 MGM，24%。第一名，Dan Dreyfus 的 Talen Energy，50%。

然后是主持人的投票：第四名 Kyle，第三名 Oleg，第二名 Dan Dreyfus，第一名 Aaron Cowen。跟观众名次几乎一样，只把前两名的顺序对调。查马斯当场说这是一次「大爆冷」，是把观众投票翻了个面。

节目在一片插科打诨里收尾。杰森坚持要颁一个他自己起名的「异性恋阿尔法男性奖」，奖杯是两个男人尴尬拥抱的雕塑，还现场演示了「保持五秒，两分钟时催产素释放」的拿奖仪式。Dreyfus 抱着奖杯回了句：我拿到奖了，也拿到龙舌兰了。

这场比赛的真正价值不在谁赢。它把四位风格完全不同的投资人放进同一个时间窗，逼他们用最锋利的方式讲清楚：为什么是这只标的、赔率在哪、什么情况下会输。同一个台上，一边是「人总会赌」的确定性，一边是「AI 需要电」的长周期，还有两个关于癌症和 AI 物理世界的期权。看完这段，比读十份卖方研报更接近真实的下注逻辑。

我们不指望这四位在 CNBC 上出现。你会在这里越来越常看到他们。（查马斯）

可以带走的几条

  • 看这类路演，先分清「能赚钱的生意」和「划算的赔率」这两件事。MGM 的吸引力来自下行被收购要约封住，Talen 的吸引力来自重置成本与自由现金流倍数，两者逻辑完全不同，别混为一谈。
  • 学查马斯的仓位思维：先问「我能下多大注」。流动性差的标的，哪怕 alpha 再高，也没法承载大资金，彩票式的 10 倍上行必须用小仓去配。
  • 把「隐藏资产」当成一个筛查工具。MGM 的日本 Osaka、迪拜的空间，公司自己在演示里都轻描淡写，这类信息不对称往往就是超额收益的来源。
  • 基建类投资要看时间尺度和利率敏感度。Talen 的故事很漂亮，但要放进「十年」这个尺度里，而且要意识到它的估值倍数会随利率波动，这是弗里德伯格点出的关键风险。
  • 生物科技的护城河要看「能不能被复制」。放射性药物因为同位素供应和放射性属性，天然挡住了仿制药和中国复制，这类结构性壁垒比单纯的疗效数据更值得重视。
  • 判断一个加密资产，先找它的价值捕获机制。GEODNET 用 80% 收入回购代币，一切在链上可验证，这比「叙事」更接近投资。但要同时警惕流动性、证券合规和代币与股权分离带来的治理风险。
  • 最后一条来自 Gavin Baker：别轻易跨到自己没有经验积累的赛道。25 年的伤疤和教训是一种资产，不滑单板，是因为清楚自己的比较优势在哪。
文稿来自 All-In 官网（allin.com）发布的官方 Transcript，说话人按官网分轨数据标注，个别段落可能张冠李戴。
Jason Calacanis (00:00:00): Maybe you could tell us a little bit about how you selected our presenters and your vision for this. Chamath Palihapitiya (00:00:05): I mean, for any of you guys who've been involved in Ira's SOAN, this is a gentleman that passed away from cancer far too young. And his family created this thing called the Sone Foundation and they would host this event. And it started in Lincoln Center and they would ask these managers and so at the time I was like a young venture investor and I got this invite and I showed up in New York at Lincoln Center in 2015 and I said Amazon's going be a trillion dollar company. And I was laughed out of the room. David Einhorn, who's a friend of mine, but who was totally wrong, said I know trillion dollar companies, this is not a trillion dollar company. Wrong. It turned out to be a great bet. I went back, I did Tesla in 2016, we picked the converts. And then in 2017, was like, all right, this is my this is this is it. This is my magnum opus. And I said, AI is the future. And then I picked Box. I was like, if I had just picked NVIDIA, I would have been an Iris Dan Dreyfus (00:01:05): A legend. Chamath Palihapitiya (00:01:06): Legend that I could have retired. Anyway, so we wanted to recreate Iris own and start to get these great managers who are making great picks, making a ton of money for their LPs. They don't get the distribution. And so it's just the chance to like get to know some Jason Calacanis (00:01:19): of these names. You don't have to see them on CNBC. Chamath Palihapitiya (00:01:21): You'll see them here more and more often. And we can just get to opine. Jason Calacanis (00:01:24): Roll the video. Yeah. Ladies and gentlemen, welcome to the best ideas pitch. Let's meet our contestants. Kyle Samani (00:01:35): Anyone should be able to trade any asset, anywhere in the world, anytime, twenty four seven with just an Internet connection and a phone in their pocket. We're building a new financial system from the ground up here. Aaron Cowen (00:01:45): People are gonna wanna own equities, and it's gonna be fun in the next couple of years. Companies gonna innovate and create products Jason Calacanis (00:01:51): and Aaron Cowen (00:01:51): applications and that's where hopefully long short managers like us can make a boatload of money. Jason Calacanis (00:01:57): My fund EcoR One Capital, which is Oleg Nodelman (00:01:59): based in San Francisco, thinks of investing in biotech in a slightly different way. We're looking for unfollowed, unloved, misunderstood biotech companies. Aaron Cowen (00:02:05): It's an amazing moment in time for those types of companies. Dan Dreyfus (00:02:08): There's been a structural and permanent perception shift where both sides of the aisle are gonna be leaning into nuclear in a big way. Massively optimistic. You know, all of this leads me to just the maximum risk on. Aaron Cowen (00:02:29): And thanks to the besties for having me, and this is obviously a fabulous event you guys have put on. I'm happy to be here. For those of you who don't know me, I run a $4,000,000,000 firm in New York called Sarepta Capital. Before founding my firm, I ran the equities business for George Soros. I was then CIO for Steve Cohen. And I've been doing hedge funds now for twenty nine years. So definitely on the older edge of my my peer group. So I was thinking about you know, I run a generalist fund and, you know, we own a bunch of tech stocks, but, you know, given this audience here, for me to pitch a tech stock would be absolutely completely stupid. So I was thinking about what else. And, you know, obviously, the theme of this conference besides tech is poker. So I'm gonna pitch to you MGM. Now most of you know MGM as you would think about it as the Vegas company. They own 13 properties in Vegas. They're one them and Caesars are the two largest owners of casino assets in Vegas. Now if you notice the other day, Caesars got taken out. And so we think Vegas is actually starting to improve. But I'm not here to pitch MGM because of Vegas. What I'm gonna tell you is there's couple of things we noticed. One is this company has been very aggressively been stock's been aggressively acquired by Barry Dillard lately. Barry now owns 26% of the company. Now I put this presentation together two weeks ago. Yesterday, he actually bid for the company. Okay? So when I put the presentation together, the stock was about $37. It's now high forties. He bid $48. Okay? I would not sell my shares to him. Kyle Samani (00:04:19): Did we Chamath Palihapitiya (00:04:20): get this presentation? Did we get it early enough to transfer? Aaron Cowen (00:04:23): I would not sell his I would not sell my stock to him for a second. And the reason is also, besides him buying the stock, the company's company's also also been been buying buying the the stock. Stock. Rarely have I ever seen a company in six years buy half their float back. So you have Barry Diller, who's the legend, aggressively buying the stock, and it's also now 80% of his NAV. Okay? So you most people think of Barry Diller as the ABC producer. He did IAC, which owned assets like Expedia, and now he's a casino guy. What is going on here? So we spend a lot of time asking ourselves why. And why is MGM has two hidden assets. Okay? The first one is and this is sort of our the punchline of what we think the stock is worth. So you add the Vegas assets plus China, you get about little low sixties. So from $48 or 37 when I started this, great return. What is they what they have now is a license to open a casino in Osaka, Japan. Japan, a couple years ago, went through a whole referendum around the country. They have prefectures. The prefectures voted. The only one that decided to own open a casino is Osaka. Now Osaka is and this is what the asset's gonna look like. It's gonna open in 2030. If you go to the company slide presentations, they sort of mention this, but they're not really talking about it. Japan, just for you people sorry. I don't know. It's very slow. Japan actually has a reasonably large gambling market. They have pachinko parlors, and they have horses. That's about a $40,000,000,000 market. If you look at the market in Macau, that's 30,000,000,000. And if you look at Vegas, it's only $10,000,000,000. So this could be a massive opportunity. You know, we're estimating they'll do about $2,000,000,000 of EBITDA. They own 40% of the property. They also get a management fee for this. Aaron Cowen (00:06:36): If you also look at where Osaka is located, it's a great you know, the so the Japanese like to gamble, but the Chinese really gamble. Okay? So if you look at where it is from Shanghai, it's shorter than Macau and Singapore, which is the two big gaming options in Asia. And for us and from Beijing, about the same distance as Macau and obviously much shorter than going to Singapore. So if you wanna go gambling for a weekend and you live in Shanghai, Shanghai, live in Beijing, Osaka is great. It's also a first world nation. And if you think about it as an investor, where would you want to have your money? Look, Macau has issues. It's a low multiple business. This is Japan. It's a first world country. So we think Barry Diller is understands gambling. He understands casinos. But what he's really doing is now trying to pick off the company to get the Japanese opportunity, which we think is worth will more than double the stock. The final option, and I'm keeping this simple. What I love about this pitch is it's really simple. It's not that hard to do the math. MGM is built somebody they're they're they're they're branding they're building a property in Dubai. Okay? Now it's gonna it's a grand complex. It has an Aria. It has an MGM, and it has a Bellagio. Gambling is illegal in Dubai right now. Okay? But they have snuck in this building 300,000 square feet of of space. Well, one day, if Dubai decides to legalize gambling, guess where it's going? Right there. Next year two oh, sorry. Two years from now, Wynn is gonna open a casino in a place called Ar Marjan, which is forty five minutes away from Dubai. Now any of us who wanna go gambling in Dubai, we Armarjan's a bit of a pain in the ass to get to. We're gonna wanna go here. So we think there's a chance that especially when wind opens also, look, there's possibility of the war, you know, Dubai wants to reestablish themselves, that they open a casino in Dubai. And, you know, what that would be worth so when you take the Vegas assets, which we think are worth about 60, when you take Japan, which we think is worth about $50. If Dubai happens, that's worth another 40 or $50. So we think the stock is a triple. Aaron Cowen (00:08:59): Remember, Barry's bidding for the company. Okay? He is not a strategic buyer. He is a financial buyer, and he's doing it to get rich. So therefore, I think this company is now in play. I don't know how it's all gonna play out, but if you own shares, don't tend to them. And the riskreward's incredible right now because, you know, I'm telling you, I think the stock could be easily worth over 100, could be worth $1.50. And now you have Barry Diller who is a has a firm bid, owns 26% of the company, basically at the same price. So I think this is a cool idea. Jason Calacanis (00:09:33): Well done. Okay. Anybody let's do two questions. Questions? Think we'll yeah. We'll put I'll give you both questions at the same time Chamath Palihapitiya (00:09:42): Sure. Jason Calacanis (00:09:43): For efficiency. David Friedberg (00:09:43): How much have you looked at like the monetization of the assets outside of gambling? I had heard from someone that Barry Diller was spending a lot of time trying to reinvent the entertainment piece of the properties. He was active on the board and they were trying to identify that the entertainment properties way the entertainment value is way under monetized and they could be making a lot more per Okay. Jason Calacanis (00:10:04): Don't answer yet. That's question one. Chamath Palihapitiya (00:10:05): Okay. Then question two is how when you expand internationally, do you scale customer credit because that tends to be the thing that drives. You know, people to come back and Aaron Cowen (00:10:16): Well, obviously, MGM I'll be sorry, your question first. MGM has a massive database of customers. Right? You know, I assume the Vegas properties have guys that come from China, they come from Japan. They'll use that database to do it. They also have a loyalty program. I unfortunately made a bad investment in a company called Rio The Rio, which was in Vegas, which we bought when they separate when Caesars merged with Eldorado, they had to shed an asset. That was that was the Rio. We I did an investment with a couple of friends and we were buying the thing at $200 per square foot. The thing we forgot was when you separated from Caesars, you lost the loyalty program. And that ended up pretty bad investment. Chamath Palihapitiya (00:10:59): Two quick questions from the audience. Aaron Cowen (00:11:00): Wait, wait, wait. I gotta ask Jerry Powell's question. Chamath Palihapitiya (00:11:02): Hold on. Question then from the audience. Aaron Cowen (00:11:03): Can I get his first? David Friedberg (00:11:04): Yeah. The entertainment question. Chamath Palihapitiya (00:11:06): I don't Aaron Cowen (00:11:06): know the answer to If it's If he can make them better, it will help. But as I'm saying, this is not really a Vegas play. This is an Asian casino play that the there there and if if you look at their presentations, which is really cool, they are not they barely mention it. So, what one of the things we happened beside we were hoping one of the so, look, I worked at SAC and one of the things we focus on is Catalyst Path. Right. So, what was the Catalyst Path? The Catalyst Path was they would have an Investor Day, blah blah blah. Barry just showed his cards. So but, you know, Aaron, if it's Chamath Palihapitiya (00:11:44): two questions. Caesars left Dubai waiting for a license. Why would this be different for MGM? That's question one. Sure. And then question two is the Osaka Casino was approved in 2023. Why was the market ignoring this hidden asset until the bid? Aaron Cowen (00:12:00): Sure. Let me answer this second. So what's also cool about this idea was so I've been doing this for twenty nine years. Wynn, they opened Macao. So Wynn started as a Vegas property, then opened Macao. The market started caring about it about three years before it opened. So that the answer is they should care about it. The reality is it tends to be about three years before it opens. Well, we're almost in that time frame, which is why we think it's opportunistically the right period of time. I I you know, regarding the question with Caesars, look, this is an option. As I told you, somebody built this project for them, they are running it for them, and they were intelligent enough to leave 300,000 square feet of empty space in case they get a casino. Well, if that happens, great. If it doesn't, you know, you're still gonna double more than double your money. So, you know, if if it happens, you triple your money. Chamath Palihapitiya (00:12:56): Pre option, you're saying? Aaron Cowen (00:12:58): It's pre option. Jason Calacanis (00:12:59): Well done. Big round of applause. Thanks, guys. Kyle Samani (00:13:01): Appreciate it. Jason Calacanis (00:13:02): Nicely done, Aaron. Dan Dreyfus (00:13:04): Next up, Daniel. Long time no see. So today, we're doing Talon Energy. But first, the anatomy of a power cycle. So a power cycle typically goes like this. In normal times, power demand grows about GDP. So if GDP grows 2%, power demand grows two. If GDP grows three, power demand goes three. And there's moments in time where we get technological breakthroughs. And a lot of those technological breakthroughs are very power intensive. So power demand spikes. And once everybody adopts that technology, it trends back down to its on algorithm GDP growth. And then you go through the efficiencies phase where we say, let's try to conserve and figure out ways to consume less power, and then the cycle starts all over again. So, you know, in history, the big technological boom that sent power demand skyrocketing was appliances and air conditioning. Everybody had to get their kettles and and and the air con. Then in the seventies and eighties and nineties, demand normalized again. But then the two thousands were all about efficiencies. You know, we had like LED lighting, smart HVAC, tinted windows, smart electronics. And at the same time, as I said earlier, we were like, you know, ripping down all our power hungry infrastructure, like aluminum smelters and moving over China. So we had two decades of effectively no power demand. And now we're just coming out of it and starting a technological cycle again, where power demand is gonna really start to explode from these sort of high 2% numbers you're seeing on the screen. Now I wanna say something right now that is incredibly important. We do not need AI demand to keep the power markets incredibly tight for the next twenty years. AI demand just turbocharges. That's all it does. And it creates shortages. So just remember that. Early in my career, I was panel with Sam Zell. Interestingly, Chamath Palihapitiya (00:15:25): it was Dan Dreyfus (00:15:25): a panel on opportunities in Mongolia. I was looking at a copper mine, and he was looking at real estate. There was one thing he said that stuck with me for the rest of my career. Is he said, if you can buy an asset, a hard asset, at below replacement cost for an asset that's gonna be needed in the future where we're gonna need to build new capacity of that asset, then you buy that asset at the discount to replacement cost, you hold it, and you sell it at a big premium to replacement cost when the market wakes up. That's exactly what we did with equity office properties, sold it at the peak of the market, but bought it at a discount to replacement value. Talend Energy is a power producer. They have two gigawatts of nuclear power, and they've got six gigawatts of natural gas base load power. Today, in the stock market, as a good speculation, you could purchase this company at a $25,000,000,000 enterprise value. The replacement cost is 45,000,000,000. And because they've got debt, it means that the equity value, just to get to replacement cost, is more than a double from where it's trading today. And if you follow Sam's playbook, then we ultimately end this cycle at a big premium to replacement value. So when I see this, I say the plan for America on the power side has to be this, make America great again, copy China. If you look at China did over the last twenty years, we started out this cycle with having two x the power generation that China had. Fast forward to today, China has three times the power generation capacity that we have. Now, if you believe that artificial intelligence is gonna be responsible for scientific breakthroughs, you either have it or you don't have the scientific breakthroughs. You believe that artificial intelligence is gonna drive robotics. You either have it or you don't have that productivity from the robots. If you believe that artificial intelligence is gonna be helpful for national security and military affairs, then you either have it or you're dead. And so this is an absolutely mandatory build out that we have to do. Otherwise we're gonna fall behind. Dan Dreyfus (00:17:40): Because at the end of the day, what is a data center? You know, in my world, in the commodities world, I look at the data center as the exact same thing as a refinery. In a traditional hydrocarbon refinery, you put oil in, crude oil in, you refine it into jet fuel or gasoline for your car. With a data center, you put electricity in, and on the other end, instead of gasoline or jet fuel, it comes photons or tokens or intelligence, whatever you wanna call it. But it's the same thing. Big capital intensive asset, $50,000,000,000 per gigawatt. And power, just like electricity, just like oil, is the input to that refinery. So here's Jensen, and he was just recently quoted that we need a thousand times more power than we currently have. Now, if that's remotely true, we need every single source of power that you can imagine. We need hundreds of gigawatts of nuclear. We need solar. We need orbital. We need it all, if this is even remotely true. But the challenge, as we spoke about before, is the supply chain. Right? All of these, you know, a data center competes for the same supply chain of the critical minerals that space launches and orbital data centers do. Power plants need all the same nickel super alloys that it takes to launch rockets and the silver that goes into these photovoltaic cells. So there's gonna be shortages of everything and delays everywhere. And my point here is we are just going to need every solution that we can throw at this for the foreseeable future. So here's a little region in The US called the PJM, Pennsylvania, Jersey, Maryland. This is a forecast from the grid operator, where they say that over the next ten years, we're gonna need a 106 gigawatts of new power in the PJM, in just one little area of The US. Now in ten years, in geological time, that's like tomorrow morning. Right? We're all so used to Internet time. You press a button and you get your food delivered to you or your car picks you up in two seconds. Dan Dreyfus (00:19:44): You know, building infrastructure happens in geological time. Ten years to build out a 106 gigawatts is literally a nanosecond from now. And, you know, you see that thermal coal retirements. We ain't retiring those coal plants because there's no world where we're gonna be building a 100 gigawatts in ten years. That's the size of what Japan consumes today for one little part of The US. And, you know, what I'll say is those that understand the supply chain and what goes into building all this, everybody's in panic mode because we know that we don't have the raw materials to meet this level of demand that's coming our way. So that's gonna keep existing capacity and power prices very tight. Now, the data center and the hyperscalers are in a panic. They're trying everything they can to source as much power as they can under long term PPAs, per power purchase price agreements at fixed prices for twenty years. There's a famous example. You know, I thought Microsoft was a green company. But they went and convinced Constellation Energy, which is a company that owns a Three Mile nuclear reactor, you know, the one that melted down and created, you know, the nuclear meltdown that gave nuclear a bad name for thirty years. It was Microsoft that told them they needed to start it up. And in order to incentivize to stimulate their hand to wallet reflex to start this thing up, they said, power prices stay are $50 a megawatt hour. We'll pay you a 100 a year for twenty years minimum price for you guys to start this up. And so here we have it. 3 Mile Island brought to you by Microsoft Azure. So, you know, it's getting harder to do these deals because the regulators are saying, wait a minute. If you're taking all this power off the grid for your data center, how are we gonna heat the homes of our customers? And so, you know, it we're getting ourselves into the moment of what I call crunch time. Dan Dreyfus (00:21:42): So just to finish up, here are the numbers on Talend. The stock today is sort of in the high three hundreds. If they just do absolutely nothing, just absolutely nothing, just sit there and run the business, let their Amazon data center contract roll up, these guys will be generating $50 a share of free cash flow per year. Again, the stock is in the high three hundreds. So it's about seven times free cash flow. Good infrastructure assets in The US traded about 15 times. So that's pretty good. You get a double for basically management just sitting around and doing nothing. But if they continue to figure out ways to sign contracts with data centers at premium prices or if power prices go up I mean, the amazing thing right now is in the PJM where these guys operate, the power price is still too low to stimulate new capacity. The math still doesn't work, which is really mind boggling. So if power prices go up a bit, they do more deals, you get to $70 a share of recurring annual free cash flow, put a 15 multiple on that, that's a thousand 50. But then if they get into building power plants, and right now the regulator is telling these companies to go sit in a room, power producer, data center, come in a room, make a deal so that you build power and get a good return on it, and the data center gets their power, gets a good return on it, and is in a pole position to be able to do this. If they just build like four gigawatts of the 100 gigawatts that we need, you could get up to, you know, over a $100 a share of free cash flow. The stock's in the high 3 hundreds today. So go and buy the shares. It's a good speculation, and we can chat. Jason Calacanis (00:23:18): Alright. Not financial advice. Gavin, Gavin, go. Gavin Baker (00:23:21): I'm just just very curious, like, do you think about regulatory risks here? Nobody likes their electricity prices going up. AI is an increasingly political issue. Just, Jason Calacanis (00:23:32): like, how do you think about that risk? We need AI and we need Dan Dreyfus (00:23:36): to figure this out. And so, there's different ways to skin a cat here. Right? My personal view is during peak hours, right? If you go drive down a highway at four in the morning, you know, you would sit there and say, why do we have all this highway capacity? This is crazy. But then you go on that same highway at rush hour, you're like, oh, we don't have enough highway capacity. There's not enough lanes. Power is the same thing. There's only a few hours a day where you really stress the system. And so, I think the working solution to get around this regulatory issue is you do the PPAs with the data centers, you force the data centers to throw a ton of battery behind it and some peakers just to get through that really intense period. And then, that's a good band aid solution until we build more power. So, there's ways to do this. Human ingenuity is going to win here. We're going get our data centers and consumer power bills are going to be, I think, relatively under control. They're going to go up, but they're going be under control. Chamath Palihapitiya (00:24:29): Okay, Dan. I have three questions from the audience, really good ones. Number one, does your thesis actually need behind the meter colocation to clear or is it just a bet that clean firm baseload is scarce enough that it doesn't matter whether power flows in front of or behind the meter? Dan Dreyfus (00:24:46): It's the latter. And that's why I gave three scenarios. Right? The $50 a share of earnings per share, again, high $300 stock. Right? $50 a share of earnings, nothing has to happen. You just sit. Right? And then you double your money. Now, if you get more behind the meter or even front of the meter, that's how you get up to that $70 a share of earnings from 50. And then if you get up to the 70 but start building new capacity, then you get to the $100 plus. Chamath Palihapitiya (00:25:11): Okay. Question two from Brad. How do you think about competition for power from things like fuel cells, gas turbines, aeroderivate turbines, orbital compute, and other sort of IPPs, independent power producers? Dan Dreyfus (00:25:24): We need all of it. We need all of it. We we know, fuel cells and, you know, the caterpillar solar turbines, these are fantastic bridge solutions, but the cost to run these things, the LCOE is like through the roof. But you you know, look, to build a $50,000,000,000 data center, don't want it to sit idle for three years waiting for your baseload CCGT. So, you do whatever it takes. You don't give a crap what you pay for that bridge solution. And so, we're finding ways through fuel cells, through Caterpillar solar turbines, hopefully through orbital data centers where we can alleviate this because I want AI to happen in a really big way and we're gonna need all the above. Chamath Palihapitiya (00:25:59): Okay. Question three. By the way, great questions, guys. Thank you for these. What is the right terminal multiple for Talend if the business mix shifts from merchant IPP to contracted infrastructure? Dan Dreyfus (00:26:11): Fabulous question. Chamath Palihapitiya (00:26:13): And the addendum here, and what percentage of EBITDA needs to be contracted before the market should re rate it? Dan Dreyfus (00:26:20): So that's a great question. And I only had six minutes to do this and I think I blew through my time, so I couldn't get into this kind of detail but it's something I would have really wanted to get into. So, whoever asked that, thank you. I just use the 15 multiple because it's sort of a blended multiple between the contracted stuff which will get a big premium multiple because it's a bond like cash flow stream and bond like cash flow streams trade at a small spread to treasuries and so treasuries if they're at 5% should trade at 20 times plus some growth or whatever plus or minus. The uncontracted stuff, the merchant stuff that has spot market exposure is more volatile, less visible, that should trade at a lower multiple. We can get into the minutiae, but just suffice to say, the more contracts, the higher the multiple, the less the lower the multiple, use 15 times as a good rule of thumb and you'll probably get to the right answer, which is what I used. Chamath Palihapitiya (00:27:07): Last question from Daniel Schurr, thank you for that. Dan, thank you. That was great. Dan Dreyfus (00:27:12): Oh, great. Thanks. My Oleg Nodelman (00:27:18): name is Oleg Nodelman. I'm the founder and managing director of Ecolor One Capital, a San Francisco based value oriented biotech fund that I started about thirteen years ago. Thanks a lot to the besties for having me here. I'm a huge fan of the pod like I'm sure all of us are. And I know how challenged Science Corner can get. So I wrote this in a way that even David Sachs would appreciate and pay attention to if he were here. Jason Calacanis (00:27:45): Well, paradoxically, he's taking a nap, which is what he normally does during Science Corner. Oleg Nodelman (00:27:49): Exactly. Generally speaking, investing in biotech companies is a horrible idea, sandwiched somewhere between movies, wineries and SPACs. In fact, our sector often feels a lot more like a casino than an actual financial market. And most of the tourists who are investing are playing the slots. Of course, at EcoRI, we consider ourselves poker players. In a sector where virtually everyone else is a momentum investor, betting on science, we focus on margin of safety. We're one of the few funds not managed by PhDs or MDs, and that's by design, because we don't want to fall in love with the science. We fall in love with the risk reward. And like the slide says, we want to monetize other kids' science projects. This is my twenty fifth year investing in biotech. I started my career with an eleven year stint at another fund and launched EcoR1 in 2013, humble beginnings, 13,000,000. Since inception, we've 10x ed to our investors and annualized at 20%. And today, have about 2,500,000,000 under management. We're lucky to have long term partners, many of whom are biotech entrepreneurs themselves and have been with us since day one. And we recently reopened for the first time in four years. Today, I'm going tell you about a company that's on the front lines of the war on cancer. Military terminology has been used when describing treatments for the disease since the early 70s when President Nixon signed the National Cancer Act. The warfare analogy is actually perfect. The warfare is actually perfect for cancer because both domains are trying to accomplish the exact same thing: find the enemy, figure out the best weapon to kill them and have minimal unwanted casualties along the way. First, a quick history of how this war has evolved. Early surgical cancer treatment and radiation was akin to a medieval siege: leveled the entire castle, burned the surrounding village and hoped the enemy was left somewhere in the rubble. Chemo actually evolved from an accidental observation during World War I, that mustard gas killed rapidly dividing tissue. Tumor cells divide fast, so doctors would flood a patient's body with chemo and hope it killed the enemy faster than it killed allies. Unfortunately, hair, skin, gut and marrow cells also divide quickly, and the poison doesn't discriminate. First generation targeted therapies were next, like a GPS guided munition. Oleg Nodelman (00:30:25): Instead of carpet bombing every dividing cell, you identify the enemy's command and control center and destroy it. The problem, like with any weapon, is that the enemy adapts and hides. And in cancer, these are called resistant mutations. Immunotherapy was first introduced to patients a decade ago. With IO, you don't send in your own troops, you recruit local allies, also known as T cells, and let them do the fighting for you. Spectacular when it works, but highly dependent on the terrain or the tumor microenvironment. This brings me to the reason we're here today. Modern day radiopharmaceuticals. Like a swarm of microdrones, small enough to navigate the bloodstream and find their target by molecular recognition, then detonate a precisely warhead with a blast radius of a 100 microns or the diameter of a single cell, an autonomous assassination with the force of a bunker buster, and minimum collateral damage. The company I'm gonna tell you about today is Aktus Oncology. The ticker is AKTS. The company has a billion dollar market cap, a $500,000,000 enterprise value, and a stockpile of cash which should last them over three years. Long past critical milestones that are coming next year. Actis was started five years ago, but recently went public with a $300,000,000 IPO that was 18 times oversubscribed and backstopped with a $100,000,000 order by Eli Lilly, the folks who bring you all the weight loss drugs. The company has designed a platform that can carry any radioactive payload, is complex enough to go after a variety of targets and small enough to clear your body with minimal side effects. The beautiful thing about this approach is that physicians can verify target engagement in early clinical trials with imaging. This significantly derisks clinical development because you know the drug is getting to the tumor. Another derisking strategy. For their first few programs, Actis chose known valid targets like nectin-four and B7 H3. Nectin-four is critical in bladder cancer, and the company's second program targeting B7 H3 is even more ambitious, expressed on every major solid tumor, including the big three: prostate, colorectal and lung. Oleg Nodelman (00:32:33): Actis started clinical trials last year and is publicly guided to initial clinical data in both of these lead programs in 2027, with Nectin-four coming as early as Q1. So you won't have to wait long. If either program shows a signal, the company is likely to get value not only for those programs, but the entire mini protein platform. This is the holy grail in biotech, getting value simply for the promise of what might come. What's even more compelling is there's an amazing amount of interest in radiotherapy from pharma. The big ones, including Bristol, Novartis, Bayer and Lilly, who backstopped the Axis IPO, have been building radiotherapy capabilities and they're hungry for assets to add to their pipelines. There's been $15,000,000,000 in M and A and deal making in radiotherapy in the last few years, and we're very much in the early innings. The neatest thing about this modality is that it's very hard to replicate. Generics generally don't traffic in radiopharma, and because the class involves radioisotopes, it's off limits to China. So unlike most of biotech, there's a real moat. And now the obligatory safety warning. Axis is not for everyone. You should consult your biotech analyst before purchasing Axis. Initiating a position may cause Jason Calacanis (00:33:48): increased anxiety to reduce sleep to the night. Sirius sometimes triggers drops in stock rates occur in biotech. Immediately after investing, you may experience sudden volatility due to handling your competitors. Stock declines are experienced for formal reason, call your broker immediately to increase your position. Remember, serious safety concerns have arose in other companies' clinical development programs. Well, those safety concerns occur with any Axis clinical program date, they may in the future. The use of many proteins liberating from a sales could not have been proven. Axis has no market products and thus no referring revenue. Solution for equity offerings may occur. In the event of a secondary offering, immediately schedule a call Oleg Nodelman (00:34:12): with Axis management team to discuss, place, get order. It's notoriously challenging to value biotech companies because when you risk adjust and discount back, you pretty quickly get to zero. For earlier stage opportunities like this, we like to triangulate. We think Actis could be worth $10,000,000,000 or $200 per share if even one of their programs makes it to market. And in this case, you have a lot of outs. David Friedberg (00:34:41): I'm not familiar with why radioisotopes are off limits to China. Oleg Nodelman (00:34:45): So in this particular case, Actis' radioisotope payload is Actinium. And Actinium is manufactured from Radium-two 33, which was used in our own nuclear programs in The US in the 50s and 60s. So it's a waste product from there. So Bactinium is not even available in other countries like China because they had a completely different their own program was completely different with enriched uranium and plutonium. David Friedberg (00:35:14): Risk for a lot of biotech and China replication came about that Amgen Sanofi Supreme Court case, didn't it? Where you could make a small because it basically said all patents are composition of matter patents. So you could change one amino acid, get around the patent. And China's basically done that with a lot of biologics that are patented in The US and Europe. They just rip them off and then you attach the radio emitting radioisotope to the molecule and you can kind of chase it. That's kind of why a lot of biotech's been depressed. Is that not true? Oleg Nodelman (00:35:45): Yeah. So with radioisotopes, again, because you to have a manufacturing supply that you have to source locally in The US, We haven't seen any competition coming from China at Jason Calacanis (00:35:58): all. David Friedberg (00:35:58): And if they have a successful readout though, would it not be like the case that someone in China would say, hey, let's go get some of the necessary radioisotopes? Oleg Nodelman (00:36:09): I'm sure they can do it for the Chinese market, but in terms of then transfer transferring that over here, we haven't we haven't seen it or kind of any wind of it at all. David Friedberg (00:36:19): And so then my last question, I'm sorry for monopolizing. Why do you think the markets discounted the value so much since the IPO? Oleg Nodelman (00:36:25): Oh gosh. Given the return in biotech It's pretty classic biotech. So it's traded flat since the IPO. Biotech investors are so insanely short term oriented that even though we're now call it eight or nine months from data, that's still way too long. And so our expectation is that folks will start accumulating this in the second half in anticipation of the data coming in the first quarter. Chamath Palihapitiya (00:36:46): Gavin, you had a question? Gavin Baker (00:36:47): Yeah. Sure. In the distant past, I ran a biopharmaceutical fund It's a very hard job. Congratulations on those numbers. But I ran that fund right after the human genome had been sequenced. And there was an expectation that the sequencing of the genome was explosion in therapies, personalized medicines, etcetera, etcetera. And I don't think, broadly speaking, we've made as much progress over the last twenty five years as maybe people thought in the early two thousands. And my hypothesis is that the genome is too big of a problem space for the human mind or software written by humans, and AI is going to unlock a lot of kind of revolutionary therapies. So my question to you, I will just admit, is a selfish question. It is not about your stock pitch, which was great. It's what do you think the odds are that in the lifetimes of everyone in this room, the average human lifespan in a developed country extends well past a 100, to a 125, a 150? Oleg Nodelman (00:37:49): I would take the over on that in no small part because we already have one of the best longevity drugs out there and folks don't even realize it in the GLP-1s and the obesity drugs. So one of the only things that's ever been shown in actual data to extend life is caloric restriction and that's literally what all the obesity drugs do. So I'm sure half the people in this room are on one of them and that's just the beginning because it's trained people that you can inject yourself with something and have healthy living through pharmaceuticals. So I think that's only going to continue. Chamath Palihapitiya (00:38:19): Oh, I got two questions from the audience. First one, as the launch costs per kilogram continue to fall, is there a credible pathway to use space and microgravity as a therapeutic variable given that cancer cells appear to behave differently in low gravity environments? Oleg Nodelman (00:38:35): That is a great question. It's probably not applicable to this. Chamath Palihapitiya (00:38:38): Okay. And then the second question, what would be a technological breakthrough that could disrupt precision radiotherapy as a result of AI at scale to drug development and precancer screening? Oleg Nodelman (00:38:50): Yeah. Awesome question. There's a small skunk's work project within Actus AI project. So with all these biotech companies, they have their little proprietary data sets that they hope to leverage with various insights. So a company like this with their many proteins and everything else they're trying to accomplish, they have their own little tiny group of PhD data scientist nerds who are seeing if they can leverage that in a pretty decent way. David Friedberg (00:39:14): Is it so it's been really hard to get CAR T in solid tumors. Is it the case that these kind of personalized peptide based immunotherapies are showing some efficacy in some solid tumors? Is that space that's going to expand and intersect here? Oleg Nodelman (00:39:33): What's most promising that I think a lot of folks have probably heard of is a new drug for pancreatic cancer from a company called RevMed with just another targeted therapy. For now, there's not a huge amount of progress from David Friedberg (00:39:46): peptide And have you looked at D proteins before? These kind of right handed proteins that seem to be able to penetrate solid tumors? Oleg Nodelman (00:39:53): One of the neat things about these mini proteins is they're hopefully of the right size to be able to deliver their payload inside of the tumors. Incredible. Chamath Palihapitiya (00:40:02): Oleg, thank you. Thank you. Well done. By the way, somebody just Yo LO'd into the stock while Oleg was on stage. It's up 6%. Like the lie. Don't do that while we're all trying to buy as well, please. Come on. Kyle Samani (00:40:19): Morning, everyone. My name is Kyle Simani. Thank you for being with us at the All On Liquidity today. Thanks to the besties for organizing. Today, we're gonna be talking about a little known asset, a little crypto asset called Geodet, which is building the rails for AI. So let's jump in. Quick bit about me. I founded a firm called MultiCoin Capital about eight and a half years ago. I stepped down a few months ago. And in my time there, was probably most well known for, leading all three rounds of investment in Solana prior to Solana's network launch, in 2020. I've been deep in the crypto space for a very long time, and I thought this would be a very natural forum to talk about a very interesting investment at the intersection of crypto and AI. Also, big shout out to David Sachs. Unfortunately, he's not here, but, David did seed Multicoin, back in the day. So thank you, David, for believing in me very early. Alright. Let's get into Geonet. So the way to understand, Geonet first is to look at GPS. Probably everyone in this room has been in in the situation on the left where you're using your phone and your phone is in the wrong spot, facing the wrong way. Right here, you can see this guy looks like he's facing a wall according to his phone. Geonet, it fundamentally is a new is uses a technology called RTK or real time kinematics where you can localize your location down to about two centimeters. For context, GPS, roughly, the the precision is about two meters. So you're getting about a 100 x accuracy for very precise geolocation. As you can imagine, any form of kind of robotics can make use of RTK, drones being the very obvious example. I'll touch on a few more. There's in a couple minutes here. Today, Geonet is the world's largest RTK network in the world, and it's also the fastest growing. The three companies you see on left here, Trimble, Hexagon, and Topcon, have all been building RTK networks in some form or fashion for, call it, twenty to thirty years. All of them combined have roughly 12,000 base stations deployed around the world. Geonet was founded in 2021, began building out the network in 2022, and today, they are roughly twice the size of the next three guys combined. Today, Geonet is live in a 150 countries around the world, more than 11,000 cities, and covers roughly 80% of the global population excluding some sanctioned countries. Kyle Samani (00:42:31): So this thing is really growing quickly. You might say, how did these guys build this network so fast? And the key is really this decentralized crypto model. So here we're looking at literally a photo of a Geonet base station on the roof of someone's house. The the global Geonet network, those 22,000 nodes, are not being built and deployed by someone that looks like AT and T or Verizon. Those base stations are being deployed by any random guy or hobbyist or professional or small business owner who wants to make some extra money. You can go on the GeoNodes website today. You can buy one of these base stations. They're a few $100. You put it on your roof of your house or your small business. It broadcast radio waves. You make money. You actually get paid in GEO tokens, which is the really cool part about this incentive system to bootstrap this thing to get it off the ground. So the Geonet network started about four years ago doing this. Today, it's now the largest, fastest growing in the world by a pretty wide margin. If you want a sense of scale, here we're looking at their coverage in United States. Obviously, every single major metro is covered, but even if you look at most of the rural parts of the country, you're covering actually the vast majority of even the rural areas. Let's talk about some of the customers and use cases for this. We'll start with agriculture first. The USDA actually launched a couple years ago a program to encourage farmers and ranchers to use precise ag technologies, including RTK networks. Today, actually, geo the USDA is now actually subsidizing many farmers and ranchers all over the country to adopt high precision ag, most of which is powered by GeoNet. Getting into some specific examples of that. Here, we're looking at what's called a robotic mule. This is made by a company called Burrow. Obviously, this is transporting some grapes. Kyle Samani (00:44:04): You could put anything on this. Has pretty obvious application for bay almost any farm or ranch you can imagine. With the advent in computer vision, CPUs, batteries, all the other AI stuff, these things are growing like hotcakes. All of them are gonna be powered by Geonet or something like it. Here, we're looking at John Deere. They have a new service that they rolled out recently called global unmanned spraying systems or GUS. These things drive around. They literally spray plants with pesticides and other things of that like that. I did actually confirm this morning there are wineries here in Napa that are actually using John Deere Gus vehicles. That was pretty cool. So if you have some wine tonight, maybe it was powered by by Gus. It was powered by GeoNet. Obviously, autonomous vehicles has a pretty obvious application for this. TomTom is one of Geonet's customers. TomTom is a supplier to basically every every AV program in the world, excluding maybe a couple. And today, TomTom is using GeoNet's data to update their maps to get them more accurate and precise as they need to cover every square inch basically around the planet. One of my favorite use cases are kinda the next wave of consumer robotics, which are getting a lot of hype these days. I think the most obvious one are robotic lawnmowers. I don't think anyone loves to mow their lawn. Robotic lawnmowers are now actually rolling out at pretty good scale. They're estimated they're gonna sell 1,000,000 robotic lawnmowers this year, made by companies like Yarbo, Sunseeker, and others. All of those these guys are all powered by Geonet. Next up, let's get to drones. The world's largest drone manufacturer, DJI, is a Geonet customer. It's not in all of their models, but it is in a lot of their models. And so, obviously, DJI is sending a ton of traffic now over GeoNet in the coming, you know, months and years as DJI rounds down in The US and you have new a wave of American drone manufacturers pop up. I'm gonna venture to guess that most, if not all of them, are gonna end up on the GeoNet network as well. The Geonet team is based in The US, has deep roots here. What love about Geonet is it's a very obvious network effect networks effect network effects business. This thing looks like a natural telecom. Kyle Samani (00:46:04): Right? You have base stations kinda all over the world. You gotta cover the whole planet. Telecoms naturally form monopolies historically. I think the same is likely to be true here. Today, Geonet is the world's largest and fastest growing network with also the lowest cost structure by a very wide margin because of this decentralized nature where people just put these things on top of their house. In terms of where the business at, the business just crossed about $11,000,000 in annualized run rate a few days ago, and it's growing more than three x year over year. I think it's gonna probably more than triple over the next twelve months. What's really cool about Geonet is how capital efficient it is and how they're actually returning capital to token holders. So today, the Geonet network is taking of that 11,000,000 in revenue, roughly eight excuse me. 80% of it is being used to make open market purchases of GEO tokens, and this is all visible on the Solana blockchain. They all have the addresses or published and stuff, so it's all verifiable in real time. That means $8,800,000 right now per year is going into buying an you know, geo net tokens on the open market. What's amazing is that last 20% is they're they're covering all their r and d costs and scaling out now their business development team. With a business like this, of course, like, it's a pretty small network of customers. The guys who work at John Deere know the guys who work at DGI, who know the guys who work at TomTom, and so this thing is now growing virally amongst this kind of core community of of customers. And as you can imagine, with customers who sign up for a service like this, they tend to ramp up their usage of that service over time. So once someone starts rolling out GeoNet, in the first year, they're usually spending about $60,000 per year. After two years, though, they're usually spending about a $170,000 per year. So the average Geonet customer is growing their their revenue with Geonet about three x in that second year. Kyle Samani (00:47:42): Obviously, then we look at their just their customers. They've signed up in the last two years. You can see they five x their customer base last year. Those are net net new customers. So applying some pretty simple math here, you can see they have a very clear path to more than three x this year as this thing ramps up. Just wrap things up summary. JUnit is the world's RTK network, growing the fastest. It has really obvious network effects and is is likely to be a very natural monopoly, growing three x year over year with a bunch of flagship customers and brands that you all know. Obviously, we have this huge physical AI tailwind behind us now, robotics, and all the other amazing stuff happening, and they're returning capital to shareholders. The token does trade on the Solana blockchain. If you wanna buy it trades $24.07. The ticker is geode, g e o d. So if you wanted to actually get some geode tokens, I encourage you to sign up for a a crypto wallet, a Solana wallet, and you can go ahead and buy Jio tokens from there. And with that, I think we are ready for some q and a. David Friedberg (00:48:36): Awesome. No problem. Jason Calacanis (00:48:37): What's the market cap? Kyle Samani (00:48:39): Oh, sorry. It's trading about a 150,000,000 on a fully diluted basis. If you were to go look Jason Calacanis (00:48:43): at any other crypto price websites like coin.gov or coinmarketcap, they're Kyle Samani (00:48:46): gonna show you something like 60 or 70,000,000. That's because not all of the tokens are floating yet, but the fully diluted number is about one Jason Calacanis (00:48:51): Is there a corporation behind it? Or is this just like a project in the Cayman Islands, in Panama with a board that nobody knows who's on it? Tell us about governance. Kyle Samani (00:49:03): So the GeoNet team is a US based corporation. There are Jason Calacanis (00:49:06): four teams in San Francisco. The CEO's name is Mike Horton. Kyle Samani (00:49:09): Really, really good guy. He's been building in this kind of IoT smart, you know, device based The Jason Calacanis (00:49:14): relationship between the corporate entity and the token, and which one should we own? Kyle Samani (00:49:21): You should own the token because I own a lot of token as as you might imagine. I don't own any of the equity. The relationship is Geonet, the cost you know, the company is facing John Deere, DJI, all these companies, and they have a contractual relationship with the Geonet Foundation use 80% of their revenues to buy tokens off the open market. Jason Calacanis (00:49:39): That's Has that corporation raised venture capital or anything? Kyle Samani (00:49:42): Yes. My prior company, Voltage Point, actually led around in Geonet previously. Chamath Palihapitiya (00:49:47): Okay. Kyle, I have I have many questions from the audience, so bear with me. Question one. Do you like Helium as much, which is GeodNet for five g signal? Kyle Samani (00:49:55): Yes. I actually led MultiCoin's investment in Helium six or seven years ago, and I continue to be a very big long term believer. They actually had big news go out this morning. But, yeah, I'm a big Helium fan. Chamath Palihapitiya (00:50:06): Question two. There's a long list of deep in projects that have failed because people just don't value the token rewards. Why is this any different? Kyle Samani (00:50:15): I mean, they're returning capital to shareholders. This thing is, you know, returning $8,800,000 to shareholders, it's trading at a $150,000,000 valuation and it's going to grow at three x this year. It's unbelievably cheap asset. It's just people aren't paying attention because it's crypto bear market right now. Chamath Palihapitiya (00:50:29): Okay. From Sam. Sorry. Can I David Friedberg (00:50:31): just understand, it's a securitized interest in the cash flows from the customers? Kyle Samani (00:50:35): Effectively, yes. It's it is a revenue. It is a David Friedberg (00:50:38): It's a revenue share token. Kyle Samani (00:50:39): Correct. Chamath Palihapitiya (00:50:40): Eight Okay. So the more John Deere pays GEO Net the company, the more you basically deprecate the tokens which should cause the token Kyle Samani (00:50:50): buying tokens in the open market. Correct? Yes. Chamath Palihapitiya (00:50:52): Yes. Okay. From Sam, what accrues value? The equity or the token? Similarly, we just covered that. How does the value accrual mechanisms square or not with current securities laws or what's contemplated in the Clarity Act? Kyle Samani (00:51:04): Yeah. So, the one answer to your question is the the tokens are the ones accruing value because they're taking 80% and buying. The other 20% is obviously funding operations. They have engineers, salespeople, all that stuff. So, that's all there and being funded. There's just securities laws. The Clarity Act passing is certainly very good for GeoNet. I'm not a lawyer, so I'm not going to tell you that, you know, it it passes the bar set in the Clarity Act, but I can tell you I'm an optimist and I've been very involved in the Clarity Act and I'm not too worried about it. David Friedberg (00:51:30): Okay. Can I can I ask about the business just real quick? Yep. So John Deere, I know this space somewhat well. I used to manage a company called Precision Planting in Agriculture. And there was John Deere makes their own RTK systems. So when you, like, run a a piece of equipment that relies on RTK, you're buying in the construction industry, Topcon or Leica or Trimble or John Deere, and you install the RTK base stations and you run your equipment. Why would John Deere and others wanna rely on this system as a diff diff like, is it better than, like, big systems that they're already using? It wasn't quite clear to me. Kyle Samani (00:52:07): I mean, CapEx versus OpEx. Right? Like, right, these networks are all over the world now. They're running at very low cost. GeoNet is is probably a third to half sorry, a third to a quarter the price than buying up your own CapEx and and doing it, and it's just available everywhere. So now it just reduces the sales cycle time for John Deere when they just say, buy the tractor. It's good. David Friedberg (00:52:26): There's another big push right now for MicroSATs to be an alternative to GPS in a way that they can actually provide sub centimeter resolution effectively replacing both GPS and RTK using a mesh network from SpaceX launched or actually SpaceX I don't know if SpaceX looked at doing this, but I know that there is a very well funded company that is trying to put up microstats to basically replace GPS and RTK. Doesn't that ultimately kind of wash out the need to have all these earth based base stations? Kyle Samani (00:52:58): There's no chance they can compete on cost because just sending things to space, satellites, that's so I mean, these Geonet base stations are a few $100. Like, you're just not gonna compete on cost with with Geotab. Chamath Palihapitiya (00:53:09): You you don't think that this is a viable replacement at scale and saturation for GPS itself? Kyle Samani (00:53:15): No. GPS is definitely very different and David Friedberg (00:53:18): Because the SLA is different. You've gotta have ubiquity for, for GPS for GPS alternative, which is why you have to have the satellites Satellites. Everywhere. You gotta have enough set. But if you get enough satellites, you can actually get to RTK Precision and you don't need to have the big expensive GPS. Chamath Palihapitiya (00:53:32): You'd have a hybrid situation where you have a bunch of Geo and LEO plus a bunch of base stations all over the place. That hybrid situation probably. David Friedberg (00:53:41): You could actually get the LEO alone can replace all of the the GEO stuff. Chamath Palihapitiya (00:53:46): Right. David Friedberg (00:53:46): That's the goal. And then if you get enough of them, which ASX unlocks. Yeah. Chamath Palihapitiya (00:53:51): And Kyle, what what about, like, other tokens when you think about other compute tasks like work to be done? For example, there's a bunch of tokens that have emerged in distributed training. How did you hone in on this and exclude the others? Or I mean Not Meaning prefer this over that? Kyle Samani (00:54:08): I mean, I met the GeoNet founder years ago. He pitched us, I've gotten to know him and followed it. The distributor training stuff, there's a whole bunch of people trying it. I'm pretty skeptical. I don't think any of it's gonna work. Distributed inference stuff is is possible, although it has not worked as well as we would have hoped. I did put some money behind that a few years ago. It's it's working, but not not a plus. One last thing actually, David, on on your prior question I wanna highlight is also energy use. Going to space just consumes way more energy than going to a base station that's, know, on the ground. And so, yeah, for a tractor, maybe that doesn't matter. But for a drone or for any other battery sensitive application, ground is always gonna be the preferred solution. Chamath Palihapitiya (00:54:45): Super interesting. Well done, Kyle. Thank you. Thank you so much. Super interesting. David Friedberg (00:54:49): Alright, guys. Before we vote, Jamal, give your feedback. Chamath Palihapitiya (00:54:53): Here's what I like. I I I apply the Stan Druckenmiller School of invest and investigate. I really believe in it. Yes. If you don't have any skin in the game, you don't care. And this is the kind of stuff that I love. I love hearing ideas like this. I love all four. My my difference is in sizing. So, you know, there's there's certain asymmetric alpha that each one of these exhibits, and then there's very different downside risk for each of them. And then there's also liquidity issues. So for example, like, I love Kyle's idea. The problem is I I could not get enough working for me where so you I don't even think I could get a million dollars in today. It would to scale in, it would move the market. So I would have to I'd have to probably I'd be like $10.20, 30,000 and then maybe Yeah. Start to buy into it. Talend, I think that you could absorb tens of millions and people wouldn't buy it an eyelash. The biotech company, the issue there is that I think that there is, as you said, Friedberg, this discontinuous illiquidity zero risk, but then there's the 10 x upside. Yeah. So there's just like huge Lily will bid for it. And then MGM, I think, is just that so I think MGM and Talend are the ones you could have huge sizing in. Right. And then the other ones, I think you have a piece because they're like lottery tickets. I think your point on MGM Okay. Wait. Hold on. Let me just review. So company number one was MGM. That was Ameren. And resorts. Okay? Company number two? Talon. Talon Energy. Company number three? Actis. Jason Calacanis (00:56:23): Actis. Chamath Palihapitiya (00:56:25): Then? Geodet. Geodet, not company, but I guess token. Yeah. Company number four, Geodet. Jason Calacanis (00:56:31): And you're buying the token, not the company. David Friedberg (00:56:33): Do you think the maybe for you too Gavin, like the Chamath Palihapitiya (00:56:37): Gavin, you ranked them. Jason Calacanis (00:56:39): Well, Even before you rank, just tell us what you think of the format and then assess the companies. We'll do ranking at the end. We're gonna do four three two. We're gonna do four three two one on stage, but give me your general ideas about the pitches, what you liked, what you did. Gavin Baker (00:56:54): I thought the pitches were great. I thought the format was amazing. I would for sure expand it next year. There are platforms that you guys could have a all all in basket or ETF that people could trade in. So like maybe that's something Will you do it next year? Will I pitch next year? Yeah. J Cal, I'll do anything you can do. Chamath Palihapitiya (00:57:14): Well, I locked. He's locked. Here's what I would ask Gavin to put you on the spot. Next year, I would I think we would all learn and benefit if you would do silicon and memory super cycle. Gavin Baker (00:57:27): Sure. Chamath Palihapitiya (00:57:28): Would you be willing to do that for us? Kyle Samani (00:57:29): The talk. Gavin Baker (00:57:29): I'll do it. Sign me up. Perfect. Kyle Samani (00:57:32): Lock up. Gavin Baker (00:57:32): Sign me up. Chamath Palihapitiya (00:57:33): Okay. Thank you. Gavin Baker (00:57:33): So keep going. Well, As far as the pitches, I do think I think it's important to disaggregate like what was a really great entertaining pitch versus what I think is is a really good risk reward. I thought Oleg and Kyle did a great job with the pitches, but I'm not a healthcare investor nor am I a crypto investor. I thoroughly enjoyed the presentations. I actually thought GNET was very interesting. I'm happy to learn from Oleg that I might lift well into my one hundreds. That's good news for me and everybody in the room. I enjoyed all the the military terminology and analogies. Chamath Palihapitiya (00:58:09): Yeah. That was really great, That was great. Really great. Gavin Baker (00:58:11): I do think from a pure risk reward perspective, I thought MGM was the best. Your downside is really capped because of the Barry Dillard bid. And then you have Japan and Dubai as, I think, very valuable future sources of value. And do think talent is also a very compelling risk reward. I just think everything in AI is going to need to grapple with increasing regulatory risk, which we talked about last time that that I was on the pod with you guys. And I don't know how to dimensionalize that. And, you know, I've been Chamath Palihapitiya (00:58:48): Like, the big negative externality for Talon is nothing to do with Talon. Gavin Baker (00:58:52): Nothing to do with Talon. Chamath Palihapitiya (00:58:53): It's like something over the top from the US government caps prices, something something. Yeah. You have Nationalizes the lab. Kyle Samani (00:58:59): You have Gavin Baker (00:58:59): a change in administration. You have a change in congress. There's laws that are passed that I think make it hard for terrestrial compute which changes the utility supply demand. But I actually think outside of that, Talend was super compelling and Jason Calacanis (00:59:12): So you got MGM, you got Talend, now you have the other two. Gavin Baker (00:59:15): I I I thought they were both great pitches. Can I tie them for third? Jason Calacanis (00:59:19): Just Well, no. Don't even give the score. Just any feedback on those two ideas or those are just a little bit lottery ticket for you or? Gavin Baker (00:59:28): No. I thought Actis was was very compelling. They're trying to trying to do something different. As Oleg said, if you ever get a biotech company that can become a platform and they have a mechanism, whether it's of drugging, whether it's targeting. If you have something that is broadly applicable, that is when you can get these really, really big $100,000,000,000 plus outcomes in biotech, which are rare. So I thought that part of Jason Calacanis (00:59:55): Right. Gavin Baker (00:59:55): Actis was super compelling. And You don't play crypto. I don't play crypto, but I thought the entire Gino Geonet discussion was fascinating. Jason Calacanis (01:00:04): I'm Is there anything happening in the off the bench and make you jump into the crypto game or it's just yours why are you not playing the crypto game? Gavin Baker (01:00:14): I feel about crypto exactly the way I do about snowboarding. Okay? I'm I'm not a very good athlete. I've spent a lifetime learning how to ski, and I'm okay. And just the idea of getting on a snowboard, having, you know, thousands of hours of ski instruction Jason Calacanis (01:00:33): You don't want the pain David Friedberg (01:00:34): in the game. Gavin Baker (01:00:35): And I have twenty five years of lessons, learnings, pain, scars from from investing in equities and public securities. And just crypto, it's a little bit like snowboarding for me, but like, you know Jason Calacanis (01:00:49): Yeah. Gavin Baker (01:00:49): 200 everybody who wants to snowboard, that's great. Everybody wants to do crypto, that's great. Jason Calacanis (01:00:54): Just please don't go sideways down the mountain and ruin the powder, David. I think Your assessment of MGM talent Okay. David Friedberg (01:01:01): I think MGM I look at the kind of return upside, the downside and the timeline. MGM's like probably a three x. I think it's also missing this point that I've heard a lot about on you can actually upgrade the monetization on these Vegas properties. We were talking to a friend of ours in Vegas. They're they're making a million bucks a day in incremental EBITDA every day that they have a show at The Sphere at the at the Venetian Hotel which is an unbelievable statistic which tells you that when you have the entertainment draw, the gambling revenue just flies. Flies. And so, Barry Diller, I have heard separately has been spending a lot of time on trying to reinvent the entertainment at these properties and thinks he has an idea on how to do it, which will cause the gambling revenue to fly. So, I think even if you discount the upside on these new locations, there's probably a lot of work to be done. Yeah. And I do like MPM. Floor on the bid and then you've got call it three x in two years even if this bid goes nowhere and they keep the thing running and they're like, we're going to reject the bid and keep running independently. Talend is maybe 3x upside, 5x upside but it's eight years out and I think one of the other challenges with Talend that I would kind of use as a valuation metric is think it's more interest rate sensitive than MGM is because the power purchase agreements really are where a lot of the revenue comes from. So, you're going get a discount rate that's a multiple that's a function of where interest rates are sitting. So, I think if interest rates shoot up which some might argue there's risk there, you actually get margin compression from that 15x outlook that he has for Talon. So, that would be my kind of downside scenario on Talon in the time ahead. In Actus, I do worry because I'm an investor in a that's got a deep protein conjugate that shows really strong efficacy into getting solid tumors. I think that there are new modalities for therapeutics for solid tumors that are being discussed that may kind of put this at risk. I think the China risk is legit because I've seen it across the board in biotech. Everything gets ripped off and people go to China but they could have a hit and Lilly could bid on it in six months if they actually get a good readout. So, there's certainly upside but the downside is probably 75% if they get a bad readout or China or some new modality comes out. David Friedberg (01:03:10): So, think the ranking is probably MGM Talon Actis and then the, for me the GeodNet piece, I just think the space thing is likely the path, it's going to replace all RTK and all GPS in the next decade. It's inevitable piggyback on systems that are already going up. Jason Calacanis (01:03:26): All right, great. So, I think I've got everybody. For me, I put them into two buckets. I think AKTS and God, those are like lottery tickets could be crazy returns, but there's a big probability of a zero there if they don't actually work. And then MGM and Talend obviously got the downside protection and those feel like people will always gamble and leave the lights on. So I kinda like both of those. I put 200 k into each in real time. Gavin Baker (01:03:57): I think people will leave the lights. Jason Calacanis (01:03:58): So that's just like my David Friedberg (01:03:59): I don't have a public Did you buy? Jason Calacanis (01:04:01): I'm just day trading. Chamath Palihapitiya (01:04:03): I I bought half of his action. Okay. Okay. Don't have a Robinhood account. I have to call my office. So I was like, just I'll take lose, buddy. I'll take you half of David Friedberg (01:04:12): I'm up seven you I didn't see you with your thumbs. Jason Calacanis (01:04:14): I'm up 7% across the portfolio. So I don't think I can include steps David Friedberg (01:04:19): to buy I did. I I waited the three in the order I Jason Calacanis (01:04:21): said on my So anyway, I'll just give mine really quick. I'll I'll go MGM, Talon, GOD, AKTS. Gavin's only gonna make Okay. $50,000,000,000,000. Chamath Palihapitiya (01:04:33): Let's bring Jason Calacanis (01:04:33): the let's bring our four pictures out. Chamath Palihapitiya (01:04:35): You have a thousand dollar get Jason Calacanis (01:04:36): the two men hugging statue. Wait. Wait. No. Before you announce it, I need the extremely alpha male heterosexual trophy. The all in heterosexual alpha male trophy, please. And I need our four pictures to come on stage. Because we it may makes it more exciting. It's like makes it uncomfortable when, like, they show the five people for best actor. You put those on the table. But wait. Where's my award? Chamath Palihapitiya (01:05:12): You're doing full marketing. David Friedberg (01:05:13): You guys have the award. Jason Calacanis (01:05:14): Please bring me the extremely osexual alpha male award. You'll see why when I show you the award. Pessiness. Aaron Cowen (01:05:21): Slide over Lumer. Jason Calacanis (01:05:22): Alright. Bring me that award. Let me show you how we three d model this. David Friedberg (01:05:26): No one wants to see this award. Look at this. Jason Calacanis (01:05:28): Is two men uncomfortably hugging. And the way we did this Chamath Palihapitiya (01:05:32): See it. Jason Calacanis (01:05:32): Is the best he works. David Friedberg (01:05:33): Come here, Freberg. I'm not doing it with you. Come on, Freberg. You do it tomorrow. Freberg. Okay. Fine. Extremely Chamath Palihapitiya (01:05:38): comfortable. That's David and I. Jason Calacanis (01:05:40): It's David you, but let's show them how we modeled this. We just did a long This is uncomfortable, and we hold it for five extra seconds. At two minutes, you get the release of oxytocin, and there it is. Okay? So gentlemen, this is it. David Friedberg (01:05:53): Do you guys have the results? Go ahead. Audience award. Jason Calacanis (01:05:55): Okay. Ready? Audience award. Chamath Palihapitiya (01:05:57): So based on 150 votes from the audience, do I just go four to one? Kyle Samani (01:06:03): Four to one. Four to one is more exciting. Chamath Palihapitiya (01:06:05): Okay. Fourth place with 5% of the vote was Kyle Simani. Jason Calacanis (01:06:09): Okay. Well done on the board. Chamath Palihapitiya (01:06:12): A very close second place. No. Third. Third. Third place with 21% of the vote. Oleg, Brock? Jason Calacanis (01:06:22): Leg. Oh, boy. We're closing in here. Here I go. Chamath Palihapitiya (01:06:24): Okay. Very dramatic. And with fifty fifty, 50% number two? I'm gonna go no. You say number one now. Jason Calacanis (01:06:33): Okay. Okay. Chamath Palihapitiya (01:06:33): No. Okay. Well okay. Sorry. Yeah. You're right. With 24% of the vote in second place, Aaron Cowan for MGM, number one with 50% of the vote, Dan Dreyfus. Hey. Wow. Jason Calacanis (01:06:48): Unbelievable. Give it up. Nicely done. Chamath Palihapitiya (01:06:51): Now the bestie Jason Calacanis (01:06:52): But hold on before you do the bestie. How do you feel right now having won this pass him the award. You guys look so uncomfortable. Yeah. Exactly. Gosh. It's really nice. Very But pass him his award for a second and let him hold it. Gavin Baker (01:07:05): Give an Academy Award. Thank everybody how you got to this place. Jason Calacanis (01:07:09): Say a few words. Dan Dreyfus (01:07:11): Got my award. I got my tequila. Yeah. Thank you. Jason Calacanis (01:07:14): There you go. Alright. Well done. Okay. Chamath Palihapitiya (01:07:17): Now. Okay. Jason Calacanis (01:07:18): Pass the award. 4321. Chamath Palihapitiya (01:07:20): It's relatively similar here. Fourth place was Kyle Simani. Okay. Third place was Oleg. Second place, Dan Dreyfus. First place, Aaron Cullen. Big upset. Flip the audience vote. Jason Calacanis (01:07:34): There you go. Alright. So MGM wins. Aaron Cowen (01:07:36): Alright. Okay. Thanks, guys. David Friedberg (01:07:37): This was amazing. Alright. Thank you all for participating. Thank you very much. Jason Calacanis (01:07:42): Thank you so much for coming, and we'll see David Friedberg (01:07:44): you Good night.

本集资源与出处