MGM, Talen Energy Top All-In Investment Pitch Competition

MGM与Talen Energy斩获All-In投资推介大赛头名 对冲基金经理强调MGM Resorts和Talen Energy是具有下行保护的优质股票标的，同时将生物技术与加密资产定性为高上行空间的“彩票型”投资。 本次大赛汇聚了四位低调的投资经理，推介跨越不同领域的极高确信度交易标的，涵盖从机构价值标的到投机性科技资产。Suvretta Capital的Aaron Cowen率先登场，阐述了MGM Resorts尚未被市场充分定价的国际资产潜力；随后，Bornite Capital的Daniel Dreyfus提出了关于Talen Energy借力核能与数据中心电力超级周期的宏观投资逻辑。转向生物技术领域，EcoR1 Capital的Oleg Nodelman概述了Actis Oncology靶向放射治疗平台的巨大上行空间；最后，Multicoin Capital的Kyle Samani介绍了GEODNET，这是一个颠覆传统地理定位基础设施的去中心化加密网络。本期节目以评审团点评和观众投票收尾，最终评选的最高荣誉由兼具稳定性的MGM与具备高增长潜力的Talen Energy平分秋色。 “最佳投资理念推介大赛”旨在重现Sohn基金会大会的精神，聚焦那些为有限合伙人创造了高额回报却缺乏广泛分销渠道的低调基金经理。 (0:00) Suvretta Capital的Aaron Cowen推介了MGM Resorts International，认为其与Caesars在拉斯维加斯大道上的双头垄断主导地位被市场低估，且得到了媒体大亨Barry Diller持股26%的有力支持。 (2:29) MGM拥有显著的未定价上行空间，源于其获得了将于2030年在日本大阪开业的大型赌场度假村的独家牌照；与澳门复杂的政治环境相比，日本是一个更为稳定的监管市场。 (5:02) MGM在迪拜持有一项零成本期权，已在一处新建酒店综合体中预留了30万平方英尺的毛坯空间，一旦当地博彩合法化，即可迅速改建为赌场。 (7:31) Bornite Capital的Daniel Dreyfus推介了Talen Energy，概述了当前的技术电力超级周期，指出对清洁基荷能源容量的需求正在爆发式增长，远超历史基准水平。 (13:04)

At an All-In pitch competition modeled after the Sohn Foundation conferences, four fund managers presented their highest-conviction long trades across hospitality, power infrastructure, biotechnology, and decentralized physical infrastructure. MGM Resorts: Value Realization and International Expansion (2:29) Aaron Cowan of Serreta Capital pitched MGM Resorts International as an undervalued asset with strong downside protection and long-dated optionality. While the market views MGM as a Las Vegas Strip operator, Cowan highlighted an active catalyst: media mogul Barry Diller accumulated a 26% stake (representing 80% of his net asset value) and tendered an offer for the entire company at $48 per share, while MGM concurrently repurchased 50% of its share float over six years. Cowan argued against accepting the offer, valuing baseline domestic operations and China assets in the low $60s per share. The thesis relies on two unpriced growth vectors: an exclusive license for an integrated resort in Osaka, Japan, and an unannounced option in Dubai. The Osaka project, scheduled for a 2030 opening, addresses a domestic Japanese gambling market estimated at $40 billion—surpassing Macau ($30 billion) and Las Vegas ($10 billion). MGM holds a 40% equity stake plus management fees, which Cowan expects will yield roughly $2 billion in EBITDA. In Dubai, MGM is constructing a non-gaming hospitality complex containing 300,000 square feet of undeclared shell space positioned for rapid conversion should regional gaming legalize following rival Wynn Resorts' development on Al Marjan Island. Cowan estimated fair value between $100 and $150 per share. Talen Energy: Clean Baseload and Replacement Costs (13:04) Daniel Dreyfus pitched independent power producer Talen Energy, framing the company as a pure play on an emerging electricity demand super-cycle. Dreyfus argued that after two decades of flat domestic demand driven by energy efficiency and manufacturing offshoring, baseload power consumption is detaching from historical GDP growth. Artificial intelligence workloads exacerbate these supply constraints, requiring round-the-clock power that grid operators cannot quickly build given multi-year infrastructure lead times, supply chain shortages in critical alloys, and regional deficits like PJM's projected 106-gigawatt requirement. Applying Sam Zell's distressed asset playbook, Dreyfus noted that Talen owns 2 gigawatts of nuclear and 6 gigawatts of natural gas baseload capacity. At an enterprise value of $25 billion, Talen trades at a steep discount to its estimated physical replacement cost of $45 billion. Financed with leverage, Dreyfus projected Talen can generate $50 per share in annual free cash flow simply by executing existing operations and contracts—trading at roughly 7 times cash flow versus a standard 15 times infrastructure multiple. If power pricing tightens or premium hyperscaler contracts expand, cash flow could reach $70 to $100 per share, implying equity upside between $1,050 per share and higher. Actis Oncology: Precision Targeted Radiopharmaceuticals (27:18) Oleg Nodelman of EcoR1 Capital presented Actis Oncology (AKTS), an early-stage biotechnology firm capitalized with a $1 billion market cap, $500 million enterprise value, and three years of cash runway following an oversubscribed IPO backed by Eli Lilly. Nodelman characterized cancer therapies as evolving from indiscriminate poisons like chemotherapy toward targeted radiopharmaceuticals—deploying micro-doses of radiation straight to cancer cells using proprietary mini-proteins that bind specifically to tumors while rapidly flushing from the body. Actis's lead clinical programs target Nectin-4 (prevalent in bladder cancer) and B7-H3 (broadly expressed across major solid tumors), with initial human data readouts slated for 2027. Nodelman highlighted structural defensibility: Actis relies on actinium-225 payloads derived from legacy Cold War nuclear waste products, a specialized supply chain unavailable to foreign competitors like China. While early-stage clinical investments carry binary development risks, Nodelman asserted that clinical success in either target would validate the underlying mini-protein platform, potentially scaling the firm to a $10 billion valuation ($200 per share). Geonet: Decentralized Infrastructure for Physical AI (40:19) Kyle Samani presented Geonet (GEOD), a decentralized physical infrastructure network providing Real-Time Kinematics (RTK) geolocation services. Unlike conventional GPS, which carries a 2-meter accuracy variance, RTK achieves 2-centimeter precision, a tolerance required for autonomous agricultural equipment, robotics, delivery drones, and high-definition vehicular mapping. Whereas legacy incumbents like Trimble, Topcon, and Leica collectively built 12,000 base stations over thirty years, Geonet bootstrapped over 24,000 stations in three years by incentivizing independent hosts with GEOD tokens, circumventing institutional capital expenditure. Geonet operates at an $11 million annualized run rate with recurring commercial customers including John Deere, TomTom, and DJI. Samani emphasized that 80% of top-line revenue is contractually mandated to purchase GEOD tokens from the open market via Solana smart contracts, effectively functioning as programmatic capital return. Trading at a fully diluted valuation of roughly $150 million, Samani argued the network's low-cost base stations make ground-based RTK vastly cheaper and less energy-intensive than proposed Low Earth Orbit satellite constellations. Panel Deliberation: Liquidity, Duration, and Final Rankings (63:19) Evaluating the pitches, panel members Chamath Palihapitiya, Gavin, and David contrasted institutional execution against early-stage liquidity. Chamath noted that while Actis and Geonet offer asymmetric upside, their limited capital absorption profiles make them operate like high-beta options where large funds cannot deploy significant capital without driving price impact. In contrast, MGM and Talen offer capacity for tens of millions of dollars with explicit downside floors. Gavin prioritized MGM due to Diller's existing cash bid, warning that Talen carries external political and regulatory exposure to federal power interventions and changing administrations. David noted Talen's valuation upside operates on an eight-year horizon vulnerable to interest-rate shifts on long-term power purchase agreements, whereas MGM's operational catalysts could play out in two years. In the final tally, the 150-person audience vote awarded first place to Daniel Dreyfus (Talen Energy) with 50% of ballots, followed by Aaron Cowan (MGM, 24%), Oleg Nodelman (Actis, 21%), and Kyle Samani (Geonet, 5%). The host panel reversed the top ranks in their official judgment, selecting Aaron Cowan's MGM pitch for first place, shifting Dreyfus to second, and retaining Nodelman and Samani in third and fourth. The pitches highlighted a clear bifurcation between high-capacity, downside-protected assets tied to traditional physical constraints and early-stage ventures with large potential upsides but significant binary trial or token liquidity risks.

在 All-In 播客举办的最佳投资点子路演竞赛中，四位基金经理分别推介了各自具备高确信度的标的，涵盖了博彩资产重估、电力基荷紧缺、靶向放射性药物以及去中心化高精度定位网络。 美高梅的隐性亚洲博彩与中东期权 (2:29) Serreta Capital 的 Aaron Cowan 指出，市场普遍将美高梅国际酒店集团（MGM）视为单纯的拉斯维加斯赌场运营商，但其真正的重估动力来自媒体大亨 Barry Diller 持股 26% 并开出每股 48 美元的全面收购要约，而美高梅过去六年已回购了半数流通股。Cowan 主张股东不应接受该报价，认为公司仅拉斯维加斯基准资产每股即值 60 美元。 增量价值的核心来自两项未被充分定价的资产：其一是美高梅持有 40% 股权、计划于 2030 年开业的日本大阪综合度假村，预计每年贡献约 20 亿美元 EBITDA，能够利用日本 400 亿美元的本土博彩基础并吸纳中国高端客流，该项目对应每股 50 美元价值；其二是迪拜在建综合体中预留的 30 万平方英尺博彩备用空间，若阿联酋实现博彩合法化，可为股价再添 40 至 50 美元，整体合理估值在 100 至 150 美元之间。 Talen Energy 与电力超级周期的重估逻辑 (13:04) Daniel 阐述了电力需求的历史周期，指出在经历二十年由节能技术和重工业外迁导致的零增长后，西方正因技术革命迈入电力超级周期。即使没有人工智能需求，未来 20 年的电力供应依然极度紧张，而 AI 数据中心相当于每吉瓦造价约 500 亿美元的现代能源精炼厂，将直接加剧短缺；例如 PJM 电网预测未来 10 年仅其管辖区域就需要新增 106 吉瓦发电容量，等同于日本全国的电力消耗。 Talen Energy 拥有 2 吉瓦核电与 6 吉瓦天然气基荷产能。Daniel 借用地产投资家 Sam Zell 低于重置成本买入稀缺硬资产的策略，指出 Talen 当前企业价值为 250 亿美元，但从零重建这套发电机组的物理重置成本高达 450 亿美元。即便公司维持现状、让现有的亚马逊数据中心合约自然放量，每年也能产生每股 50 美元的自由现金流；若按照基建资产常规的 15 倍估值乘数或进一步签署溢价购电协议（PPA），股价存在成倍甚至达到 1,050 美元的上涨空间。 Actis Oncology 与放射性药物的技术壁垒 (27:18) Eco R1 Capital 创始人 Oleg Nodelman 回顾了肿瘤疗法的演进历程，认为化疗和初代靶向药分别类似地毯式轰炸与精确制导武器，但癌细胞容易产生耐药突变，而放射性药物则如同微型无人机蜂群，通过微型蛋白递送细胞级放射性载荷直接破坏肿瘤内部。Actis Oncology（AKTS）拥有 10 亿美元市值与 5 亿美元企业价值，账上持有可支撑三年以上的充裕现金储备。 Actis 的先导临床项目靶向 Nectin-4（膀胱癌靶点）和广泛存在于实体瘤中的 B7-H3，临床数据预计于 2027 年公布，最早的数据读出可能在第一季度。Nodelman 强调，该公司的护城河不仅在于其微型蛋白对实体瘤的穿透力，还在于其采用源自美国 20 世纪五六十年代核计划废料提取的锕放射性同位素；由于中国历史核项目主要集中于浓缩铀和钚，缺乏相应供应链，使 Actis 在该领域几乎免受海外仿制竞争。若其中一个项目成功推向市场，其估值有望达到 100 亿美元（即每股 200 美元）。 Geonet 去中心化 RTK 网络的飞轮效应 (40:19) Multicoin Capital 前合伙人 Kyle Samani 推介了构建实体人工智能基础设施的加密网络 Geonet。传统 GPS 存在约 2 米的定位误差，而 Geonet 借助实时动态差分技术（RTK）将精度提升至 2 厘米，这是自动驾驶农业机械、无人机和消费级割草机器人所必需的基础设施。相较于传统巨头历时数十年仅建立约 1.2 万个基站且基建昂贵，Geonet 通过代币激励让独立运营者以数百美元自建屋顶基站，规模已迅速达到前三大传统竞争对手总和的两倍以上。 商业化方面，Geonet 的年化营收已突破 1,100 万美元，年增长率达到 3 倍，合作客户包括 John Deere 的自动化喷洒系统、大疆无人机以及为全球自动驾驶车企提供地图的 TomTom。根据机制，其 80% 的网络总收入（约 880 万美元）会通过 Solana 智能合约直接在公开市场上回购 GEOD 代币，剩余 20% 用于研发与业务扩张。Samani 表示，尽管目前全面稀释市值约 1.5 亿美元，但强劲的负客户流失率和资本效率使其成为被市场低估的现金流型加密资产。 评委评估：流动性约束与投资组合分歧 在路演后的评审环节中，Chamath 等评委从头寸管理和流动性角度进行了批判性分析。评委指出，Geonet 和 Actis 属于不对称回报但归零风险较高的“彩票型”投资，尤其是 Geonet 代币当前的流动性较浅，机构资金甚至难以在不引起价格大幅波动的情况下建仓 100 万美元；相比之下，Talen Energy 和 MGM 则具有坚实的下行保护，能够轻易承接数千万美元的机构配置。 针对具体下行风险，评委 Gavin 认为 MGM 拥有 Diller 现金要约的底部支撑，风险收益比最为优秀；而 Talen Energy 尽管商业逻辑扎实，但与数据中心高度绑定的长周期购电协议面临不可控的联邦政治、监管干预以及利率波动的敏感性。最终评选出现戏剧性反差：现场观众投票中，Talen Energy 的演讲者 Dan Drafus 以 50% 得票率斩获第一；而在“Bestie”专家评委团投票中，推介 MGM 的 Aaron Cowan 逆转夺冠。 本场路演展示了从传统重资产重估到前沿科技基础设施的不同投资范式，但讨论同时表明，高回报潜力往往伴随着流动性容量限制、长期监管博弈或二元临床风险，资金在选择纯粹成长弹性还是兼具下行保护的防御资产上存在鲜明分歧。

## Background 本期视频由知名播客《All-In Podcast》于 2026 年 6 月 27 日发布在 YouTube 平台上。视频记录了 All-In 首届“最佳投资构想路演大赛（Best Ideas Pitch Competition）”，邀请了来自对冲基金、风险投资及加密领域的四位顶尖投资经理现场分享各自最具前瞻性的重磅持仓逻辑，并由 Besties 主持团与现场观众共同评选最终赢家。 ## TL;DR 四位投资人分别展示了横跨传统资产与前沿科技的核心标的：Suvretta Capital 的 Aaron Cowen 剖析了博彩巨头美高梅（MGM Resorts）在回购、收购要约及日本大阪与迪拜项目带来的多倍增量价值；Bornite Capital 的 Dan Dreyfus 阐述了 AI 数据中心供电荒下泰伦能源（Talen Energy）被严重低估的基荷电力与核能资产；EcoR1 Capital 的 Oleg Nodelman 推介了专注微型蛋白放射性药物的抗癌先锋 Aktis Oncology；Multicoin Capital 的 Kyle Samani 则展现了利用 DePIN 去中心化模式颠覆传统高精度 RTK 定位网络的 GEODNET。在最终评选中，Dan Dreyfus 斩获现场观众票选第一，而 Besties 主持团则通过综合风险收益比的大逆转，将全场冠军授予推介美高梅的 Aaron Cowen。 ## Atmosphere & style 整场活动洋溢着高水准华尔街顶级对冲基金研讨与硅谷科技沙龙交织的独特氛围，既有极为硬核的财务建模、重置成本测算与供应链博弈，又充斥着 All-In 招牌式的幽默互怼与现场“即兴下单建仓”的狂欢感。台上台下在严肃的投资推演与恶搞奖杯颁奖的松弛感之间自如切换，不仅充斥着极度乐观的冒险精神与智力激荡，更让观众沉浸式体验到真金白银博弈与思想碰撞带来的肾上腺素飙升。 ## Bulletin - [00:00](https://www.youtube.com/watch?v=fO5sC7qS04E&t=0s) All-In 播客举办首届最佳投资构想路演（Best Ideas Pitch），四位顶尖基金经理分别推介了美高梅（MGM）、泰伦能源（Talen Energy）、Aktis Oncology 以及去中心化定位网络 GEODNET。 - [01:06:22](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3982s) Bornite Capital 的 Dan Dreyfus 凭借泰伦能源（Talen Energy）拿下 50% 观众票选夺得观众选择奖第一名，强调 AI 数据中心将对早已紧俏的电网造成极端的超级电力周期缺口。 - [01:07:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=4038s) 四位 Besties 评委出现大逆转评选，最终将冠军授予推介美高梅（MGM）的 Aaron Cowen，看重 Barry Diller 要约收购提供的下行安全边际与日本大阪、迪拜项目的翻倍期权。 - [54:50](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3290s) 评委将四个标的精辟拆解为两大组合：具备数十亿美元承载力与强现金流的“基石重仓股”（MGM 与 Talen），以及具备非对称高赔率但流动性受限的“彩票型标的”（Aktis 与 GEODNET）。 ## Key takeaways - [00:00](https://www.youtube.com/watch?v=fO5sC7qS04E&t=0s) Chamath 创设该赛事的灵感源于索恩投资大会（Sohn Conference），旨在为公开市场顶尖基金经理提供摆脱传统财经媒体束缚、展现颠覆性大仓位逻辑的舞台。 - [02:29](https://www.youtube.com/watch?v=fO5sC7qS04E&t=149s) Aaron Cowen 推介美高梅（MGM），指出传媒大亨 Barry Diller 已持有 26% 股份并发起 48 美元要约收购，而公司在过去 6 年已回购注销近 50% 流通股。 - [05:00](https://www.youtube.com/watch?v=fO5sC7qS04E&t=300s) 美高梅手握两大隐藏王牌资产：预计 2030 年开业的日本大阪综合度假村独占 400 亿美元存量市场，以及迪拜滨海综合度假体暗中预留的 30 万平方英尺潜在赌场空间。 - [13:07](https://www.youtube.com/watch?v=fO5sC7qS04E&t=787s) Dan Dreyfus 提出“电力周期的结构剖析”，认为即便没有 AI，未来二十年电力市场也将极度紧缺，而 AI 的爆发直接给该趋势加装了涡轮增压。 - [16:39](https://www.youtube.com/watch?v=fO5sC7qS04E&t=999s) 数据中心物理与经济本质即当代的“炼油厂”：每吉瓦投入高达 500 亿美元，输入电力并输出 Token 与机器智慧，受制于特种合金等严峻供应链瓶颈。 - [19:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1158s) 超大规模云服务商陷入抢电恐慌，以微软为代表的巨头甚至主动以每兆瓦时 100 美元的翻倍溢价签署长期 PPA，推动三哩岛核电站历史性重启。 - [21:12](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1272s) Talen Energy 坐拥 2GW 核电与 6GW 天然气基荷产能，以 15 倍估值倍数测算，基准躺平情景现金流即对应翻倍目标价，自建产能情景下每股 FCF 可超 100 美元。 - [27:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1638s) EcoR1 Capital 的 Oleg Nodelman 秉持冷酷的德州扑克安全边际策略，推介放射性药物研发平台 Aktis Oncology，视其为抗癌“微型无人机”。 - [30:50](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1850s) Aktis 研发搭载锕-225 的微型蛋白平台，爆炸半径仅 100 微米实现单细胞级定向爆破，并依靠美国战后遗留的特殊同位素供应链建立了极高的海外仿制壁垒。 - [41:07](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2467s) Kyle Samani 推介 DePIN 项目 GEODNET，依托去中心化基站众包飞轮，将卫星定位精度由传统 GPS 的 2 米跃升 100 倍至 2 厘米级的 RTK 水准。 - [45:57](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2757s) GEODNET 已成长为全球最大的 RTK 网络，其 ARR 达 1100 万美元且 80% 的营收协议规定用于在公开市场透明回购代币，深度绑定无人机、自动驾驶与精准农业。 - [54:50](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3290s) Chamath 引用斯坦利·德鲁肯米勒法则进行仓位复盘：MGM 与 Talen 具备充沛流动性可承载数千万美元重仓，Aktis 与 GEODNET 则是面临流动性或研发风险的高赔率期权。 - [01:01:00](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3660s) 评委深入比较估值安全垫与兑现周期：美高梅下行被收购价封死且具备 2 年 3 倍潜力；Talen 变现周期长达 8 年且受制于利率与政策监管风险。 - [01:06:22](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3982s) 观众投票中 Dan Dreyfus（Talen）以 50% 得票率斩获第一，但最终 Besties 评委将桂冠颁给风险收益比极具确定性的 Aaron Cowen（MGM）。 ## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=fO5sC7qS04E&t=0s) 创设 Best Ideas Pitch 的渊源与灵感 Chamath Palihapitiya 首先向听众阐述了这个最佳投资构想路演（Best Ideas Pitch）环节的创立初衷。他提到了投资界著名的索恩投资大会（Sohn Investment Conference）——该大会由已故投资人 Ira Sohn 的家人创立，旨在为抗击癌症筹款。多年来，索恩大会在纽约林肯中心举办，邀请全球顶尖基金经理登台分享最具前瞻性的个股策略。Chamath 回忆道，自己 2015 年作为一个年轻的风投人受邀登上林肯中心的舞台，大胆预言“亚马逊（Amazon）将成为一家市值达到一万亿美元的公司”，结果引得全场哄堂大笑；他的好友、著名对冲基金经理 David Einhorn 当时甚至反驳道：“我很清楚万亿美元公司是什么样子，这绝不是一家万亿级公司。”事实证明 Einhorn 彻底看走眼了，这是一笔极其出色的下注。 乘胜追击的 Chamath 在 2016 年推介了特斯拉（Tesla）的可转换债券，同样取得了辉煌战绩。到了 2017 年，他信心满满地准备奉献自己的“代表作（magnum opus）”，高呼“人工智能就是未来”，然而最终选中的标的却是 Box。Chamath 自嘲道：“要是我当年选的是英伟达（Nvidia），我现在就是索恩大会的传奇人物，早就直接退休了。”这次 All-In 举办 Best Ideas 比赛，正是希望复刻这种高水准的思想碰撞，让那些真正在公开市场为 LP 赚大钱、却鲜少在 CNBC 等传统财经媒体露面的顶尖基金经理走上前台，把他们的重磅持仓逻辑展现在大众面前。 ```mermaid flowchart TD accTitle: Chamath 的索恩大会推介历史与本次活动定位 accDescr: 展示 Chamath 在索恩大会的三次关键个股推荐演变，以及本次 All-In 最佳投资构想路演的创办目标。 subgraph SohnHistory["Chamath 索恩大会历史战绩"] A["2015: 预测 Amazon 达万亿市值<br/>（遭全场嘲弄，但预测成功）"] --> B["2016: 推介 Tesla 可转债<br/>（大获成功的交易）"] B --> C["2017: 押注 AI 未来，误选 Box<br/>（若选 Nvidia 即可封神退休）"] end subgraph BestIdeasGoal["All-In Best Ideas Pitch 目标"] D["顶级基金经理"] --> E["摆脱传统财经媒体限制"] E --> F["展现颠覆性投资逻辑与高赔率标的"] end SohnHistory -.-> BestIdeasGoal ``` --- ### [01:25](https://www.youtube.com/watch?v=fO5sC7qS04E&t=85s) 参赛选手先导片：重塑金融、逆向生物科技与核能复兴 随后现场播放了一段节奏明快的选手推介片，4 位来自不同投资领域的基金经理依次登场，亮出鲜明的投资哲学： - 有加密与金融科技领域投资人提出未来愿景：“世界上任何人在任何地方、任何时间（24/7），仅凭一部手机和网络连接就能交易任何资产，我们正在从底层重建一套全新的金融系统。” - 多空股票基金经理表示，未来几年各行各业将迎来产品与应用层面的爆炸式创新，这正是对冲基金做多优质资产、做空落后标的以“赚大钱”的黄金期。 - 来自旧金山 EcoR1 Capital 的 Oleg Nodelman 则聚焦于生物科技领域特立独行的“捡漏”战术，专门搜寻那些“无人关注、不受待见、被市场严重误解的生物医药企业”。 - 宏观与能源赛道的投资人则指出，跨党派对于发展核能已经形成不可逆转的共识，当下正是重仓这类硬资产的历史性拐点。全体参赛者均展现出极度乐观与“顶格承担风险（maximum risk on）”的进攻姿态。 --- ### [02:29](https://www.youtube.com/watch?v=fO5sC7qS04E&t=149s) Suvretta Capital 的 Aaron Cowen 登场：为何在科技峰会上推介 MGM Suvretta Capital Management 创始人兼首席投资官 Aaron Cowen 作为第一位选手登台路演。Cowen 掌管着 40 亿美元规模的纽约基金，行业资历极深：他曾掌管乔治·索罗斯（George Soros）的股票业务，随后担任史蒂夫·科恩（Steve Cohen）的首席投资官，在对冲基金行业深耕长达 29 年。作为一个全行业通吃（generalist）的基金掌舵人，Cowen 直言，虽然自己也重仓科技股，但在这个科技大佬云集的峰会上推介科技股“简直愚蠢透顶”；而峰会的另一大招牌主题是德州扑克，因此他决定推介博彩巨头——美高梅度假村国际集团（MGM Resorts International）。 大多数投资者对 MGM 的认知局限于拉斯维加斯，MGM 与凯撒娱乐（Caesars Entertainment）是赌城最大的两家资产持有者，其中 MGM 在拉斯维加斯大道拥有 13 家超级赌场物业。前不久凯撒娱乐传出被收购传闻，也印证了赌城基本面的强劲回暖。但 Cowen 强调，自己看好 MGM 绝非仅仅因为拉斯维加斯，而是因为公司内部正在发生不同寻常的资本运作： - **巴里·迪勒（Barry Diller）的激进建仓**：传媒大亨 Barry Diller 近期大举扫货，目前已持有 MGM 26% 的股份，这笔持仓甚至占据了其控股旗舰 IAC 资产净值（NAV）的整整 80%。 - **突发要约收购**：就在路演的前一天，Barry Diller 正式向 MGM 发起了每股 48 美元的全面收购要约。两周前 Cowen 准备 PPT 时股价还在 37 美元，如今已飙升至 40 美元高位。现场主持人 Jason Calacanis 随即幽默插话：“我们提前拿到 PPT 了吗？拿到得够早去买股票吗？” - **公司史上罕见的激进回购**：除 Diller 外，MGM 自身在过去 6 年内回购注销了将近一半（50%）的流通股。Cowen 表示自己极少在美股市场看到回购力度如此惊人的公司。 - **坚决不卖给 Diller**：Cowen 呼吁所有股东绝对不要以 48 美元的价格把筹码贱卖给 Diller，“连一秒钟的考虑都不要有”。 --- ### [05:00](https://www.youtube.com/watch?v=fO5sC7qS04E&t=300s) 隐藏资产一：日本大阪赌场的巨大潜力 Cowen 剖析道，大众总以为 Barry Diller 只是当年 ABC 电视台的制作人或者打造了 Expedia 的互联网老兵，如今重仓下注赌场看似不可思议，但其背后其实深藏着市场未能充分定价的两大隐藏资产。 按照最保守的分部估值法（Sum-of-the-parts, SOTP），单看拉斯维加斯加上澳门博彩资产，MGM 的股票就值 60 美元出头，远高于 Diller 提出的 48 美元收购价。而第一大被忽视的隐藏王牌，是 MGM 手握的**日本大阪综合度假村（IR）特许经营牌照**： - **历史性政策开放**：日本数年前在全国各都道府县进行博彩公投，最终唯一通过设立赌场的行政区就是大阪。MGM 的综合度假区预计于 2030 年正式开业。 - **惊人的本土博彩基数**：许多人忽视了日本是极度狂热的博彩大国，仅凭柏青哥（Pachinko）与赛马市场规模就达 400 亿美元；作为对比，澳门博彩市场约为 300 亿美元，而拉斯维加斯仅为 100 亿美元。 - **财务回报与地理优势**：MGM 拥有该物业 40% 的股权并收取高额管理费，保守估计该项目能带来高达 20 亿美元的税息折旧及摊销前利润（EBITDA）。大阪的地理位置极具战略意义：从上海飞大阪的时间比飞澳门和新加坡还要短；从北京飞大阪的距离与澳门相当，远短于飞新加坡。 - **第一世界法治溢价**：与近年来面临估值倍数压缩和合规风险的澳门相比，日本作为发达经济体，法治透明度极高，资本愿意给予更高的市盈率倍数。Cowen 认为仅日本这一项增量期权，就能让 MGM 的内在价值“再翻一番（增值约 50 美元/股）”。 --- ### [07:32](https://www.youtube.com/watch?v=fO5sC7qS04E&t=452s) 隐藏资产二：迪拜的隐蔽布局与估值翻倍空间 第二张极具想象力的王牌则隐藏在中东阿联酋——**迪拜项目（The Island）**。MGM 正在迪拜开发包含阿丽拉（Aria）、美高梅（MGM）与贝拉吉奥（Bellagio）三大豪华品牌的滨海综合度假体： - **偷渡的赌场空间**：尽管迪拜目前在法律层面上严禁赌博，但 MGM 在建筑设计中极其隐秘地“偷偷预留了 30 万平方英尺的未指定功能空间”。只要迪拜监管松绑博彩合法化，这块区域一夜之间就能变成顶级赌场。 - **中东博彩军备竞赛逼近**：永利度假村（Wynn Resorts）耗资数十亿美元的综合度假村预计两年后在哈伊马角（Ras Al Khaimah，阿联酋境内，距离迪拜车程约 45 分钟）开门迎客。由于去往哈伊马角交通极其不便，迪拜绝不可能眼睁睁看着博彩客流和数十亿美金流失。为了在地区局势与金融竞争中重塑霸主地位，迪拜放开博彩几乎是必然之势。 - **SOTP 终极测算**：拉斯维加斯和中国澳门底仓资产值 60 美元，日本大阪项目值 50 美元，一旦迪拜赌场牌照落地，又将带来 40 至 50 美元的额外增量。因此，MGM 的潜在价值在 100 至 150 美元之间，相较起步价是一笔稳健的三倍股（Triple）投资机会。 - **资本博弈结局**：Barry Diller 不是产业买家，而是一个“想赚大钱的纯粹财务买家”，这表明 MGM 已经彻底处于公开竞购的“在局状态（in play）”。Cowen 再次强调现阶段风险收益比极其优秀，现有股东切勿交出筹码。 | 业务分部 / 潜在增量 | 预估每股贡献价值 | 关键驱动因素与催化剂 | |:--- |:--- |:--- | | **拉斯维加斯核心资产 + 澳门持股** | 约 $60 / 股 | 赌城复苏势头强劲、行业整合溢价、过去 6 年注销 50% 流通股 | | **日本大阪综合度假村 (IR)** | 约 $50 / 股 | 2030年开业；独占 400 亿美元本土存量市场；20 亿美元 EBITDA 预期；中国一线城市短途航程覆盖 | | **阿联酋迪拜项目期权** | $40 - $50 / 股 | 30 万平方英尺预留赌场空间；永利哈伊马角项目倒逼迪拜博彩合法化 | | **总体估值目标** | **$100 - $150 / 股** | **三倍看涨空间**（Barry Diller 提出 $48 要约，构成极佳安全边际） | --- ### [09:36](https://www.youtube.com/watch?v=fO5sC7qS04E&t=576s) 问答环节：非博彩娱乐资产变现与跨国信贷扩张 路演结束后进入评委与嘉宾提问环节。为了提高效率，现场决定连续提出两个问题让 Cowen 一并作答： 1. **非博彩娱乐资产的变现空间**：有评委提出，据传 Barry Diller 深度参与了 MGM 董事会，并正花费大量精力重构物业的“娱乐消费场景”，因为他们发现非博彩娱乐价值被严重低估，通过娱乐业态变现能够大幅拉升单客消费额。 2. **跨国客户信用额度（Credit Lines）与客户留存**：另一位评委指出，当赌场业务跨国扩张到日本、中东等海外市场时，如何有效搭建和风控高净值赌客的信贷额度体系，这也是驱动高价值 VIP 客户持续回流的核心壁垒。 Cowen 率先解答了信贷与客户黏性问题：MGM 拥有全世界最顶级的赌客数据库体系与全球会员忠诚度积分网络。拉斯维加斯常年沉淀了大量来自中国本土、日本以及亚太其他地区的超级 VIP 赌客，无论是在日本大阪开业还是在中东布局，MGM 根本无需从零拓客，直接调用既有的客户网络与成熟的跨国信贷追踪系统即可完成精准转化。 ### [10:33](https://www.youtube.com/watch?v=fO5sC7qS04E&t=633s) 美高梅投资逻辑答辩：大阪赌场催化与迪拜的“免费期权” 在问答环节的开头，发言者先分享了一段早年在拉斯维加斯投资里约酒店赌场（The Rio）的惨痛教训：当时凯撒娱乐（Caesars）与埃尔多拉多（El Dorado）合并，被迫剥离资产，他和朋友以每平方英尺仅 200 美元的极低价格买下了里约酒店，却忽略了一个致命问题——脱离凯撒体系后，该物业瞬间失去了凯撒庞大的会员忠诚度计划（Loyalty Program）支持。随后现场观众与评委抛出了两个核心质询：第一，凯撒娱乐此前苦等牌照未果最终退出了迪拜，美高梅凭什么能有所不同？第二，日本大阪赌场早在 2023 年就已获批，为什么市场在此次竞标前一直对这一隐藏资产视而不见？ 对此，阿隆·考恩（Aaron Cowen）凭借近 30 年的从业经验作出了回应。他指出，以永利度假村（Wynn Resorts）进军澳门为例，市场通常只会在实体项目正式开业前大约 3 年才开始真正计入其估值，而目前大阪项目恰好步入了这关键的“3 年倒计时”窗口期，此时介入正是最具性价比的时机。针对迪拜的牌照风险，他强调美高梅持有的是一个“免费期权”（Free Option）：整个项目是由合作方出资建设并由美高梅代为运营，且美高梅极其精明地预留了 30 万平方英尺的闲置空间作为未来赌场的备用场地。“如果博彩牌照落地，你的投资能翻三倍；即使牌照没有拿到，光凭现有业务也足以让你赚上一倍多。” 这一精彩的论述赢得了全场掌声。 - **里约酒店（The Rio）的前车之鉴**：以 $200/平方英尺的超低估值购入，却因失去凯撒忠诚度网络而陷入被动。 - **市场定价时滞规律**：大型海外赌场（如当年的永利澳门）通常在开业前 3 年才被资本市场充分 Price-in，大阪项目当前正处于估值重塑拐点。 - **迪拜项目的轻资产结构**：由合作方全额出资兴建，美高梅负责运营并预留 30 万平方英尺博彩空间，下行有保障、上行附带三倍回报的免费期权。 --- ### [13:07](https://www.youtube.com/watch?v=fO5sC7qS04E&t=787s) 丹·德雷福斯登场：电力周期的演变与“山姆·泽尔法则” 博奈特资本（Bornite Capital）的丹·德雷福斯（Dan Dreyfus）随即登台推介泰伦能源（Talen Energy）。他首先从底层逻辑梳理了“电力周期的结构剖析”（The Anatomy of a Power Cycle）。在常态下，电力需求的复合增速与 GDP 保持同步（即 GDP 增长 2%～3%，电力需求就增长 2%～3%）；随后会出现高耗能的技术突破引发电力需求陡增；当技术被全社会普及后，增速回归常态；接着整个社会进入以节能减排为核心的“能效优化阶段”（如 LED 照明、智能暖通 HVAC、隔热玻璃等技术的普及），同时伴随高耗能基础设施向海外转移。过去二十年里，由于大量电解铝厂等用能大户迁移至中国，美国经历了两整代人几乎“零电力需求增长”的平淡期；如今，随着新一代颠覆性技术的爆发，电力需求正从高基准再度大幅抬升。 ```mermaid flowchart TD accTitle: 电力周期演变模型 accDescr: 展示常态电力增长、颠覆性技术爆发、渗透回归常态与能效优化四个循环阶段 subgraph Phase1 ["第一阶段：常态与激增"] A["GDP 平衡期: 需求随 GDP 同步增长"] --> B["颠覆性高耗能技术爆发: 电力需求垂直跃升"] end subgraph Phase2 ["第二阶段：回归与重塑"] B --> C["技术全面普及: 增速回归基础算法"] C --> D["能效优化与外迁: 需求停滞二十年"] D --> E["新科技浪潮崛起: 开启新一轮超级电力周期"] end ``` 德雷福斯抛出了一个极为关键的论断：“我们根本不需要 AI 的需求爆发，未来二十年的电力市场本身就会极其紧缺；AI 的出现只是给这一趋势加装了涡轮增压（Turbocharges），直接制造了极端供需缺口。”他回忆起早年与传奇地产投资家山姆·泽尔（Sam Zell）在蒙古铜矿与地产论坛同台时学到的一课：当一项未来社会必不可少、且必须追加新建产能的核心重资产出现时，只要你能以远低于其重置成本（Replacement Cost）的价格折价买入并坚定持有，等市场猛然惊醒时，就能以大幅溢价卖出——这正是当年泽尔在 Equity Office Properties 上的经典战役。 - **泰伦能源（Talen Energy）核心资产**：手握 2 吉瓦（GW）核电基荷产能与 6 吉瓦（GW）天然气基荷产能。 - **巨大的重置成本折价**：当前资本市场给予泰伦能源的企业价值（EV）约为 250 亿美元，而其实际重置成本高达 450 亿美元；在杠杆债务效应下，仅回归重置成本就意味着股价有超过一倍的翻倍空间。 --- ### [16:39](https://www.youtube.com/watch?v=fO5sC7qS04E&t=999s) “数据中心即炼油厂”：中美电力竞赛与供应链危机 谈及美国的能源安全格局，德雷福斯毫不客气地呼吁：“让美国再次伟大（MAGA）的唯一路径，就是在电力规划上向中国看齐。”二十年前本轮周期伊始，美国的发电装机容量是中国的大约两倍；时至今日，中国的总发电装机规模已反超达到美国的三倍之多。在大国博弈的背景下，人工智能直接决定了前沿科学突破、具身智能机器人所转化的全要素生产率，乃至国家安全与军事霸权的生死存亡。“你要么拥有它，要么直接出局。” 为了让普通投资者理解这一赛道的本质，大宗商品出身的德雷福斯提出了一个精妙的比喻——“数据中心在物理与经济逻辑上就是一座当代的炼油厂”： - **传统石化炼油厂**：重资本投入，输入原油（Crude Oil），经过裂解提炼，末端输出供汽车和飞机燃烧的汽油或航空煤油。 - **当代 AI 数据中心**：每吉瓦投入高达 500 亿美元的超重资产，输入电力（Electricity），经过算力吞吐与计算，末端输出光子（Photons）、Token 或所谓的机器智慧。 英伟达 CEO 黄仁勋近期直言人类未来需要的电力是现有规模的“1000 倍”。德雷福斯指出，若该判断哪怕有一丝成真，我们就必须倾尽所有能源解决方案——不管是数百吉瓦的核电、光伏还是太空轨道能源基地。但现实的致命瓶颈在于物理层面的供应链：数据中心、商业航天、火箭发射和尖端发电机组，全部在争夺同一批关键矿产与高精材料，例如耐超高温的镍基高温合金（Nickel Superalloys），以及光伏电池必须耗费的巨量白银，材料的绝对短缺与工程周期的全线延误已无法避免。 --- ### [19:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1158s) PJM 电网的迫切缺口与超大规模科技巨头的抢电恐慌 德雷福斯将焦点拉回具体区域——涵盖宾夕法尼亚、新泽西与马里兰等地的美国中大西洋电网组织（PJM）。根据区域电网运营商的最新测算，仅 PJM 这一个局部区域在未来 10 年内就需要新增惊人的 106 吉瓦（GW）装机容量。在习惯了“按一下手机屏幕外卖和专车就能立刻送达”的互联网时代，人们往往忘记了构建重工业基础设施是按照“地质时间”（Geological Time）来推进的；在工业建设维度，10 年不过是一眨眼的纳秒瞬间，要在 10 年内平地起建 106 吉瓦——相当于整个日本全国当下的总用电负荷。 在这一残酷现实面前，既定的退役计划已成空谈：“原本计划退役的燃煤发电机组绝不可能如期关停，因为如果不保留它们，世界上没有任何可能在 10 年内填补上 100 吉瓦的电力真空。”深谙底层工业制造的人群目前已经陷入了集体恐慌，因为市场上根本没有足够的原材料去支撑如此庞大的装机增量，这意味着现有的存量电厂与基荷电价将在极长周期内保持极度紧俏。 为了争夺生存权，微软（Microsoft）、亚马逊等超大规模云服务商（Hyperscalers）正不惜一切代价疯抢基荷电力，试图通过锁定长达 20 年的长期购电协议（PPA）固定电价。以一向标榜“绿色零碳”的微软为例，为了抢占确定性电力，甚至主动登门说服了星座能源（Constellation Energy），促使其重启因严重堆芯熔毁事故导致核电产业冰封三十年的三里岛（Three Mile Island）核电站。为了撬动对方掏钱复活这座具有历史阴影的核反应堆，微软直接开出了极具吸引力的溢价——在基准电价仅为 50 美元/兆瓦时的情况下，微软承诺以每兆瓦时 100 美元的翻倍天价长期包销。 ### [21:12](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1272s) 算账 Talen Energy：三种情景与自由现金流估值 Dan Dreyfus 进一步回顾了科技巨头买电的标志性事件：为了重启三哩岛（Three Mile Island）核电站，微软 Azure 愿意签署长达 20 年的最低保底购电协议。然而，眼下这类交易正变得愈发艰难——监管机构已经开始介入质问：“等等，如果你们把电网里的电全抽走给数据中心用，普通居民冬天的取暖问题怎么解决？”能源供给正步入“紧要关头（crunch time）”。 回到对 Talen Energy 的估值测算上，Dan 给出了极其清晰的三种递进情景： ```mermaid flowchart TD accTitle: Talen Energy 投资回报与三种业务演进路径 accDescr: 展示 Talen 从“躺平”到溢价签约再到自建电厂的三种情景下每股自由现金流及目标股价估算。 subgraph BaseCase ["基准情景：原地躺平 (Do Nothing)"] A["仅维持现状运营<br/>推进既有亚马逊数据中心合同"] --> B["每股自由现金流: $50/年"] B --> C["给予 15x 基础设施估值倍数<br/>对应股价: ~$750 (当前 ~$370 翻倍)"] end subgraph ExpansionCase ["扩展情景：更高电价与新合同"] D["PJM 电价上涨<br/>以溢价签署更多数据中心合同"] --> E["每股自由现金流: $70/年"] E --> F["给予 15x 估值倍数<br/>对应股价: ~$1,050"] end subgraph GrowthCase ["超额增长情景：自建电厂"] G["与数据中心合作自建发电产能<br/>承接全美 100GW 缺口中的 4GW"] --> H["每股自由现金流: $100+/年"] H --> I["长期股价潜力具备巨大想象空间"] end ``` - **基准情景（“躺平”模式）**：哪怕管理层什么都不做，仅仅坐着维持公司日常运营，让现有的亚马逊数据中心合同逐步释放，Talen 每年也能产生每股 50 美元的自由现金流（FCF）。当前股价在 300 多美元高位（约 370-380 美元），对应估值仅约 7 倍现金流。而美国优质基础设施资产的交易倍数通常在 15 倍左右，这意味着即使管理层无所作为，投资者也能获得翻倍回报。 - **扩展情景（溢价签约与电价上涨）**：在 Talen 运营所在的 PJM 电网区域，目前的电价甚至低到无法刺激新建发电产能，这非常不可思议。若电价出现温和上涨，且公司能继续以溢价与数据中心达成更多协议，其年度经常性自由现金流将跃升至每股 70 美元。按 15 倍估值倍数计算，目标价可达 1,050 美元。 - **自建电厂情景（成为新增产能主力）**：目前监管方正把发电商与数据中心召集到同一个房间里，促成双方达成协议——让发电商获得合理回报来兴建新电厂，同时数据中心获得稳定的电力供应。Talen 处于这一格局的核心领跑地位。如果全美未来所需的 100 GW 电力缺口中，Talen 哪怕只负责承建 4 GW，其每股自由现金流就能突破 100 美元以上。 --- ### [23:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1398s) 问答环节：监管阻力、过渡方案与估值倍数之争 在 Dan 结束演讲并打趣这“不是投资建议”后，现场评委与观众提出了三个极具分量的问题： **1. 居民电费上涨与 AI 政治化带来的监管风险如何化解？** Dan 指出，社会离不开 AI，因此问题必然会解决，而“办法总比困难多”。他以高速公路做类比：凌晨四点开在空旷的高速上，人会觉得修这么多车道完全是浪费；但一到早晚高峰，大家又会抱怨车道远远不够。电力系统也是如此，一天之中真正让电网承压的只有极少数高峰小时。目前的务实解决方案是：发电商与数据中心签署电力采购协议（PPA），但强制要求数据中心在后端部署大量储能电池以及调峰发电机组（peakers），以此扛过最极端的用电高峰。这是在新电厂建成前非常优秀的“创可贴式过渡方案”，既能保障数据中心运转，又可将民用电费的涨幅控制在可接受范围内。 **2. 投资逻辑是否必须依赖“表后共置”（Behind-the-Meter Colocation）获批？** 观众提问核心在于：这项投资押注的究竟是表后直连，还是仅仅押注“清洁、稳定、基荷电力”本身的极度稀缺？Dan 明确回答是后者。表后获批只是锦上添花：在零变动的最保守情况下，纯靠现有业务就能赚到每股 50 美元现金流；若能推进表后或更多表前（front-of-the-meter）项目，现金流会迈向 70 美元；只有在启动新建产能时，才会走向 100 美元以上。 **3. 来自燃料电池、燃气轮机乃至“轨道算力”的替代竞争？** Dan 认为眼下是“全部都需要（We need all of it）”。虽然燃料电池、卡特彼勒（Caterpillar）旗下 Solar Turbines 燃气轮机等设备的平准化度电成本（LCOE）高得离谱，但对于一座造价高达 500 亿美元的数据中心而言，资本方绝不可能容忍其闲置 3 年干等联合循环燃气轮机（CCGT）基荷电厂落成。为了不让算力资产空转，数据中心愿意不惜一切代价为过渡期能源买单。 **4. 基础设施化转型的终局估值倍数是多少？** 观众 Daniel Scherer 追问：如果 Talen 从市场化独立发电商（Merchant IPP）转向长期锁定合同的基础设施，EBITDA 需要有多少比例转为合同制才能迎来重估？Dan 解释道： - **已锁定合同的业务**现金流如同债券，因其高确定性通常仅比美债收益率高出微小的利差（若美债收益率为 5%，对应合理估值往往在 20 倍左右），享有很高的溢价倍数。 - **未锁定合同的市场化电力业务**则直面现货市场的波动，能见度低，估值倍数天然折价。 - 综合来看，合同化比例越高，整体倍数就越高；在简化模型中采用 15 倍的混合估值倍数，是一个兼顾保守与合理的基准法则。 --- ### [27:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1638s) EcoR1 Capital 登台：把生物科技当德州扑克而非老虎机 EcoR1 Capital 创始人兼常务董事 Oleg Nodelman 随后登场，推介放射性药物研发公司 Aktis Oncology。Oleg 率先展现了幽默感，调侃道自己是播客的忠实粉丝，并深知“科学角落（Science Corner）”环节经常让人听得头大，因此他特意调整了叙述方式，“连 David Sacks 都能听懂并且感兴趣——如果他没睡着的话”。主持台随即打趣配合称，Sacks 此时正好在打瞌睡，这恰恰是他面对科学话题时的常态。 Oleg 坦言，在宏观层面上，投资生物科技通常是个极糟糕的主意，“其糟糕程度大概介于投资电影、酒庄和 SPAC 壳公司之间”。这个行业往往更像赌场而非严肃的金融市场，涌入其中的大多数“行业游客”实际上是在玩老虎机，盲目下注于虚无缥缈的技术风口。 与依赖动量交易（Momentum Investing）、容易深陷于科学情怀的传统机构不同，成立于 2013 年的 EcoR1 始终把自己定位为冷酷的德州扑克选手，死磕**安全边际（Margin of Safety）**与风险收益比： - **拒绝科学执念**：基金刻意不让医学博士（MD）或哲学博士（PhD）掌舵，核心策略是“将其他小孩的科学课作业商业化变现（monetize other kids' science projects）”，避免因对科研细节产生情感共鸣而遮蔽理性的商业判断。 - **长期亮眼业绩**：Oleg 投身生科投资已有 25 年，早年在另一家基金工作 11 年后创办 EcoR1，从 1,300 万美元的微小规模起步，如今管理规模已达约 25 亿美元。自成立以来基金为投资者带来了 10 倍回报，年化收益率达 20%，并在封闭 4 年后近期首次重新开放接纳资金。 --- ### [29:08](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1748s) 癌症治疗的演进史：从地毯式轰炸到精准武器 Oleg 正式引出 Aktis Oncology，将其形容为癌症战线最前沿的尖兵。自 1970 年代初尼克松总统签署《国家癌症法案》（National Cancer Act）以来，医学界就一直使用军事术语来描述抗癌过程。这种战争隐喻极为贴切，因为抗癌与现代军事作战的目标如出一辙：**发现敌人、选出最致命的武器消灭敌人、并在过程中将附带平民伤亡降到最低**。 ```mermaid flowchart TD accTitle: 癌症治疗手段的历史演进 accDescr: 展现癌症疗法从早期的无差别破坏，逐步演进为精确制导与动员盟友的过程。 Stage1["早期手术与放疗：中世纪围城战<br/>夷平整座城堡、焚毁周围村庄，寄希望于将敌人埋在废墟下"] Stage2["化疗：地毯式毒气轰炸<br/>利用一战芥子气原理毒杀快速分裂细胞，同时误伤毛发、皮肤与骨髓"] Stage3["初代靶向疗法：GPS 制导导弹<br/>精确摧毁指挥中枢，但癌细胞会通过基因突变产生抗药性隐藏躲避"] Stage4["免疫疗法 (IO)：本土联盟作战<br/>不再直接派兵，而是激活人体本土盟友（T 细胞）对抗癌细胞"] Stage1 --> Stage2 --> Stage3 --> Stage4 ``` Oleg 梳理了人类对抗癌症战术的四个历史发展阶段： - **早期手术与传统放射治疗**：如同“中世纪的围城战（medieval siege）”，战术简单粗暴——直接夷平整座城堡并烧光周边的村庄，只盼望着敌人也一同死在瓦砾堆里。 - **化疗时代**：起源于第一次世界大战中对芥子气毒气特性的偶然发现，即该物质能够摧毁快速分裂的组织。由于肿瘤细胞分裂极快，医生便向患者体内灌注化学毒药，期望在杀死正常组织前先消灭肿瘤。然而，人体的毛发、皮肤、肠道和骨髓细胞同样具有快速分裂的特性，毒药无法分辨敌我，造成了极其惨烈的全身毒副作用。 - **第一代靶向治疗**：如同“GPS 制导武器”，医生不再地毯式轰炸每一个分裂细胞，而是识别出癌细胞的指挥控制中枢并加以精确摧毁。但正如战争中的敌人一样，癌细胞具有极强的适应性与隐匿能力，会演化出耐药性突变（resistant mutations）。 - **肿瘤免疫疗法（IO）**：约十年前进入临床应用。IO 改变了思路，不再直接由外部向体内投放部队，而是动员患者体内的“本土盟友”——即自身的免疫 T 细胞，借助人体内生力量去击杀肿瘤。 ### [30:50](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1850s) 现代放射性药物：抗癌微型无人机平台 Aktis Oncology 传统的癌症免疫疗法虽然在见效时表现亮眼，但极度依赖肿瘤微环境和生理地形。EcoR1 Capital 的创始人 Oleg Nodelman 由此引出他本次推介的核心主角——现代放射性药物（Radiopharmaceuticals）。他将这种疗法生动地比作一组在血液中穿梭的微型无人机：体积足够微小以在血管系统巡航，依靠分子识别技术精准锁定肿瘤靶点，随后引爆一颗爆炸杀伤半径仅有 100 微米（相当于单个细胞直径）的高精度弹头。这种疗法犹如实施了一场"具有钻地弹威力的自主暗杀行动"，既能以毁灭性力量摧毁癌细胞，又将周围健康组织的连带损伤降至最低。 Oleg 推介的公司是 Aktis Oncology（股票代码：AKTS）。该公司目前的市值约为 10 亿美元，企业价值（EV）约为 5 亿美元，账上储备的现金足以支撑其运营 3 年以上，完全能够覆盖关键临床里程碑。Aktis 成立于五年前，近期完成了一轮 3 亿美元的 IPO。该发行获得了 18 倍的超额认购，并由减肥药巨头礼来（Eli Lilly）领投 1 亿美元认购作为后盾支持。 Aktis 的核心技术在于设计了一个能够搭载任何放射性载荷的微型蛋白（mini-protein）平台。其分子结构既足够复杂，能精准识别各类癌症靶点；又足够轻巧微小，能迅速从人体内代谢清除，将全身性副作用降到最低。该方案最大的优势在于临床去风险能力：医生在早期临床试验中便可通过成像技术直观验证靶向结合情况（target engagement），直接确认药物是否真实送达肿瘤部位。 ```mermaid flowchart LR accTitle: Aktis Oncology 靶向放射治疗工作流程 accDescr: 展示微型蛋白搭载锕-225精准递送至肿瘤细胞并由成像技术验证的过程 subgraph S1["设计与组装"] A["微型蛋白平台 (Mini-protein)"] --> B["放射性弹头 (锕-225/Actinium-225)"] end subgraph S2["体内递送与成像"] B --> C["血液循环巡航 & 分子靶向识别"] C --> D["临床早期正电子/核医学成像验证"] end subgraph S3["精准爆破与清除"] D --> E["100微米单细胞级爆炸半径 (钻地弹杀伤)"] E --> F["快速肾脏代谢排出 (极低毒副作用)"] end ``` - **靶点布局**：Aktis 首批管线明智地选择了已验证的成熟靶点。首发管线靶向 Nectin-4（在膀胱癌中起关键作用）；第二条管线则更具雄心，靶向在绝大多数实体瘤（包括前列腺癌、结直肠癌和肺癌“三巨头”）中高表达的 B7-H3。 - **数据催化剂时间表**：公司已于去年启动临床试验，官方指引将在 2027 年公布这两个先导项目的初步临床数据，其中 Nectin-4 的读出最早可能在 2027 年第一季度公布。 - **平台价值溢价**：生物制药领域的“圣杯”在于平台溢价——一旦其中任何一条管线展现出疗效信号，市场的重估不仅局限于单个项目，更将辐射至其整个微型蛋白底层平台。 --- ### [33:04](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1984s) 制药巨头的并购狂潮、仿制药壁垒与“免责幽默” 大药企对放射治疗领域的并购胃口正在疯狂扩张。Bristol Myers Squibb（百时美施贵宝）、Novartis（诺华）、Bayer（拜耳）以及为 Aktis 站台的礼来，都在紧锣密鼓地搭建放射治疗产能，渴求吞并优质管线。仅过去几年，放射治疗领域的并购和交易总额就已达到 150 亿美元，而整个行业仍处于发展的“早期阶段”。 与常规生物药相比，该赛道拥有坚不可摧的护城河：仿制药商通常不涉足放射性药物，且由于其核心涉及特殊的放射性同位素供应链，对中国等海外仿制者存在天然壁垒，不易被反向工程或轻易复制。在估值探讨前，Oleg 模仿处方药广告的语速来了一段极具反讽幽默的“合规风险警告”，引得全场大笑： > “Aktis 并不适合所有人。购买 Aktis 前请咨询您的生物科技分析师。建立头寸可能会导致焦虑加剧、夜间失眠；生物科技股有时会出现断崖式暴跌；投资后可能会因缺乏竞争对手而遭遇剧烈波动。如果股价无故下跌，请立即致电您的经纪人加仓……微型蛋白向细胞递送技术尚未完全证实，Aktis 目前无任何商业化产品因此无经常性收入。若发生二次增发，请立即与 Aktis 管理层通话认购。” 对于估值难题，Oleg 坦言生物科技早期的现金流折现（DCF）模型在经过风险调整后很容易算成零。因此团队采用交叉比对法测算：只要 Aktis 旗下任何一条管线最终成功获批上市，该公司的估值就可能跃升至 100 亿美元，即对应每股 200 美元。 --- ### [34:41](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2081s) 现场激辩：同位素的地缘供应链壁垒与后 IPO 估值低迷 现场评委和嘉宾针对中国竞争以及专利问题展开了追问。有嘉宾质疑，在 Amgen 诉 Sanofi 案确定了物质成分专利（composition of matter patents）的高门槛后，很多生物药被海外公司通过修改单个氨基酸进行绕道仿制，放射性药物是否也会重蹈覆辙？ Oleg 详细解释了放射性同位素的独特地缘制造壁垒： Aktis 所采用的放射性同位素弹头是锕（Actinium，即锕-225）。锕必须通过镭-223（Radium-223）加工制备，而这是美国在 20 世纪 50 至 60 年代核计划留下的工业副产物。中国当年的核计划完全走的是浓缩铀和钚的路线，根本没有这套特定同位素的原料积累与供应链。由于必须在本土完成采购与放射性合成加工，目前在欧美市场完全看不到来自海外的实质性竞争。 关于为何公司 IPO 后股价表现平淡的疑问，Oleg 指出这属于典型的生物科技板块特征：二级市场投资者的持有周期极短。虽然距离关键数据读出仅有不到一年时间，但对短期交易者来说依然显得漫长。他预计机构投资者将在下半年开始针对次年一季度的数据读出逐步建仓。 --- ### [36:47](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2207s) 问答交锋：AI 赋能、人类寿命突破与实体瘤递送边界 问答环节延伸至更宏大的生物医药命题。前生物医药基金经理 Gavin 提出，过去 25 年人类基因组计划后的治疗突破并未达到千禧年预期的爆炸式增长，他询问在 AI 算力突破下，人类预期寿命是否能在这一代人眼中突破 125 岁甚至 150 岁。 Oleg 毫不犹豫地选择“押注超过这一数字（take the over）”。他解释称，目前人类其实已经手握最强大的长寿药物——以 GLP-1 为代表的减肥药。科学上唯一被数据严格证实的延寿机制就是热量限制（caloric restriction），而这类药物在生理上达成的正是这一效果。更重要的是，它改变了大众的心理认知，让人们接受通过规律的药物注射实现健康生活。 针对现场观众与嘉宾的后续技术问题，Oleg 也给出了明确回应： - **太空与微重力癌症治疗**：有观众询问发射成本下降是否会使微重力治癌成为可能，Oleg 直言该方向虽然很有趣，但并不适用于 Aktis 的研发逻辑。 - **内部 AI 团队运作**：Aktis 内部设有一个小型的类似“臭鼬工厂”（Skunk Works）的项目，由专门的博士数据科学家团队负责，利用专有数据挖掘微型蛋白的新构型与新靶点。 - **实体瘤穿透难题**：针对 CAR-T 等细胞疗法在实体瘤领域的停滞不前，Oleg 指出靶向疗法（如 Revolution Medicines 在胰腺癌上的进展）正成为新的中坚力量。而 Aktis 微型蛋白的核心优势，正是利用其恰到好处的分子尺度穿透实体瘤屏障，将放射性弹头安全运送至肿瘤核心。 ### [40:11](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2411s) 讲台交接与 Kyle Samani 登场：加密与 AI 的交汇点 在 Oleg 刚结束对 Aktis Oncology 的路演时，台上台下传来一阵骚动，有人在 Oleg 还在台上演讲时就已经“梭哈”（YOLO）买入，导致相关标的直接暴涨了 6%。主持人们开玩笑地抗议道：“别在我们大家也想买的时候抢跑啊，拜托！”随后，Multicoin Capital 联合创始人 Kyle Samani 走上讲台。Kyle Samani 在大约八年半前创立了知名加密投资基金 Multicoin Capital，并于数月前卸任。在风险投资生涯中，他最为人熟知的战绩便是在 2020 年 Solana 主网上线之前，连续领投了 Solana 的全部三轮融资。 Kyle 特别向当天未能到场的 David Sacks 致谢，感谢 Sacks 当年作为种子轮投资人出资支持了 Multicoin 的创办。Kyle 表示，这次在全流动性论坛上，他要推介的是一个相对小众却极其硬核的加密资产——GEODNET。该项目正处在加密货币（DePIN）与人工智能（AI）的黄金交叉点上，致力于为未来的物理 AI 和自主机器人系统搭建关键的基础设施轨道。 ### [41:07](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2467s) 从传统 GPS 到 RTK：厘米级高精度定位的技术飞跃 要理解 GEODNET 的价值，首先必须看清现有 GPS 技术的硬伤。几乎每个人都经历过手机导航失灵的窘境：定位点偏离了实际位置，导航箭头甚至显示人正对着一堵墙。普通的民用 GPS 定位精度大约只有 2 米，这对于人类找路或许勉强够用，但对任何精密机器人系统来说都远远不够。 GEODNET 采用的是一种名为 RTK（Real-Time Kinematics，实时动态差分定位）的技术。通过遍布地面的基站校准卫星信号，RTK 能够将定位精度大幅提升至 2 厘米级别。相较于传统 GPS，其精度实现了约 100 倍的跨越式提升。无论是农田作业的自主拖拉机、穿梭农场的运输车，还是空中飞行的无人机，任何形态的物理机器人都高度依赖 RTK 带来的高精度地理定位服务。 ```mermaid flowchart TD accTitle: RTK 与传统 GPS 精度及应用对比 accDescr: 对比传统 GPS 的米级精度与 RTK 厘米级精度，展示 RTK 支持的多类自主机器人场景。 subgraph Positioning["定位技术对比"] GPS["传统卫星 GPS 定位<br>精度: 约 2 米<br>(手机导航漂移/误差大)"] RTK["RTK 实时动态差分定位<br>精度: 约 2 厘米<br>(100 倍精度提升)"] end subgraph Robotics["物理 AI 与机器人应用场景"] Agri["精准农业<br>(John Deere GUSS / 农用运输车)"] AV["自动驾驶与高精地图<br>(TomTom 地图更新)"] Consumer["消费级机器人<br>(智能割草机器人)"] Drone["低空经济与无人机<br>(DJI 大疆等航拍与作业设备)"] end RTK --> Agri RTK --> AV RTK --> Consumer RTK --> Drone ``` ### [41:55](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2515s) 去中心化 DePIN 颠覆巨头：GEODNET 的全球闪电扩张 目前，GEODNET 已经是全球规模最大、增速最快的 RTK 网络。在它出现之前，Trimble（天宝）、Hexagon（海克斯康）和 Topcon（拓普康）三家传统行业巨头已经深耕 RTK 网络二三十年，但三家公司在全球部署的基站总数加起来也仅有约 12,000 个。而成立于 2021 年、从 2022 年才开始正式铺设网络的 GEODNET，其全球节点数已突破 22,000 个，达到了前三大传统巨头总和的两倍左右。如今，GEODNET 已覆盖全球 150 个国家、11,000 多座城市，触达了全球约 80% 的人口（除部分受制裁地区外）。 GEODNET 之所以能用四年时间完成传统巨头耗时三十年都未能达到的网络规模，核心秘密在于其去中心化的加密激励模型（DePIN，去中心化物理基础设施网络）： - **非中心化重资产投入**：网络不是由 AT&T 或 Verizon 这样的传统电信运营商耗费巨资统一建设，而是由全球各地的爱好者、专业技术人员或小型企业主自发部署。 - **低成本自购部署**：任何人都可以登录 GEODNET 官网，花费几百美元购买一个轻便的基站设备，将其安装在自家屋顶或商铺上方。 - **代币经济正向激励**：基站向周边广播无线电校准信号提供服务，部署者即可获得项目的 GEOD 原生代币作为报酬，以此驱动网络飞轮以极低成本迅速向全球扩张。 - **全美无死角覆盖**：在美国本土，GEODNET 不仅全面覆盖了每一个主要大都会区，甚至连绝大部分偏远农村和农田区域也实现了网络覆盖。 ### [43:34](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2614s) 核心落地用例：精准农业、自动驾驶、割草机与大疆无人机 随着计算机视觉、算力芯片和电池技术的进步，物理实体机器人正在爆发式增长，而 GEODNET 为这些机器人提供了底层定位基础设施： - **精准农业（Agriculture）**：美国农业部（USDA）数年前便启动了专项计划，鼓励农场主和牧场主采用高精度农业技术与 RTK 网络，目前 USDA 正在直接出资补贴农户采用高精度农业方案，其中绝大部分由 GEODNET 提供支持。实际落地案例包括由 Burro 制造的“机器骡”（用于葡萄园等农场搬运物资的小型自主车辆），以及农业机械巨头约翰迪尔（John Deere）推出的 GUSS（Global Unmanned Spraying Systems，全球无人喷洒系统）。Kyle 提到，在演讲当天的纳帕谷（Napa）酒庄中，就有实际运行的约翰迪尔 GUSS 农用车在利用 GEODNET 进行精准农药与养料喷洒。 - **自动驾驶（Autonomous Vehicles）**：全球导航电子地图巨头 TomTom 是 GEODNET 的正式客户。除极个别自研体系外，TomTom 为全球绝大部分自动驾驶研发项目提供高精地图数据，目前 TomTom 正全面采用 GEODNET 的实时差分数据来高频校准并更新其全球地图。 - **消费级机器人（Consumer Robotics）**：以智能割草机器人为代表的家用设备正在大规模放量，今年预计全球出货量将达 100 万台。包括 Yarbo、Sunseeker 等头部割草机品牌，其高精度定位能力均完全由 GEODNET 驱动。 - **无人机系统（Drones）**：全球最大的无人机制造商大疆（DJI）也是 GEODNET 的客户，其旗下众多无人机型号都在使用 GEODNET 的差分服务，为网络贡献了海量流量。Kyle 强调，随着美国本土无人机制造浪潮的兴起，基于供应链安全和本土化考量，未来新崛起的美国无人机厂商几乎毫无疑问也将全面接入 GEODNET。 ### [45:57](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2757s) 商业爆发力与股东回报：80% 收入用于公开市场代币回购 GEODNET 具备极为清晰的网络效应。从商业本质上看，它类似于一个天然的电信级基础设施垄断网络，而电信网络在历史上往往具备赢家通吃的垄断特性。得益于分布式个人节点的众包模式，GEODNET 不仅坐拥全球最大的规模和最快的增速，更享有同行难以企及的极低成本结构。 在财务表现和代币经济学设计上，GEODNET 展现出了惊人的自我造血与资本回报能力： - **营收规模与增速**：就在演讲前几天，GEODNET 的年化经常性收入（ARR）刚刚突破 1,100 万美元大关，年同比增长超过 3 倍（3x YoY），Kyle 预计未来 12 个月该业务将继续保持三倍以上的增速。 - **80% 现金流回购代币**：在约 1,100 万美元的年化收入中，有整整 80%（即每年约 880 万美元）被直接用于在公开市场回购 GEOD 代币。所有回购操作在 Solana 区块链上完全透明、地址公开，投资者可以进行 7x24 小时的实时链上验证。 - **极度轻量与高毛利**：剩余 20% 的收入便足以覆盖网络所有的底层研发支出，并支撑其商务拓展团队的扩张。由于农机、无人机、高精地图等行业的客户圈层重合度极高（约翰迪尔、大疆、TomTom 之间彼此熟识），业务在核心客户群中正通过口碑病毒式裂变。 - **客户高留存与三倍净扩张**：客户第一年接入 GEODNET 服务的平均客单价约为 60,000 美元；到了第二年，随着使用场景深化，其平均支出将飙升至 170,000 美元，单客收入净扩张接近 3 倍。在客户基数层面，GEODNET 去年的净增新客户数量增长了整整 5 倍。 Kyle 总结道，GEODNET 是一家集“全球最大且增速最快的 RTK 网络”、“天然的电信级垄断壁垒”、“物理 AI 与机器人浪潮顺风车”以及“扎实的现金流回购机制”于一身的罕见标的。该项目的原生代币 GEOD 在 Solana 区块链上提供 7x24 小时交易，投资者只需配置一个 Solana 钱包即可直接参与。 ### [48:28](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2908s) GEODNET 的代币经济学与企业架构：80% 现金流回购机制 Kyle Samani 结束宣讲后，现场随即进入问答环节。主持人和嘉宾首先追问 GEODNET 的估值水准。Kyle 解释称，按完全稀释估值（FDV）计算，GEOD 代币目前整体市值约为 1.5 亿美元。如果投资者通过 CoinGecko 或 CoinMarketCap 等主流加密货币行情网站查询，显示的流通市值大约只有 6000 万到 7000 万美元，这是因为大量代币尚处于锁仓或未释放状态，尚未进入二级市场流通。 面对现场对于加密项目是否属于“设立在开曼群岛或巴拿马、董事会成员无人知晓的皮包机构”的合规质问，Kyle 明确澄清了 GEODNET 的实体治理结构。该团队是一家总部位于美国的正规公司，核心团队常驻旧金山，CEO 是在物联网（IoT）与智能硬件领域深耕多年的连续创业者 Mike Horton。 在被问及投资者究竟应该买入股权还是持有代币时，Kyle 毫不掩饰地亮明自身持仓立场：“你应该买代币，因为正如你能想到的那样，我手里持有大量代币，而完全没有持有该公司的股权。”他详细拆解了企业与代币之间的价值传导逻辑：GEODNET 企业实体直接面向 John Deere（约翰迪尔）、DJI（大疆创新）等企业级 B 端客户签订商业合同；同时，该商业实体与非营利性的 GEODNET 基金会（Geodnet Foundation）签署了具有法律约束力的协议，必须将全部营收的 80% 拿出，直接在公开二级市场上回购 GEOD 代币。剩下的 20% 资金则保留在公司内部，用于支付工程师薪酬、销售开支和日常运营支持。 ```mermaid flowchart LR accTitle: GEODNET 业务与代币价值捕获飞轮 accDescr: 展示 B 端企业客户付款后，资金在 GEODNET 企业与二级市场代币回购之间的流动闭环。 subgraph Clients["企业客户层"] B["John Deere / 大疆 DJI 等"] end subgraph Entity["GEODNET 实体公司"] C["GEODNET 运营实体"] OPEX["20% 营收：研发与销售支出"] end subgraph Market["代币飞轮与基金会"] F["GEODNET 基金会"] BUY["80% 营收：公开市场回购 GEOD"] TOKEN["GEOD 代币持有人"] end B -- 支付服务费 --> C C --> OPEX C -- 80% 收入协议转移 --> F F --> BUY BUY -- 买压拉动 / 价值沉淀 --> TOKEN ``` ### [49:47](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2987s) 现场交锋：DePIN 赛道折戟魔咒与加密合规红线 主持人随后抛出线上观众的尖锐提问。首先是关于 Helium（去中心化 5G / 物联网网络）的对比。Kyle 坦承自己早在六七年前就主导了 Multicoin Capital 对 Helium 的领投，至今仍然是其坚定信仰者，并透露 Helium 当天早晨刚公布了重要利好。但观众更核心的疑虑在于：整个 DePIN（去中心化物理基础设施网络）赛道横尸遍野，绝大多数项目失败的原因都是“除了不可持续的代币通胀奖励外毫无实质价值”，GEODNET 凭什么能够幸免？ Kyle 强力反驳称，本质区别在于 GEODNET 是一个已经能产生真实利润并向持有人返还资本的实体业务。它当前以 1.5 亿美元的 FDV 交易，年化返还给代币持有人的资本就高达 880 万美元，而且今年预计将实现 3 倍的营收增长。在加密货币处于熊市周期、市场注意力严重匮乏的当下，这类资产在估值上显得“不可思议的便宜”。 现场评委一针见血地点出核心实质：“这本质上是对终端客户现金流的证券化权益，是一种 80% 的收入分成代币（Revenue Share Token）。”这意味着约翰迪尔等大客户采购服务的支出越多，公开市场上被买入并沉淀的代币就越多。对于观众 Sam 提出的《清晰法案》（Clarity Act）及潜在证券法监管风险，Kyle 坦言自己并非专职律师，无法从法理层面对条文作出定论，但他深度参与了相关立法倡导，心态保持乐观，并不为此过度担忧。 ### [51:30](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3090s) 技术路线大辩论：地基 RTK 站 vs. 低轨微卫星网络 讨论随后进入硬核技术层面的交锋。Jason Friedberg 结合自己早前管理农业高科技企业 Precision Planting 的行业经验提出技术质疑：在精准农业和建筑测绘领域，John Deere、Topcon（拓普康）、Leica（徕卡）和 Trimble（天宝）等巨头长期自己搭建专用 RTK 基站系统，为什么客户会放弃已有成熟设备转而依赖 GEODNET？此外，当前市场上正兴起利用 SpaceX 等火箭密集发射低轨微卫星（LEO Microsats）组建网状网络的浪潮，试图绕过传统地基网络，直接提供亚厘米级定位，这是否会彻底颠覆并淘汰地基基站？ Kyle 从商业财务模型与物理硬件极限两个维度进行了系统性防御： - **资本支出（Capex）转向运营成本（Opex）**：自建专用 RTK 基站是沉重的高额资本开支。GEODNET 依托全球分布式矿工节点，将服务资费直接降到了传统自建方案的三分之一甚至四分之一。这种“即买即用”的现成网络显著缩短了 John Deere 销售智能农机的成单周期。 - **制造成本与太空发射成本的鸿沟**：低轨卫星不仅制造发射成本极其高昂，单星寿命有限；而 GEODNET 的地面基站单台硬件仅需数百美元，在成本量级上太空方案根本无从竞争。 - **能耗与电池敏感型硬件的天然限制**：Kyle 特别指出终端设备的功耗问题。农用重型拖拉机或许不在意能耗，但对于无人机（Drone）、手持勘测仪或任何依靠电池供电的移动设备而言，接收高强度卫星微波所耗费的电量远高于向近处地面基站通信。在长寿命和轻量化场景中，地基站具有无法撼动的物理优势。 谈及去中心化算力等其他代币板块时，Kyle 表现得极其审慎。他直言自己对当前大热的“去中心化分布式模型训练”高度怀疑，认为技术路径极大概率无法跑通；而去中心化推理（Distributed Inference）虽然 Multicoin 几年前曾注资布局，但落地表现目前仅算尚可，远称不上顶尖。 ### [54:50](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3290s) Chamath 复盘：斯坦利·德鲁肯米勒法则与四大标的仓位艺术 随着四场路演全部结束，Chamath Palihapitiya 率先分享了他的宏观研判框架。他直言自己高度奉行传奇投资家斯坦利·德鲁肯米勒（Stanley Druckenmiller）的投资哲学：“投资必须有真金白银的切身利益（Skin in the game），否则你根本不会真正关心。” 在 Chamath 看来，四位投资人带来的都是极具吸引力的顶级构想，但作为资产配置者，真正的胜负手不在于选谁，而在于**仓位管理（Sizing）、流动性承载力与非对称阿尔法（Asymmetric Alpha）**的匹配： - **巨额仓位标的（MGM Resorts 与 Talen Energy）**：这两家企业具备庞大的二级市场流动性，可以容纳数千万美元的单笔资金下注而“市场连眼皮都不会眨一下”。无论是 Aaron Cowen 提出的 MGM 资产负债表与回购重估，还是 Dan Dreyfus 带来的 Talen 能源数据中心红利，都属于适合下重注的核心底仓。 - **高赔率彩票型标的（Aktis Oncology 与 GEODNET）**：这两处投资更类似于赔率极高的期权。Oleg Nodelman 推荐的核药研发公司 Aktis 存在流动性断层风险，研发失败可能血本无归，但一旦临床成功并被礼来（Eli Lilly）等医药巨头竞购，就是十倍以上的非对称回报；Kyle 的 GEODNET 逻辑非常漂亮，但当前代币市场流动性极浅，哪怕想建仓 100 万美元都会直接把盘面价格打飞，机构投资者只能以两三万美元的分批限额极其缓慢地逐步建仓。 | 演讲者 | 推介标的 | 核心投资属性 | Chamath 的仓位定性 | |:--- |:--- |:--- |:--- | | **Aaron Cowen** | MGM Resorts (MGM) | 传统博彩巨头、自由现金流回购、低估值蓝筹 | 重仓承载型资产（可容纳数千万美元级规模） | | **Dan Dreyfus** | Talen Energy (TLN) | 独立电力供应商、核电直连数据中心、AI 基建算力风口 | 重仓承载型资产（流动性极充裕） | | **Oleg Nodelman** | Aktis Oncology (私有) | 靶向α疗法（TAT）、偶联核素抗癌、大药企潜在并购 | 赔率型期权（面临二元结果与流动性锁死风险） | | **Kyle Samani** | GEODNET (GEOD) | DePIN 物理硬件网络、企业收入80%回购、厘米级 RTK | 赔率型期权（代币流动性受限，需小规模建仓） | ### [56:09](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3369s) 投资大乱斗的终局预热与 All-In ETF 畅想 主持人 Jason Calacanis 现场理清了本场“Best Ideas”大赛的四家对决标的：MGM Resorts、Talen Energy、Aktis Oncology 以及代表 Web3 去中心化基建的 GEODNET。 面对即将揭晓的最终排名，现场评委与特邀嘉宾 Gavin 展开互动。Gavin 盛赞整场路演的内容水准与赛制设计，建议主办方明年必须进一步扩大这一赛事的规模。他甚至现场提议：“你们几位完全可以联合金融交易平台，直接发行一个由 All-In 推荐组合构成的公开交易一篮子指数（All-In Basket）或 ETF，让全球普通观众也能一键同步参与这些顶级专业投资人的首选标的。”这一充满雄心的商业构想立刻引发现场热烈共鸣，全场随即进入评委揭晓前最紧张的最终打分与定夺环节。 ### [57:17](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3437s) 投资理念碰撞：娱乐性路演 vs. 风险收益比 现场主持人在开场时半开玩笑地给嘉宾下战书，提议明年让他来主讲一场关于“芯片与存储器超级周期”（silicon and memory super cycle）的深度分享，对方当场欣然应允并直言“算我一个”。随后话题迅速切回对四位投资人路演方案的深度盘点。评审们指出，评估这些路演时必须把“极具观赏性与娱乐性的表达”和“扎实优异的风险收益比（risk-reward）”区分开来。尽管 Oleg 和 Kyle 的路演十分精彩，但如果评审本身并非专注于医疗健康或加密货币领域的投资人，就很难跨界下注。不过，大家对路演内容给予了高度评价：GEODNET 的去中心化物理基础设施网络（DePIN）构想令人耳目一新，而 Oleg 关于放射性药物（radiopharmaceuticals）的科普更是让全场兴奋地意识到人类未来或许能健康长寿地活到上百岁，其报告中运用的军事术语和战略类比也极其抓人眼球。 从纯粹的风险收益比角度来看，MGM Resorts（美高梅）展现出了极强的确定性。由于有 Barry Diller（IAC 董事长）的收购底线支撑，美高梅的下行风险几乎被彻底封死，同时其在日本和迪拜的新项目则构成了极为可观的未来增量催化剂。至于 Talen Energy，虽然商业逻辑十分诱人，但评审们一致认为围绕 AI 数据中心供能的投资都必须直面日益严峻的监管政策风险——这并非 Talen 本身的经营缺陷，而是悬在头顶的巨大负面外部效应：美国联邦政府介入控价、电力市场监管、实验室或关键设施收归国有，或是国会与政府换届带来的电力供需政策重构，这些宏观变数在当下都极其难以量化。 - **半导体超级周期约定**：现场嘉宾受邀并正式承诺将于下一届峰会深入讲解芯片与内存的超级周期。 - **美高梅（MGM Resorts）的核心护城河**：Barry Diller 的潜在收购报价提供了坚实的估值底线（floor），日本与迪拜项目则构成长期估值看涨期权。 - **泰伦能源（Talen Energy）的系统性风险**：独立于企业执行力之外的监管变数、政府公用事业电价干预政策及 AI 能源管制的不可预测性。 --- ### [59:15](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3555s) 生物科技平台价值与“滑雪 vs. 单板”的加密资产隐喻 在谈到 Aktis Oncology 时，评审们强调了生物医药领域“平台型公司”（biotech platform）的巨大爆发力。正如 Oleg 所言，一旦一家生物科技公司攻克了一种具备通用机制的成药方式或精准靶向递送技术，它就不再仅仅是一款药物的单点突破，而是能演变为广泛适用的底层平台。这类在生物医药领域极为罕见的底层突破，往往能催生出千亿美元级别的超级巨头，这也是 Aktis 最具吸引力的地方。 当被问及为何始终不肯下场参与加密货币投资、究竟需要怎样的催化剂才会跨界时，一位坚持传统股权投资的嘉宾用了一个生动的滑雪比喻： > “我对加密货币的看法，跟我对单板滑雪（snowboarding）的感觉一模一样。我天生不是顶尖的运动员，我花了一辈子的时间去学双板滑雪（ski），才勉强掌握得还行。在投入了成千上万个小时的双板练习之后，一想到要从零去踩单板，我就不想受那个罪。我在二级市场和股票投资领域摸爬滚打了 25 年，积累了无数的教训、痛苦和伤疤。加密货币对我而言就像单板滑雪——想玩单板的人尽情去玩，大家开心就好，只要别横着滑下山把雪道上的粉雪给刮毁了就行。” 这种观点引发了全场大笑，但也清晰勾勒出传统资深公开市场投资者对新兴代币经济与 DePIN 概念所持有的敬畏与审慎边界。 --- ### [01:01:00](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3660s) 深度估值拆解：现金流、利率敏感度与下行风险 紧接着，评委们围绕四项资产的收益空间、潜在下行幅度与兑现周期展开了极其严密的量化比较。在美高梅方面，估值不仅在于日本与迪拜等海外市场的落地，更在于拉斯维加斯存量核心物业在商业变现上的大幅升级。以威尼斯人酒店（The Venetian）为例，得益于毗邻拉斯维加斯球体馆（Sphere）的顶级娱乐演出引流，场馆在有演出的日子里每天能多赚 100 万美元的息税折旧摊销前利润（incremental EBITDA）；演出带来的客流直接带动了博彩收入的暴涨。Barry Diller 深度参与并重构这些物业的娱乐体验，有望使博彩流水实现二次腾飞。即便管理层拒绝收购并维持独立运营，依托极高的下行安全垫，美高梅有望在 2 年内实现约 3 倍回报。 相比之下，Talen Energy 虽有 3 到 5 倍的上涨空间，但其变现周期长达 8 年，且其业务模式对利率极其敏感：因为长周期购电协议（PPA）构成了其收入的核心支柱，贴现率直接受制于基准利率走势。一旦利率攀升，其远期 15 倍的估值倍数将会面临剧烈的利润率压缩。而在 Aktis 方面，固体瘤疗法面临着新兴治疗模态（如新型抗体/降解蛋白偶联物）的迭代冲击，更面临生物科技行业普遍存在的中国市场仿制与知识产权外流风险（China risk）；若临床读出数据优异，礼来（Eli Lilly）等医药巨头可能在 6 个月内发起天价收购，但若数据折戟或遭新技术替代，下行空间恐高达 50% 至 75%。对于 GEODNET，评委们则认为从技术演进上看，基于卫星和太空系统的架构极大概率将在未来十年全面取代现有的 RTK 与地面 GPS 体系，成为大型空间基础设施的必然产物。 ```mermaid flowchart TD accTitle: 投资标的风险与回报特征分类 accDescr: 按照收益预期与风险特征将四个路演标的划分为稳健防御类与高赔率彩票类两大阵营 subgraph Defense ["稳健防御与强安全垫型 (Downside Protection)"] MGM["<b>MGM Resorts (美高梅)</b><br>• 预期：2年约 3x 回报<br>• 安全垫：Barry Diller 报价兜底<br>• 催化剂：Sphere 演出引流博彩/日本迪拜扩张"] TLN["<b>Talen Energy (泰伦)</b><br>• 预期：8年 3-5x 回报<br>• 优势：AI 算力电力长期刚需<br>• 风险：高利率压制估值、政策强监管"] end subgraph Lottery ["高赔率高波动彩票型 (Lottery Tickets)"] AKTS["<b>Aktis Oncology</b><br>• 预期：潜在百亿/千亿平台或被大药企收购<br>• 风险：临床读出失败存在 50-75% 下行、竞争与仿制"] GEOD["<b>GEODNET</b><br>• 预期：颠覆性高精度时空 DePIN 网络<br>• 风险：加密机制与代币风险、太空卫星替代竞争"] end ``` - **Sphere 的吸金效应**：威尼斯人酒店在球体馆办演出的当天，能斩获 100 万美元/天的超额 EBITDA。 - **美高梅回报预测**：具有坚实安全边际，预计 2 年内可实现约 3x 投资回报。 - **泰伦能源的软肋**：周期长（看至 8 年）、PPA 现金流贴现对高利率环境极度敏感。 - **Aktis 潜在下行**：受制于临床读出与技术路线替代，坏消息下存在 50%-75% 的回撤风险。 --- ### [01:03:26](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3806s) 现场真金白银建仓、搞怪奖杯与大众投票揭榜 在最终总结阶段，主持与评委团队直接将四个标的精炼归纳为两大类别：“彩票池”（Aktis 和 GEODNET，可能产生惊人回报，但也伴随着归零风险）与“基石池”（美高梅与 Talen，分别代表了人类永恒的需求——赌博娱乐和保障供电）。有评委更是透露，自己已经在场下各豪掷 20 万美元实时买入了美高梅和 Talen 的股票。现场随即爆发了一连串精彩的互怼：没有下载 Robinhood 股票账户的评委急忙打电话给家族办公室下单，甚至要求“分一半仓位”，却被对方以“投资组合刚刚已经涨了 7%”为由当场拒绝，场面极为欢乐。最终三位评委的综合排名均高度统一：美高梅稳居第一，Talen 紧随其后，Aktis 和 GEODNET 分列三四位。 ```mermaid flowchart LR accTitle: 现场大众投票名次（已揭晓部分） accDescr: 展示四位选手路演在现场观众投票中的排名情况 P4["<b>第四名 (5% 得票率)</b><br>Kyle Samani (GEODNET)"] P3["<b>第三名 (21% 得票率)</b><br>Oleg Nodelman (Aktis Oncology)"] Top2["<b>冠亚军对决 (74% 得票率)</b><br>美高梅 (Aaron) vs. 泰伦能源 (Dan)"] P4 --> P3 --> Top2 ``` 在一阵哄笑声中，四位参赛投资人被请回主舞台。为了营造类似奥斯卡最佳男主角颁奖的紧张刺激感，主持人特意要求搬出峰会标志性的“极具阳刚气质的纯爷们奖杯”（extremely heterosexual alpha male award）——一座由 3D 打印制作的“两名成年男子极度尴尬拥抱”的金色雕像。主持人甚至拉来同伴在台上现场还原如何保持身体僵硬拥抱数秒并打趣称“拥抱满 2 分钟身体才会分泌催产素（oxytocin）”，令全场捧腹。随后，基于现场观众投出的 150 张选票，大屏幕正式自下而上公布最终名次：Kyle Samani（GEODNET）以 5% 的得票率位列第四名，而 Oleg Nodelman（Aktis Oncology）则斩获 21% 的得票率稳坐第三名。最终的冠亚军角逐，正式锁定在美高梅与 Talen Energy 之间。 ### [01:06:22](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3982s) 观众投票揭晓：Dan Dreyfus 携 Talen Energy 斩获半数选票 现场气氛在最后的结果揭晓时刻被推向了最高潮。主持人宣布首先揭晓现场观众的投票结果。在全场观众的投票中，Suvretta Capital Management 的 Aaron Cowen 凭借关于美高梅度假村（MGM Resorts）的推介获得了 24% 的选票，位列观众票选的第二名。而 Bornite Capital 的 Dan Dreyfus 则以高达 50% 的压倒性得票率荣登榜首，凭借其对能源公司 Talen Energy 的深刻阐述毫无争议地夺得观众选择奖的第一名。 伴随着全场的一片欢呼与惊叹，主持人们起哄让获胜的 Dan Dreyfus 走上前接过奖杯，并打趣现场的参赛嘉宾们看起来都有些拘谨和不适应。主持人催促着把奖杯递到 Dan 手中，调侃让他发表一段类似奥斯卡获奖感言一样的致辞，感谢帮助他走到今天这一步的所有人。Dan 捧着奖杯幽默地回应道：“奖杯我拿到了，龙舌兰酒也有了，非常感谢大家。” - **现场观众投票第二名**：Aaron Cowen（Suvretta Capital Management，标的：MGM Resorts），获得 24% 选票。 - **现场观众投票第一名**：Dan Dreyfus（Bornite Capital，标的：Talen Energy），获得 50% 选票。 ### [01:07:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=4038s) Besties 最终评选发生大逆转：Aaron Cowen 夺得第一 紧接着，主持人宣布由四位 Besties 亲自评选的最终排名结果。与观众投票的走向不同，四位主持人的评判带来了一场戏剧性的大逆转： Multicoin Capital 的 Kyle Samani（推介 GEODNET）在 Besties 评选中位列第四；EcoR1 Capital 的 Oleg Nodelman（推介 Aktis Oncology）位列第三；而在观众票选中以 50% 得票率拔得头筹的 Dan Dreyfus，在 Besties 心目中屈降至第二位；最终，Aaron Cowen 凭借对 MGM 的推介成功反超，逆转拿下 Besties 评选的冠军宝座。随着主持人们宣告“MGM 获胜”，全场在对所有参赛投资人的感谢与掌声中圆满收官，为这场首届“Best Ideas”投资推介大赛画上了句号。 ```mermaid flowchart TD accTitle: 观众投票与 Besties 投票结果对比 accDescr: 对比观众投票与 Besties 评选的排名结果，展示 Aaron Cowen 的逆转夺冠。 subgraph Audience["观众投票结果 (Audience Vote)"] A1["第一名: Dan Dreyfus (50% 选票)<br/>Talen Energy"] A2["第二名: Aaron Cowen (24% 选票)<br/>MGM Resorts"] end subgraph Besties["Besties 评选结果 (Hosts' Choice)"] B1["第一名 (大逆转): Aaron Cowen<br/>MGM Resorts"] B2["第二名: Dan Dreyfus<br/>Talen Energy"] B3["第三名: Oleg Nodelman<br/>Aktis Oncology"] B4["第四名: Kyle Samani<br/>GEODNET"] end Audience -.->|"评选翻转"| Besties ``` - **Besties 评定第四名**：Kyle Samani（GEODNET） - **Besties 评定第三名**：Oleg Nodelman（Aktis Oncology） - **Besties 评定第二名**：Dan Dreyfus（Talen Energy） - **Besties 评定第一名**：Aaron Cowen（MGM Resorts，成功逆转观众投票夺得冠军） ## Notable quotes > [00:00](https://www.youtube.com/watch?v=fO5sC7qS04E&t=0s) 要是我当年选的是英伟达（Nvidia），我现在就是索恩大会的传奇人物，早就直接退休了。 — Chamath Palihapitiya > [01:25](https://www.youtube.com/watch?v=fO5sC7qS04E&t=85s) 世界上任何人在任何地方、任何时间（24/7），仅凭一部手机和网络连接就能交易任何资产，我们正在从底层重建一套全新的金融系统。 — 参赛选手 > [02:29](https://www.youtube.com/watch?v=fO5sC7qS04E&t=149s) 连一秒钟的考虑都不要有。 — Aaron Cowen > [10:33](https://www.youtube.com/watch?v=fO5sC7qS04E&t=633s) 如果博彩牌照落地，你的投资能翻三倍；即使牌照没有拿到，光凭现有业务也足以让你赚上一倍多。 — Aaron Cowen > [13:07](https://www.youtube.com/watch?v=fO5sC7qS04E&t=787s) 我们根本不需要 AI 的需求爆发，未来二十年的电力市场本身就会极其紧缺；AI 的出现只是给这一趋势加装了涡轮增压（Turbocharges），直接制造了极端供需缺口。 — Dan Dreyfus > [16:39](https://www.youtube.com/watch?v=fO5sC7qS04E&t=999s) 让美国再次伟大（MAGA）的唯一路径，就是在电力规划上向中国看齐。 — Dan Dreyfus > [19:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1158s) 原本计划退役的燃煤发电机组绝不可能如期关停，因为如果不保留它们，世界上没有任何可能在 10 年内填补上 100 吉瓦的电力真空。 — Dan Dreyfus > [27:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1638s) 其糟糕程度大概介于投资电影、酒庄和 SPAC 壳公司之间。 — Oleg Nodelman > [27:18](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1638s) 将其他小孩的科学课作业商业化变现（monetize other kids' science projects）。 — Oleg Nodelman > [33:04](https://www.youtube.com/watch?v=fO5sC7qS04E&t=1984s) Aktis 并不适合所有人。购买 Aktis 前请咨询您的生物科技分析师。建立头寸可能会导致焦虑加剧、夜间失眠；生物科技股有时会出现断崖式暴跌；投资后可能会因缺乏竞争对手而遭遇剧烈波动。如果股价无故下跌，请立即致电您的经纪人加仓……微型蛋白向细胞递送技术尚未完全证实，Aktis 目前无任何商业化产品因此无经常性收入。若发生二次增发，请立即与 Aktis 管理层通话认购。 — Oleg Nodelman > [48:28](https://www.youtube.com/watch?v=fO5sC7qS04E&t=2908s) 你应该买代币，因为正如你能想到的那样，我手里持有大量代币，而完全没有持有该公司的股权。 — Kyle Samani > [54:50](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3290s) 投资必须有真金白银的切身利益（Skin in the game），否则你根本不会真正关心。 — Chamath Palihapitiya > [59:15](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3555s) 我对加密货币的看法，跟我对单板滑雪（snowboarding）的感觉一模一样。我天生不是顶尖的运动员，我花了一辈子的时间去学双板滑雪（ski），才勉强掌握得还行。在投入了成千上万个小时的双板练习之后，一想到要从零去踩单板，我就不想受那个罪。我在二级市场和股票投资领域摸爬滚打了 25 年，积累了无数的教训、痛苦和伤疤。加密货币对我而言就像单板滑雪——想玩单板的人尽情去玩，大家开心就好，只要别横着滑下山把雪道上的粉雪给刮毁了就行。 — 现场嘉宾 > [01:06:22](https://www.youtube.com/watch?v=fO5sC7qS04E&t=3982s) 奖杯我拿到了，龙舌兰酒也有了，非常感谢大家。 — Dan Dreyfus ## Mentioned Chamath Palihapitiya, David Einhorn, Aaron Cowen, Dan Dreyfus, Oleg Nodelman, Kyle Samani, Jason Calacanis, David Sacks, Brad Gerstner, Barry Diller, Sam Zell, 黄仁勋 (Jensen Huang), Mike Horton, Jason Friedberg, Stanley Druckenmiller, Suvretta Capital Management, Bornite Capital, EcoR1 Capital, Multicoin Capital, MGM Resorts International, Talen Energy, Aktis Oncology, GEODNET, IAC, Caesars Entertainment, Wynn Resorts, Amazon, Tesla, Box, Nvidia, Microsoft, Constellation Energy, Eli Lilly, Bristol Myers Squibb, Novartis, Bayer, John Deere, TomTom, DJI, Burro, Yarbo, Sunseeker, Trimble, Hexagon, Topcon, Leica, Helium, PJM Interconnection, 三里岛核电站 (Three Mile Island), Solana, RTK (Real-Time Kinematics), DePIN, 锕-225 (Actinium-225), Nectin-4, B7-H3, Sphere, The Venetian, Sohn Investment Conference ## Video description (0:00) Chamath explains the Best Ideas format (2:31) Suvretta Capital Management's Aaron Cowen pitches MGM Resorts (13:07) Bornite Capital's Dan Dreyfus pitches Talen Energy (27:19) EcoR1 Capital's Oleg Nodelman pitches Aktis Oncology (40:20) Multicoin Capital's Kyle Samani pitches GEODNET (54:50) The Besties recap the pitches and announce winners Follow Aaron: https://www.linkedin.com/in/aaron-cowen-0a44a450 Follow Dan: https://x.com/dreyfd https://www.linkedin.com/in/daniel-dreyfus-b65554209 Follow Oleg: https://www.linkedin.com/in/oleg-nodelman-375131 Follow Kyle: https://x.com/KyleSamani https://www.linkedin.com/in/kylesamani Follow the besties: https://x.com/chamath https://x.com/Jason https://x.com/DavidSacks https://x.com/friedberg Follow on X: https://x.com/theallinpod Follow on Instagram: https://www.instagram.com/theallinpod Follow on TikTok: https://www.tiktok.com/@theallinpod Follow on LinkedIn: https://www.linkedin.com/company/allinpod Intro Music Credit: https://rb.gy/tppkzl https://x.com/yung_spielburg #allin #tech #news [Source](https://www.youtube.com/watch?v=fO5sC7qS04E)

[0:00] Chamath Palihapitiya: 要是我当年选的是英伟达（Nvidia），我现在就是索恩大会的传奇人物，早就直接退休了。 [1:25] 参赛选手: 世界上任何人在任何地方、任何时间（24/7），仅凭一部手机和网络连接就能交易任何资产，我们正在从底层重建一套全新的金融系统。 [2:29] Aaron Cowen: 连一秒钟的考虑都不要有。 [10:33] Aaron Cowen: 如果博彩牌照落地，你的投资能翻三倍；即使牌照没有拿到，光凭现有业务也足以让你赚上一倍多。 [13:07] Dan Dreyfus: 我们根本不需要 AI 的需求爆发，未来二十年的电力市场本身就会极其紧缺；AI 的出现只是给这一趋势加装了涡轮增压（Turbocharges），直接制造了极端供需缺口。 [16:39] Dan Dreyfus: 让美国再次伟大（MAGA）的唯一路径，就是在电力规划上向中国看齐。 [19:18] Dan Dreyfus: 原本计划退役的燃煤发电机组绝不可能如期关停，因为如果不保留它们，世界上没有任何可能在 10 年内填补上 100 吉瓦的电力真空。 [27:18] Oleg Nodelman: 其糟糕程度大概介于投资电影、酒庄和 SPAC 壳公司之间。 [27:18] Oleg Nodelman: 将其他小孩的科学课作业商业化变现（monetize other kids' science projects）。 [33:04] Oleg Nodelman: Aktis 并不适合所有人。购买 Aktis 前请咨询您的生物科技分析师。建立头寸可能会导致焦虑加剧、夜间失眠；生物科技股有时会出现断崖式暴跌；投资后可能会因缺乏竞争对手而遭遇剧烈波动。如果股价无故下跌，请立即致电您的经纪人加仓……微型蛋白向细胞递送技术尚未完全证实，Aktis 目前无任何商业化产品因此无经常性收入。若发生二次增发，请立即与 Aktis 管理层通话认购。 [48:28] Kyle Samani: 你应该买代币，因为正如你能想到的那样，我手里持有大量代币，而完全没有持有该公司的股权。 [54:50] Chamath Palihapitiya: 投资必须有真金白银的切身利益（Skin in the game），否则你根本不会真正关心。 [59:15] 现场嘉宾: 我对加密货币的看法，跟我对单板滑雪（snowboarding）的感觉一模一样。我天生不是顶尖的运动员，我花了一辈子的时间去学双板滑雪（ski），才勉强掌握得还行。在投入了成千上万个小时的双板练习之后，一想到要从零去踩单板，我就不想受那个罪。我在二级市场和股票投资领域摸爬滚打了 25 年，积累了无数的教训、痛苦和伤疤。加密货币对我而言就像单板滑雪——想玩单板的人尽情去玩，大家开心就好，只要别横着滑下山把雪道上的粉雪给刮毁了就行。 [66:22] Dan Dreyfus: 奖杯我拿到了，龙舌兰酒也有了，非常感谢大家。

[1:25] anyone should be able to trade any asset anywhere in the world, anytime, 24/7 [1:25] unfollowed, unloved, misunderstood biotech companies [14:49] Daniel: We do not need AI demand to keep the power markets incredibly tight for the next 20 years. AI demand just turbocharges. That's all it does. And it creates shortages. [19:18] Three Mile Island brought to you by Microsoft Azure [27:18] Oleg Nodelman: monetize other kids' science projects [31:20] holy grail in biotech [40:19] Kyle Samani: revenue share token [40:19] Kyle Samani: you should own the token because I own a lot of the token... I don't own any of the equity. [40:19] Kyle Samani: there's no chance they can compete on cost because just sending things to space with satellites... these Geonet base stations are a few hundred bucks. [40:19] Chamath: lottery tickets [40:19] Gavin: The big negative externality for Talon is nothing to do with Talon. It's like something over the top from the US government... You have a change in administration. You have a change in Congress. There's laws that are passed. [40:19] Gavin: I have 25 years of lessons, learnings, pain, scars from investing in equities and public securities and just crypto... it's a little bit like snowboarding for me... please don't go sideways down the mountain and ruin the powder. [63:19] people will always gamble and leave the lights on. [64:32] extremely alpha male heterosexual trophy

[0:00] Maybe you could tell us a little bit [0:01] about how you selected our presenters [0:03] and your vision for this. I mean for any [0:05] of you guys who've been involved in Ira [0:09] this is a gentleman that passed away [0:10] from cancer far too young and his family [0:13] created this thing called the sone [0:14] foundation and they would host this [0:16] event and it started in Lincoln Center [0:18] and they would ask these managers and so [0:21] at the time I was like a young venture [0:23] investor and I got this invite and I [0:25] showed up in New York at Lincoln Center [0:27] in 2015 and I said Amazon's going to be [0:30] a trillion dollar company and I was [0:32] laughed out of the room. Uh David [0:34] Einhorn, who's a friend of mine, but who [0:36] was totally wrong, said, "I know [0:39] trillion dollar companies. This is not a [0:41] trillion dollar company." Wrong. [0:43] It turned out to be a great bet. I went [0:45] back. I did Tesla in 2016. We picked the [0:48] converts. And then in 2017, I was like, [0:51] "All right, this is my this is this is [0:53] it. This is my magnumopus." And I said, [0:57] "AI is the future." And then I picked [0:59] box. [1:01] I was like, if I had just picked Nvidia, [1:04] I would have been an Irish legend legend [1:06] and I could have retired. Anyway, so we [1:08] wanted to recreate Iris and start to get [1:10] these great managers who are making [1:12] great picks, making a ton of money for [1:13] their LPs. They don't get the [1:15] distribution and so it's just a chance [1:17] to like get to know some of these names. [1:19] You don't have to see them on CNBC. [1:21] You'll see them here more and more [1:22] often. And we can just get to Opine. [1:25] >> Roll the video. [1:25] >> Yeah. Ladies and gentlemen, welcome to [1:29] the best ideas pitch. Let's meet our [1:33] contestants. [1:35] >> Anyone should be able to trade any asset [1:38] anywhere in the world, anytime, 24/7, [1:40] with just an internet connection and a [1:42] phone in their pocket. [1:42] >> We're building a new financial system [1:44] from the ground up here. [1:45] >> People are going to want to own [1:46] equities, and it's going to be fun. In [1:49] the next couple years, companies going [1:50] to innovate and create products and [1:51] applications, and that's where hopefully [1:54] long short managers like us can make a [1:56] boatload of money. [1:57] >> My fund, Eco R1 Capital, which is based [1:59] in San Francisco, thinks of investing in [2:00] biotech in a slightly different way. [2:02] We're looking for unfollowed, unloved, [2:04] misunderstood biotech companies. [2:05] >> It's an amazing moment in time for those [2:07] types of companies. There's been a [2:10] structural and permanent perception [2:12] shift where both sides of the aisle are [2:16] going to be leaning into nuclear in a [2:18] big way. [2:18] >> I'm massively optimistic. You know, all [2:20] of this leads me to just the maximum [2:24] risk on. [2:29] >> Thanks the besties for having me. And [2:31] this is obviously a fabulous event you [2:32] guys have put on. I'm happy to be here. [2:34] Um, for those of you who don't know me, [2:36] um, I run a $4 billion firm in New York [2:38] called Serreta Capital. Um, before [2:41] founding my firm, I was, uh, I ran the [2:43] equities business for George Soros. Uh, [2:45] I was then CIO for Steve Cohen. Uh, and [2:48] I've been doing hedge funds now for 29 [2:50] years. So, um, definitely on the older [2:52] edge of my my pure group. Um, so I was [2:56] thinking about, you know, I run a [2:57] generalist fund. Um, and you know, we [3:00] own a bunch of tech stocks, but you [3:02] know, given this audience here, for me [3:03] to pitch a tech stock would be [3:05] absolutely completely stupid. Um, so I [3:07] was thinking about what else. And you [3:09] know, obviously the theme of this [3:11] conference besides tech is poker. So I'm [3:15] going to pitch to you MGM. Now most of [3:18] you know MGM as uh you think about it as [3:21] the Vegas um company. They own 13 [3:24] properties in Vegas. Um they're one [3:26] they're them and Caesars are the two [3:28] largest owners of uh casino assets in [3:31] Vegas. Um now if you notice um the other [3:35] day um Caesars got taken out. Um and so [3:40] we think Vegas is actually starting to [3:42] improve. Um but I'm not here to pitch [3:45] MGM because of Vegas. What I'm [3:49] going to tell you is there's a couple of [3:50] things we noticed. One is this company's [3:54] been very aggressively been stock's been [3:56] aggressively acquired by Barry Dillard [3:58] lately. Um Barry now owns 26% of the [4:01] company. Um now I put this presentation [4:04] together two weeks ago. Yesterday he [4:07] actually bid for the company. Okay. So [4:09] when I put the presentation together the [4:11] stock was about $37. It's now high40s. [4:14] He bid $48. Okay. I would not sell my [4:18] shares to him. [4:19] >> When did we get this presentation? Did [4:21] we get it early enough to trans? [4:23] >> Uh I would not sell his I would not sell [4:25] my stock to him for a second. And the [4:28] reason is also besides him buying the [4:31] stock, the company's also been buying [4:33] the stock. Rarely have I ever seen a [4:36] company in 6 years buy half their float [4:39] back. So you have Barry Diller who's the [4:42] legend aggressively buying the stock and [4:44] it's also now 80% of his NAV. Okay. So [4:48] you most people think of Barry Diller as [4:50] the ABC producer. He did I which owned [4:53] assets like Expedia and now he's a [4:56] casino guy. What is going on here? So we [5:00] spent a lot of time [5:02] um asking ourselves why um and why is [5:08] MGM has two hidden assets. Okay. The [5:12] first one [5:14] is, and this is sort of our the [5:17] punchline of what we think the stock is [5:19] worth. So you add the Vegas assets plus [5:22] China, you get about little low60s. So [5:25] from $48 or 37 when I started this, [5:29] great return. What is they what they [5:32] have now is a license to open a casino [5:36] in Aaka, Japan. Japan a couple years ago [5:39] went through a whole referendum around [5:41] the country. They have prefixures. The [5:43] prefixures voted. The only one that [5:46] decided to own open a casino is Osaka. [5:50] Now Osaka is and this is what the [5:53] asset's going to look like. It's going [5:54] to open in 2030. If you go to the [5:57] company slide presentations, they sort [5:59] of mention this, but they're not really [6:01] talking about it. Um, Japan, just for [6:05] you people, um, sorry, I don't know, [6:08] this is very slow. Um, Japan actually [6:10] has a reasonably large gambling market. [6:12] They have pachinko parlors and they have [6:14] horses. That's about a $40 billion [6:16] market. If you look at the market in [6:18] Macau, that's $30 billion. And if you [6:20] look at Vegas, it's only $10 billion. [6:22] So, this could be a massive opportunity. [6:26] Um, you know, we're estimating they'll [6:28] do about $2 billion of IBIDA. They own [6:31] 40% of the property. Um, they also get a [6:34] management fee for this. Um, if you also [6:36] look at where Osaka is located, it's a [6:39] great, you know, so the Japanese like to [6:41] gamble, but the Chinese really gamble. [6:43] Okay. So, if you look at where it is [6:46] from Shanghai, it's shorter than Macau [6:48] and Singapore, which are the two big [6:50] gaming options in Asia. And from and [6:53] from Beijing, about the same distance as [6:55] Macau, and obviously much shorter than [6:57] going to Singapore. So, if you want to [6:59] go gambling for a weekend and you live [7:01] in Shanghai, live in Beijing, Osaka is [7:04] great. It's also a first world nation. [7:06] Um, and if you think about as an [7:08] investor, where would you want to have [7:09] your money? Look, Macau has issues. It's [7:13] a low multiple business. This is Japan. [7:15] It's a first world country. So, we think [7:17] Barry Diller is understands gambling. He [7:21] understands casinos. But what he's [7:23] really doing is now trying to pick off [7:25] the company to get the Japanese [7:28] opportunity which we think is worth will [7:30] more than double the stock. The final [7:32] option and I'm keeping this simple. What [7:34] I love about this pitch is it's really [7:36] simple. It's not that hard to do the [7:37] math. MGM is built somebody they're [7:41] they're they're branding they're [7:42] building a property in Dubai. Okay. Now [7:45] it's going to it's a grand complex. It [7:48] has an Arya. It has an MGM and it has a [7:50] Bellagio. Gambling is illegal in Dubai [7:53] right now. Okay. But they have snuck in [7:58] this building 300,000 square ft of cu of [8:02] space. Well, one day if Dubai decides to [8:06] legalize gambling, guess where it's [8:09] going? Right there. [8:11] Next year, two, sorry, two years from [8:13] now, win is going to open a casino in a [8:16] place called Ararjan, which is 45 [8:19] minutes away from Dubai. Now, any of us [8:22] who want to go gambling in Dubai, we [8:25] arjan's a bit of a pain in the ass to [8:26] get to, we're going to want to go here. [8:29] So, we think there's a chance that [8:31] especially when win opens. Also look, [8:33] there's a possibility of the war, you [8:36] know, Dubai wants to reestablish [8:37] themselves that they open a casino in um [8:42] Dubai and you know what that would be [8:44] worth. So when you take the Vegas [8:47] assets, which we think are worth about [8:48] 60, when you take Japan, which we think [8:51] is worth about 50 bucks, if Dubai [8:53] happens, that's worth another $40 or [8:55] $50. So we think the stock is a triple. [8:59] Bar remember Barry's bidding for the [9:01] company. Okay, he is not a strategic [9:04] buyer. He is a financial buyer and he's [9:06] doing it to get rich. So therefore, I [9:09] think this company is now in play. I [9:12] don't know how it's all going to play [9:13] out, but if you own shares, don't tend [9:15] to them. And the riskreward is [9:17] incredible right now because, you know, [9:19] I'm telling you, I think the stock could [9:20] be easily worth over 100, could be worth [9:23] 150. And now you have Barry Diller who [9:26] is a has a firm bid owns 26% of the [9:29] company basically at the same price. So [9:32] I think this is a cool idea. [9:33] >> Well done. [9:36] Okay. Uh anybody let's uh do two [9:39] questions and uh yeah we'll put we'll [9:41] give you both questions at the same time [9:42] for efficiency. [9:43] >> How much have you looked at like the [9:45] monetization of the assets outside of [9:47] gambling? I had heard from someone that [9:49] Barry Diller was spending a lot of time [9:50] trying to reinvent the entertainment [9:52] piece of the properties. he was active [9:54] on the board and they were trying to [9:56] identify that the entertainment property [9:58] is way the entertainment value is way [10:00] undermonetized and they could be making [10:02] a lot more per [10:03] >> Okay, don't answer yet. That's question [10:05] one. [10:05] >> And then question two is how when you [10:07] expand internationally do you scale [10:10] customer credit because that tends to be [10:13] the thing that drives [10:14] >> you know people to come back and [10:16] >> well obviously MGM let me start with [10:17] your question first. MGM has a massive [10:20] database of customers, right? So, you [10:22] know, I assume the Vegas properties have [10:24] guys that come from China, they come [10:26] from um Japan. They'll use that database [10:29] to do it. They also have a loyalty [10:30] program. Um I unfortunately made a bad [10:33] investment in a company called Rio, the [10:34] Rio, which was in Vegas. Um which we [10:37] bought when they separate when Caesar's [10:40] merged with El Dorado. Um they had to [10:42] shed an asset. Um that was that was uh [10:45] the Rio. We I did an investment with a [10:48] couple of friends and we were buying the [10:50] thing at $200 per square foot. The thing [10:52] we forgot was when you separated from [10:55] Caesars, you lost the loyalty program [10:57] and that ended up two quick questions [10:59] from the audience. [11:00] >> Wait, I got to ask Kyle's question. [11:02] >> Hold on. Question and then from the [11:03] audience. [11:04] >> Let me get his first. [11:05] >> The entertainment question. [11:06] >> I don't know the answer. I don't know [11:08] the answer to that. If it's if he can [11:10] make them better, it will help. [11:13] >> But as I'm saying, this is not really a [11:15] Vegas. This is an Asian casino play that [11:20] >> the and if if you look at their [11:22] presentations which is really cool. [11:25] >> They are not they barely mention it. So [11:27] what one of the things we happened [11:28] besides we were hoping one of the cat so [11:30] looked I worked at SACE and one of the [11:32] things we focus on is catalyst path. [11:34] >> So what was the catalyst path? The [11:36] catalyst path was they would have an [11:38] investor day blah blah blah. Barry just [11:40] showed his cards. So, um, but you know, [11:44] >> Aaron, two questions. Caesars left Dubai [11:47] waiting for a license. Why would this be [11:50] different for MGM? That's question one. [11:52] >> And then question two is the Osaka [11:54] Casino was approved in 2023. [11:57] Why was the market ignoring this hidden [11:59] asset until the bid? [12:00] >> Sure. Let me answer this. So, what's [12:03] also cool about this idea was, so I've [12:05] been doing this for 29 years. um win um [12:10] they opened Macau. So Win started as a [12:12] Vegas property then opened Macau. The [12:14] market started caring about it about 3 [12:17] years before it opened. Um so that the [12:20] answer is they should care about it. The [12:22] reality is it's tends to be about 3 [12:24] years before it opens. Well, we're [12:26] almost in that time frame, which is why [12:28] we think it's opportunistically the [12:30] right period of time. Um I, you know, [12:32] regarding the question with Caesars, [12:34] look, this is an option. As I told you, [12:36] somebody built this project for them. [12:39] They are running it for them and they [12:42] were intelligent enough to leave 300,000 [12:45] square ft of empty space in case they [12:47] get a casino. Well, if that happens, [12:49] great. If it doesn't, you know, you're [12:52] still going to double more than double [12:53] your money. So, you know, you if if it [12:55] happens, you triple your money. [12:56] >> Free option. You're saying [12:58] >> it's free option. [12:59] >> Well done. Big round of applause. [13:01] >> Thanks, guys. Appreciate it. [13:02] >> Nicely done, Aaron. [13:04] Next up, Daniel. [13:06] >> Long time no see. [13:09] So, today we're doing talent energy. Uh, [13:12] but first, [13:14] the anatomy of a power cycle. [13:19] So, a power cycle typically goes like [13:22] this. In normal times, power demand [13:24] grows about GDP. So, if GDP grows 2%, [13:27] power demand grows two. If GDP grows [13:29] three, power demand goes three. And [13:32] there's moments in time where we get [13:33] technological breakthroughs and a lot of [13:36] those technological breakthroughs are [13:37] very power intensive. So power demand [13:40] spikes [13:42] and once everybody adopts that [13:44] technology [13:46] it trends back down to its onal [13:48] algorithm [13:50] GDP growth and then you go through the [13:54] efficiencies phase where we say let's [13:56] try to conserve and figure out ways to [13:58] consume less power and then the cycle [14:01] starts all over again. So, you know, in [14:03] history, the big technological boom that [14:05] sent power demand skyrocketing was [14:08] appliances and air conditioning. [14:10] Everybody had to get their kettles and [14:12] and the aircon. Then in the 70s and 80s [14:16] and 90s, demand normalized again. But [14:18] then the 2000s were all about [14:20] efficiencies. You know, we had like LED [14:22] lighting, smart HVAC, tinted windows, [14:24] smart electronics. And at the same time, [14:27] as I said earlier, we were like, you [14:29] know, ripping down all our power- hungry [14:31] infrastructure like aluminum smelters [14:33] and moving over China. So, we had two [14:35] decades of effectively no power demand. [14:38] And now we're just coming out of it and [14:40] starting a technological cycle again [14:43] where power demand is going to really [14:45] start to explode from these sort of high [14:47] 2% numbers you're seeing on the screen. [14:49] Now, I want to say something right now [14:52] that is incredibly important. [14:57] We do not need AI demand to keep the [15:02] power markets incredibly tight for the [15:07] next 20 years. [15:10] AI demand just turbocharges. [15:13] That's all it does. And it creates [15:16] shortages. So just remember that. Early [15:19] in my career, I was on a panel with Sam [15:23] Zel. Interestingly, it was a panel on [15:26] opportunities in Mongolia. I was looking [15:27] at a copper mine and he was looking at [15:29] real estate. There was one thing he said [15:31] that stuck with me for the rest of my [15:33] career is he said, "If you can buy an [15:37] asset, a hard asset at below replacement [15:40] cost for an asset that's going to be [15:42] needed in the future where we're going [15:43] to need to build new capacity of that [15:45] asset. Then you buy that asset at the [15:48] discount to replacement cost. You hold [15:50] it and you sell it at a big premium to [15:53] replacement cost when the market wakes [15:55] up. That's exactly we did with Equity [15:56] Office properties. Sold it at the peak [15:58] of the market but bought it at a [16:00] discount to replacement value. Talon [16:02] Energy is a power producer. They have 2 [16:05] gawatts of nuclear power and they've got [16:07] 6 gawatt of natural gas base load power. [16:10] Today in the stock market, as a good [16:14] speculation, you could purchase this [16:16] company at a $25 billion enterprise [16:18] value. the replacement cost is 45 [16:20] billion. And because they've got debt, [16:23] it means that the equity value uh just [16:25] to get to replacement cost uh is more [16:28] than a double uh from where it's trading [16:30] today. And if you follow Sam's playbook, [16:34] then we ultimately end this cycle at a [16:36] big premium to replacement value. [16:39] So when I see this, [16:42] I say the plan for America [16:45] on the power side has to be this. Make [16:49] America great again. Copy China. [16:54] If you look what China did over the last [16:56] 20 years, we started out this cycle with [17:00] having 2x the power generation that [17:02] China had. Fast forward to today, China [17:04] has three times the power generation [17:06] capacity that we have. Now, if you [17:10] believe that artificial intelligence is [17:12] going to be responsible for scientific [17:14] breakthroughs, you either have it or you [17:16] don't have the scientific breakthroughs. [17:17] If you believe that artificial [17:18] intelligence is going to drive robotics, [17:20] you either have it or you don't have [17:21] that productivity from the robots. If [17:23] you believe that artificial intelligence [17:24] is going to be helpful for national [17:27] security and military affairs, then you [17:29] either have it or you're dead. And so [17:32] this is an absolutely mandatory [17:36] buildout that we have to do otherwise [17:39] we're going to fall behind because at [17:40] the end of the day what is a data [17:42] center? You know in my world in the [17:44] commodities world I look at a data [17:45] center as the exact same thing as a [17:47] refinery. In a traditional hydrocarbon [17:48] refinery you put oil in crude oil in you [17:52] refine it into jet fuel or gasoline for [17:54] your car. With a data center, you put [17:58] electricity in and on the other end [18:01] instead of gasoline or jet fuel out [18:02] comes photons or tokens or intelligence, [18:04] whatever you want to call it, but it's [18:05] the same thing. Big capital inensive [18:08] asset, $50 billion per gawatt and power [18:12] just like electricity, just like oil is [18:14] the input to that refinery. So, here's [18:18] Jensen [18:19] and uh he was just recently quoted that [18:23] we need a thousand times more power than [18:27] we currently have. Now, if that's [18:30] remotely true, we need every single [18:33] source of power that you can imagine. We [18:35] need hundreds of gigawatts of nuclear, [18:37] we need solar, we need orbital, we need [18:40] it all, if this is even remotely true. [18:44] But the challenge as we spoke about [18:46] before is the supply chain, right? All [18:48] of these, you know, a data center [18:49] competes for the same supply chain of [18:50] the critical minerals that space [18:52] launches and orbital data centers do. [18:54] Power plants need all the same nickel [18:57] super alloys that it takes to launch [18:59] rockets and the silver that goes into [19:01] these photovalttaic cells. And so and so [19:04] there's going to be shortages of [19:06] everything and delays everywhere. And my [19:09] point here is we are just going to need [19:13] every solution that we can throw at this [19:15] for the foreseeable future. [19:18] So here's a little region in the US [19:20] called the PJM, Pennsylvania, Jersey, [19:22] Maryland. Uh this is a forecast from the [19:24] grid operator where they say that over [19:27] the next 10 years we're going to need [19:29] 106 gawatts of new power in the PJM in [19:32] just one little area of the US. Now in [19:35] 10 years in geological time that's like [19:37] tomorrow morning right we're all so used [19:39] to internet time you press a button and [19:40] you get your food delivered to you or [19:42] your car picks you up in two seconds you [19:44] know building infrastructure happens in [19:45] geological time 10 years to build out [19:47] 106 gawatt is literally a nancond from [19:51] now and you know you see that thermal [19:54] coal retirements we ain't retiring those [19:56] coal plants because there's no world [19:58] where we're going to be building 100 [19:59] gawatt in 10 years that's the size of [20:02] what Japan consumes today for one little [20:04] part of the US. And you know what I'll [20:07] say is those that understand the supply [20:10] chain and what goes into building all [20:12] this, everybody's in panic mode because [20:16] we know that we don't have the raw [20:19] materials to meet this level of demand [20:22] that's coming our way. So that's going [20:24] to keep existing capacity and power [20:26] prices very tight. Now the data center [20:29] and the hyperscalers are in a panic. [20:32] They're trying everything they can to [20:34] source as much power as they can under [20:36] long-term PPAs power purchase price [20:39] agreements at fixed prices for 20 years. [20:42] There's a famous example, you know, I [20:43] thought Microsoft was a green company, [20:46] but they went and convinced [20:48] Constellation Energy, which is a company [20:50] that owns the 3M island nuclear reactor, [20:52] you know, the one that melted down and [20:54] created, you know, the nuclear meltdown [20:55] that gave nuclear a bad name for 30 [20:57] years. It was Microsoft that told them [21:00] they needed to start it up. And in order [21:02] to incentivize to stimulate their hand [21:04] to wallet reflex to start this thing up, [21:06] they said, "Power prices stay $50 a [21:09] megawatt hour. We'll pay you a hundred a [21:12] year for 20 years minimum price for you [21:15] guys to start this up." And so here we [21:19] have it. Uh 3M Island brought to you by [21:21] Microsoft Azure. So, you know, it's [21:25] getting harder to do these deals because [21:26] the regulators are saying, "Wait a [21:28] minute. If you're taking all this power [21:30] off the grid for your data center, how [21:32] are we going to heat the homes of our [21:34] customers?" And so, uh, you know, it [21:37] we're getting ourselves into the moment [21:40] of what I call crunch time. So, just to [21:43] finish up, uh, here are the numbers on [21:45] Talon. Um, the stock today is sort of in [21:48] the high 300s. If they just do [21:50] absolutely nothing, just absolutely [21:53] nothing, just sit there and run the [21:55] business, let their Amazon data center [21:58] contract roll up, these guys will be [22:01] generating $50 a share of free cash flow [22:04] per year. Again, the stock is in the [22:06] high 300. So, it's about seven times [22:07] free cash flow. Good infrastructure [22:09] assets in the US, traded about 15 times. [22:12] So, that's pretty good. You get a double [22:13] for basically management just sitting [22:15] around and doing nothing. But if they [22:18] continue to figure out ways to sign [22:20] contracts with data centers at premium [22:22] prices or if power prices go up. I mean [22:25] the amazing thing right now is in the [22:26] PJM where these guys operate, the power [22:29] price is still too low to stimulate new [22:31] capacity. The math still doesn't work, [22:33] which is really mindboggling. So if [22:35] power prices go up a bit, they do more [22:37] deals, you get to $70 a share of [22:40] recurring annual free cash flow, put a [22:42] 15 multiple on that, that's,50. [22:45] But then if they get into building power [22:47] plants, right? And right now the [22:49] regulator is telling these companies to [22:50] go sit in a room, power producer, data [22:53] center, come in a room, make a deal so [22:56] that you build power and get a good [22:58] return on it. And the data center gets [22:59] their power gets a good return on it. [23:01] And Talon is in a pole position to be [23:03] able to do this. If they just build like [23:05] 4 gawatt of the 100 gawatts that we [23:07] need, you could get up to, you know, [23:09] over $100 a share of free cash flow. The [23:11] stocks in the high30s today. So, uh, go [23:14] and buy the shares. It's a good [23:15] speculation and, um, we can chat. [23:18] >> All right. Not financial advice. Gavin, [23:20] go. [23:21] >> Gav and go. [23:22] >> I just, just very curious like how do [23:23] you think about regulatory risks here? [23:25] Nobody likes their electricity prices [23:27] going up. AI is an increasingly [23:30] political issue just like how do you [23:33] think about that risk? [23:35] >> We need AI and we need to figure this [23:36] out. And so there's different ways to [23:38] skin a cat here, right? My personal view [23:41] is during peak hours, right? If you go [23:44] drive down a highway at 4 in the [23:46] morning, you know, you would sit there [23:48] and say, "Why do we have all this [23:49] highway capacity? This is crazy." But [23:51] then you go on that same highway at rush [23:52] hour, you're like, "Oh, we don't have [23:54] enough highway capacity. There's not [23:55] enough lanes." Power is the same thing. [23:57] There's only a few hours a day where you [24:00] really stress the system. And so I think [24:02] the working solution to get around this [24:04] regulatory issue is you do the PPAs with [24:08] the data centers. You force the data [24:09] centers to throw a ton of battery behind [24:11] it and some peakers just to get through [24:13] that really intense period. And then uh [24:17] that's a good band-aid solution until we [24:19] build more power. So there's ways to do [24:21] this. Human ingenuity is going to win [24:22] here. We're going to get our data [24:24] centers and consumer power bills are [24:25] going to be, I think, relatively under [24:27] control. They're going to go up, but [24:28] they're going to be under control. Okay, [24:29] Dan, I have three questions from the [24:31] audience. Really good ones. Number one, [24:33] does your thesis actually need behind [24:35] the meter collocation to clear or is it [24:38] just a bet that clean firm base load is [24:41] scarce enough that it doesn't matter [24:43] whether power flows in front of or [24:45] behind the meter? It's the latter. And [24:46] that's why I gave three scenarios, [24:48] right? The $50 a share of earnings per [24:51] share. Again, a high $300 stock, right? [24:54] $50 a share of earnings. Nothing has to [24:56] happen. You just sit, right? And you [24:58] double your money. Now, if you get more [25:00] behind the meter or even front of the [25:01] meter, that's how you get up to that $70 [25:04] a share of earnings from 50. And then if [25:06] you get up to the 70, but start building [25:08] new capacity, then you get to the $100 [25:10] plus. [25:11] >> Okay. Question two from Brad. How do you [25:13] think about competition for power from [25:16] things like fuel cells, gas turbines, [25:18] aerodyit turbines, orbital compute, and [25:21] other sort of IPs, independent power [25:23] producers. [25:24] >> We need all of it. We need all of it. We [25:26] we you know fuel cells and you know the [25:29] Caterpillar solar turbines these are [25:32] fantastic bridge solutions but the cost [25:34] to run these things the LCOE is like [25:36] through the roof but you you know look [25:37] to build a $50 billion data center you [25:39] don't want it to sit idle for 3 years [25:41] waiting for your base load CCGT so you [25:43] do whatever it takes you don't give a [25:44] crap what you pay for that bridge [25:46] solution and so we're finding ways [25:48] through fuel cells through uh you know [25:50] Caterpillar solar turbines hopefully [25:52] through orbital data centers where we [25:54] can alleviate this because I want AI to [25:57] happen in a really big way and we're [25:58] going to need all the above. [26:00] >> Okay, question three. By the way, great [26:02] questions guys. Thank you for these. [26:04] What is the right terminal multiple for [26:06] Talon if the business mix shifts from [26:08] merchant IP to contracted [26:11] infrastructure? [26:12] >> Fabulous question [26:13] >> and and and the addendum here and what [26:15] percentage of EBIDA [26:17] needs to be contracted before the market [26:19] should rerate it? So, that's a great [26:21] question and and I only had six minutes [26:23] to do this and I think I blew through my [26:24] time so I couldn't get into this kind of [26:25] detail, but it's something I would have [26:27] really wanted to get into. So, whoever [26:28] asked that, thank you. Uh, I just use [26:30] the 15 multiple because it's sort of a [26:32] blended multiple between the contracted [26:34] stuff, which will get a big premium [26:36] multiple because, you know, it's a bond [26:38] like cash flow stream and bond like cash [26:40] flow streams trade at a small spread to [26:42] treasuries and so treasuries if they're [26:43] at 5% should trade at 20 times plus some [26:46] growth or whatever, plus or minus. the [26:48] uncontracted stuff, the merchant stuff [26:50] that has spot market exposure is more [26:51] volatile, less visible, that should [26:54] trade at a lower multiple. We can get [26:56] into the minutia, but just suffice to [26:58] say, the more contracts, the higher the [27:01] multiple, the less the lower the [27:02] multiple. Use 15 times as a good rule of [27:04] thumb, and you'll probably get to the [27:06] right answer, which is what I used. [27:07] >> That last question from Daniel Sherer. [27:09] Thank you for that, [27:10] >> Dan. Thank you. That was great. [27:12] >> Great. Thanks. [27:18] My name is Oleg Nelman. I'm the founder [27:20] and managing director of Eco1 Capital, a [27:22] San Francisco based valueoriented [27:25] biotech fund uh that I started about 13 [27:28] years ago. Thanks a lot to the besties [27:30] for having me here. I'm a huge fan of [27:33] the pod like I'm sure all of us are and [27:35] I know how challenged Science Corner can [27:37] get. So, I wrote this in a way that even [27:40] David Saxs would appreciate and pay [27:42] attention to if if he were here. [27:45] >> Well, paradoxically, he's taking a nap, [27:47] which is what he normally does during [27:48] science corners. [27:49] >> Exactly. [27:51] >> Generally speaking, investing in biotech [27:53] companies is a horrible idea sandwiched [27:56] somewhere between movies, wineries, and [27:59] spaxs. In fact, our sector often feels a [28:03] lot more like a casino than an actual [28:05] financial market. And most of the [28:07] tourists who are investing are playing [28:08] the slots. [28:12] Of course, at Eco1, we consider [28:14] ourselves poker players in a sector [28:16] where virtually everyone else is a [28:17] momentum investor betting on science. We [28:20] focus on margin of safety. We're one of [28:22] the few funds not managed by PhDs or [28:24] MDs, and that's by design because we [28:27] don't want to fall in love with the [28:28] science. We fall in love with the [28:30] riskreward. And like the slide says, we [28:32] want to monetize other kids science [28:34] projects. [28:36] This is my 25th year investing in [28:38] biotech. I started my career with an [28:40] 11-year stint at another fund and [28:42] launched Eco R1 in 2013, humble [28:45] beginnings with 13 million. Since [28:47] inception, we've 10xed to our investors [28:49] and annualized at 20%. And today we have [28:51] about two and a half billion under [28:53] management. [28:55] We're lucky to have long-term partners, [28:57] many of whom are biotech entrepreneurs [28:58] themselves, and have been with us since [29:00] day one. And we recently reopened for [29:03] the first time in four years. [29:08] Today I'm going to tell you about a [29:10] company that's on the front lines of the [29:11] war on cancer. Military terminology has [29:14] been used when descri describing [29:16] treatments for the disease since the [29:17] early 70s when President Nixon signed [29:19] the National Cancer Act. [29:23] The warfare analogy is actually perfect. [29:28] The warfare actually is actually perfect [29:29] for cancer because both domains are [29:31] trying to accomplish the exact same [29:32] thing. find the enemy, figure out the [29:35] best weapon to kill them, and have [29:37] minimal unwanted casualties along the [29:39] way. [29:42] First, a quick history of how this war [29:44] has evolved. [29:47] Early surgical cancer treatment and [29:48] radiation was akin to a medieval siege. [29:51] Level the entire castle, burn the [29:53] surrounding village, and hope the enemy [29:55] was left somewhere in the rubble. [29:59] Chemo actually evolved from an [30:00] accidental observation during World War [30:02] I that mustard gas killed rapidly [30:05] dividing tissue. Tumor cells divide [30:07] fast, so doctors would flood a patient's [30:09] body with chemo and hoped it killed the [30:11] enemy faster than it killed allies. [30:13] Unfortunately, hair, skin, gut, and [30:16] marrow cells also divide quickly, and [30:18] the poison doesn't discriminate. [30:20] First generation targeted therapies were [30:22] next, like a GPS guided munition. [30:25] Instead of carpet bombing every dividing [30:27] cell, you identify the enemy's command [30:29] and control center and destroy it. The [30:32] problem, like with any weapon, is that [30:33] the enemy adapts and hides. And in [30:36] cancer, these are called resistant [30:37] mutations. [30:42] Imunotherapy was first introduced to [30:44] patients a decade ago. With IIO, you [30:46] don't send in your own troops. You [30:48] recruit local allies, also known as T- [30:50] cells, and let them do the fighting for [30:52] you. Spectacular when it works, but [30:54] highly dependent on the terrain or the [30:56] tumor micro environment. This brings me [30:58] to the reason we're here today. [31:00] Modern-day radio pharmaceuticals. Like a [31:02] swarm of micro drones small enough to [31:04] navigate the bloodstream and find their [31:06] target by molecular recognition, then [31:08] detonate a precisely sized warhead with [31:10] a blast radius of 100 microns or the [31:13] diameter of a single cell. An autonomous [31:15] assassination with the force of a bunker [31:17] buster and minimum collateral damage. [31:20] The company I'm going to tell you about [31:22] today is Actis Oncology. The ticker is [31:24] AKTS. The company has a billion dollar [31:27] market cap, a $500 million enterprise [31:30] value, and a stockpile of cash, which [31:32] should last them over 3 years. Long past [31:35] critical milestones that are coming next [31:37] year. Actis was started five years ago, [31:40] but recently went public with a $300 [31:42] million IPO that was 18 times [31:43] oversubscribed and backstopped with a [31:46] hund00 million order by Eli Liy, the [31:48] folks who bring you all the weight loss [31:50] drugs. The company has designed a [31:52] platform that can carry any radioactive [31:54] payload, is complex enough to go after a [31:56] variety of targets, and small enough to [31:58] clear your body with minimal side [31:59] effects. The beautiful thing about this [32:01] approach is that physicians can verify [32:03] target engagement in early clinical [32:05] trials with imaging. This significantly [32:08] derisks clinical development because you [32:10] know the drug is getting to the tumor. [32:12] Another de-risking strategy for their [32:14] first few programs Actis chose known [32:16] valid targets like nectin 4 and B7H3. [32:20] Nectin 4 is critical in bladder cancer [32:22] and the company's second program [32:24] targeting B7H3 is even more ambitious [32:27] expressed on every major solid tumor [32:29] including the big three prostate, [32:31] colurectyl and lung. Actis started [32:34] clinical trials last year [32:37] and is publicly guided to initial [32:39] clinical data in both of these lead [32:41] programs in 2027 with Nectton 4 coming [32:44] as early as Q1. So you won't have to [32:46] wait long. If either program shows a [32:48] signal, the company is likely to get [32:50] value not only for those programs but [32:52] the entire mini protein platform. This [32:55] is the holy grail in biotech, getting [32:58] value simply for the promise of what [33:00] might come. [33:04] What's even more compelling is there's [33:06] an amazing amount of interest in [33:08] radiotherapies from pharma. The big ones [33:10] including Bristol, Novartis, Bayer, and [33:13] Lily who backs stop the Actis IPO have [33:15] been building radiotherapy capabilities [33:17] and they're hungry for assets to add to [33:19] their pipelines. There's been 15 billion [33:22] in M&A and dealmaking in radiotherapy in [33:23] the last few years and we're very much [33:25] in the early innings. The neatest thing [33:28] about this modality is that it's very [33:29] hard to replicate. Generics generally [33:31] don't traffic in radioarma and because [33:34] the class involves radioisotopes it's [33:36] off limits to China. So unlike most of [33:38] biotech there's a real moat [33:42] and now the obligatory safety warning. [33:44] Axis is not for everyone. You should [33:45] consult your biotech analyst before [33:47] purchasing access. Initiating a position [33:48] may cause increased anxiety reduced [33:49] sleep through the night. Serious [33:50] sometimes drops in stock rates occur in [33:52] biotech. Immediately after investing may [33:53] experience sudden volatility due to lang [33:55] competitors. If stock declines are [33:56] experienced no fun reason. Call your [33:57] broker immediately to increase your [33:58] position. Remember serious safety [33:59] concerns have risen in other companies [34:00] local development programs. While no [34:01] safety concerns occur with any access [34:03] program date, they may in the future. [34:04] The use of many proteins delivered for [34:05] sales has not been proven. Access is no [34:07] market products and thus no referring [34:08] revenue solution for equity offerings [34:10] may occur. In the event of a secondary [34:11] offering, immediately schedule a call [34:12] with access management team to discuss [34:13] placing an order. [34:22] It's notoriously challenging to value [34:24] biotech companies because when you risk [34:26] adjust and discount back, you pretty [34:28] quickly get to zero. For earlier stage [34:31] opportunities like this, we like to [34:32] triangulate. We think Actis could be [34:34] worth 10 billion or $200 per share if [34:36] even one of their programs makes it to [34:38] market. And in this case, you have a lot [34:40] of outs. [34:41] >> I'm not familiar with why radioisotopes [34:43] are off limits to China. So in this [34:47] particular case um Actis' radioisotope [34:51] payload is actinium and actinium is [34:54] manufactured from radium 233 which was [34:58] used in our own nuclear programs in the [35:01] US in the in the 50s and 60s. So it's a [35:03] waste product from there. So actinum is [35:06] not even available in other countries [35:08] like China because they had a completely [35:09] different uh their own program was [35:11] completely different with enriched [35:13] uranium and plutonium. But the but the [35:14] the risk for a lot of biotech and China [35:17] replication came about that Amgen Sophi [35:20] Supreme Court case, didn't it? Where [35:22] they could you could make a small [35:23] because it basically said all patents [35:25] are composition of matter patents. So [35:27] you could change one amino acid get [35:29] around the patent and China's basically [35:31] done that with a lot of biologics that [35:33] are patented in the US and Europe. They [35:35] just rip them off and then you attach [35:37] the radio emmitting [35:38] >> uh radioisotope to the molecule and you [35:41] can kind of chase it. That's kind of why [35:42] a lot of biotech's been depressed. Is [35:44] that not true? [35:45] >> Yeah. So with with with radioisotopes [35:47] again because you have to you have to [35:48] have a a manufacturing supply that you [35:51] have to source locally in the US. Um [35:56] we haven't seen any competition coming [35:57] from China at all. [35:58] >> And if if they have a successful readout [36:00] though, would it not be like the case [36:04] that someone in China would say, "Hey, [36:05] let's go get some of the necessary [36:07] radioisotopes." And [36:09] >> they're they're I'm sure they can do it [36:11] for the Chinese market, but in terms of [36:13] then transfer transferring that over [36:15] here, we we haven't we haven't seen it [36:17] or kind of any wind of it at all. [36:19] >> And so then my last question, I'm sorry [36:20] for monopolizing. Why do you think the [36:21] markets discounted the value so much [36:23] since the IPO? [36:24] >> Oh gosh. Given the return in biotech [36:26] >> valuation, it's pretty classic biotech. [36:28] So it's traded flat since the IPO. uh [36:31] biotech investors are so insanely [36:32] short-term oriented that even though [36:34] we're now call it eight or nine months [36:37] from data that's still way too long and [36:39] so our expectation is the folks will [36:41] start accumulating this in the second [36:42] half in anticipation of the data coming [36:44] in the first quarter [36:46] >> Gavin you had a question [36:47] >> yeah sure so I in in in the distant past [36:51] I ran a biioharmaceutical fund uh and [36:54] you know it's a very hard job [36:55] congratulations on those numbers but I [36:58] ran that fund right after the human g [36:59] genome had been sequenced and there was [37:01] an expectation that the sequencing of [37:03] the genome was going to lead to this [37:05] explosion in therapies personalized [37:07] medicines etc etc and I don't think [37:10] broadly speaking we've made as much [37:12] progress over the last 25 years as maybe [37:15] people thought in the early 2000s and my [37:18] hypothesis is that the genome is too big [37:21] of a problem space for the human mind or [37:24] software written by humans and AI is [37:27] going to unlock a lot of kind of [37:29] revolution utionary therapies. So my [37:31] question to you, I will just admit it's [37:32] a selfish question. It is not about your [37:34] stock pitch, which is great. It's what [37:36] do you think the odds are that in the [37:39] lifetimes of everyone in this room, the [37:42] average human lifespan in a developed [37:44] country extends well past 100 to 125, [37:48] 150. I would take the over on that uh in [37:53] no small part because we already have [37:54] one of the best longevity drugs out [37:56] there and folks don't even realize it in [37:58] the glip ones and the obesity drugs. So [38:00] one of the only things that's ever been [38:01] shown in act with actual data to extend [38:03] life is caloric restriction and that's [38:05] literally what all the obesity drugs do. [38:07] So I'm sure half the people in this room [38:08] are on one of them. Uh and that's just [38:10] the beginning because it's trained [38:12] people that you can inject yourself with [38:15] something and have healthy living [38:16] through pharmaceuticals. So I I think [38:18] that's only going to continue. [38:19] >> Oh, I got two questions from the [38:20] audience. First one, as the launch costs [38:22] per kilogram continue to fall, is there [38:25] a credible pathway to use space and [38:27] microgravity as a therapeutic variable [38:30] given that cancer cells appear to behave [38:32] differently in lowgravity environments? [38:35] >> That is a great question that's probably [38:36] not applicable to this. [38:38] >> Okay. And then the second question, what [38:41] would be a technological breakthrough [38:43] that could disrupt precision [38:45] radiotherapy as a result of AI at scale [38:48] to drug development and precancerous [38:50] screen? [38:50] >> Yeah, another awesome question. There's [38:52] a a small skunks work project within [38:55] Actis AI project. So with all these [38:58] biotech companies, they have their [38:59] little proprietary data sets that they [39:00] hope to leverage with various insights. [39:03] So a company like this with their many [39:04] proteins and everything else they're [39:05] trying to accomplish, they have their [39:07] own little tiny group of PhD data [39:10] scientist nerds who are seeing if they [39:12] can leverage that in a pretty decent [39:13] way. Is it so it's been really hard to [39:17] get CARTT in solid tumors? Um is it the [39:21] case that these kind of personalized [39:24] uh uh peptide based imunotherap uh [39:26] therapies are showing some efficacy in [39:28] some solid tumors and is that uh a space [39:31] that's going to expand and kind of [39:32] intersect here? [39:33] >> What's most promising that I think a lot [39:35] of folks have probably heard of is uh uh [39:38] a new drug for pancreatic cancer from a [39:40] company called Revmed was just another [39:41] targeted therapy. So for now there's not [39:45] a huge amount of progress from from [39:46] peptides. And have you looked at Droins [39:48] before? These kind of right-handed [39:50] proteins that seem to be able to [39:51] penetrate solid tumors. [39:53] >> Well, so one of the one of the neat [39:54] things about these mini proteins is [39:56] they're hopefully of the right size to [39:59] be able to deliver their payload inside [40:00] of the tumors. [40:01] >> Incredible. Oleg, thank you. [40:03] >> Thank you. [40:09] >> By the way, by the way, somebody just [40:11] yoloed into the stocks while Oleg was on [40:13] stage. It's up 6%. like the L. Don't do [40:16] that while we're all trying to buy as [40:18] well. Please come on. [40:19] >> Morning everyone. My name is Kyle [40:21] Samani. Uh thank you for being with us [40:22] at the all liquidity today. Thanks to [40:24] the besties for organizing. Today we're [40:26] going to be talking about a little known [40:27] asset, a little crypto asset called [40:29] Geonet. Uh which is building the rails [40:32] for AI. So let's jump in. Quick bit [40:35] about me. Uh I founded a firm called [40:36] Multicoin Capital about eight and a half [40:38] years ago. I stepped down a few months [40:39] ago. Um, and in my time there, I was [40:42] probably most well known for, uh, [40:43] leading all three rounds of investment [40:44] in Salana prior to Salana's network [40:47] launch, uh, in 2020. Um, I've been deep [40:50] in the crypto space for a very long [40:51] time. Uh, and I thought this would be a [40:53] very natural forum to talk about a very [40:55] interesting investment at the [40:56] intersection of crypto and AI. Uh, also [40:59] big shout out to David Sax. [41:00] Unfortunately, he's not here, but, uh, [41:02] David did seed multicoin uh, back in the [41:04] day. So, thank you, David, for believing [41:05] in me very early. All right, let's get [41:07] into geoet. So the way to understand [41:10] Geonet first is to look at GPS um [41:13] probably everyone in this room has been [41:15] in the situation on the left where [41:17] you're using your phone and your phone [41:19] is in the wrong spot facing the wrong [41:21] way. Right here you can see this guy [41:23] looks like he's facing a wall according [41:24] to his phone. Um, Geonet it [41:27] fundamentally is a new is it uses a [41:30] technology called RTK or real-time [41:32] kinematics where you can localize your [41:34] location down to about 2 centimeters for [41:36] context. GPS roughly the the precision [41:38] is about 2 m. So you're getting about [41:40] 100x accuracy for very precise [41:43] geoloccation. Uh, as you can imagine any [41:45] form of kind of robotics can make use of [41:48] RTK drones being the very obvious [41:50] example. I'll touch on a few more just [41:52] in a couple minutes here. [41:55] Um, today Geonet is the world's largest [41:57] RTK network in the world and it's also [41:59] the fastest growing. Um, the three [42:01] companies you see on the left here, [42:02] Trimble, Hexagon, and Topcon, have all [42:04] been building RTK networks in some form [42:06] or fashion for call it 20 to 30 years. [42:09] Um, all of them combined have roughly [42:10] 12,000 base stations deployed around the [42:13] world. GeoNet was founded in 2021. Uh, [42:16] began building out the network in 2022. [42:18] Uh, and today they are roughly twice the [42:19] size of the next three guys combined. Uh [42:22] today's geoet is live in 150 countries [42:24] around the world, more than 11,000 [42:26] cities and covers roughly 80% of the [42:28] global population excluding some [42:30] sanctioned countries. Um so this thing [42:32] is really growing quickly. Uh you might [42:34] say, how did these guys build this [42:35] network so fast? Uh and the key is [42:36] really this decentralized crypto model. [42:38] Um so here we're looking at literally a [42:40] photo of a geoet base station on the [42:42] roof of someone's house. The the global [42:45] geoet network, those 22,000 nodes are [42:47] not being built and deployed by some one [42:49] that looks like AT&T or Verizon. Those [42:51] base stations are being deployed by any [42:53] random guy or hobbyist or professional [42:56] or small business owner who wants to [42:57] make some extra money. You can go on the [42:59] Geonodes website today. You can buy one [43:01] of these base stations. They're a few [43:02] hundred bucks. You put it on your roof [43:04] of your house or your small business. It [43:06] broadcasts radio waves. You make money. [43:08] Uh you actually get paid in geode [43:10] tokens, which is the really cool part [43:12] about this incentive system to bootstrap [43:13] this thing to get it off the ground. So [43:15] the Geonet network started about 4 years [43:17] ago doing this. Today it's now the [43:19] largest fastest growing in the world by [43:20] a pretty wide margin. If you want a [43:22] sense of scale, uh here we're looking at [43:24] their coverage in United States. [43:25] Obviously, every single major metro is [43:27] covered, but even if you look at most of [43:28] the rural parts of the country, you're [43:30] covering actually the vast majority of [43:31] even the rural areas. Um let's talk [43:34] about some of the customers and use [43:35] cases for this. We'll start with [43:36] agriculture first. Um the USDA actually [43:39] launched a couple years ago a program to [43:41] encourage farmers and ranchers to use uh [43:43] precise a technologies including RTK [43:45] networks. Um today actually geo the USDA [43:48] is now actually subsidizing uh many [43:50] farmers and ranchers all over the [43:51] country uh to adopt high precision a [43:53] most of which is powered by geodet. [43:56] Um getting into some specific examples [43:57] of that here we're looking at what's [43:59] called a robotic mule. Um this is made [44:01] by a company called Burrow. Uh obviously [44:03] this is transporting some grapes. You [44:04] could put anything on this. Has pretty [44:06] obvious application for bas almost any [44:08] farm or ranch you can imagine. With the [44:10] advent in computer vision, CPUs, [44:12] batteries, all the other AI stuff, these [44:14] things are growing like hotcakes, all of [44:16] them are going to be powered by Geonet [44:17] or something like it. Um, here we're [44:20] looking at John Deere. They have a new [44:22] service uh that they rolled out recently [44:24] called Global Unmanned Spraying Systems [44:26] or Gus. Um, these things drive around. [44:28] They literally spray plants with [44:29] pesticides and other things of that like [44:31] that. I did actually confirm this [44:33] morning there are wineries here in Napa [44:34] that are actually using John Deere Gus [44:37] uh uh vehicles. That was pretty cool. So [44:39] if you have some wine tonight, maybe it [44:40] was powered by by Gus, which powered by [44:43] Geonet. Um, obviously autonomous [44:45] vehicles has a pretty obvious [44:47] application for this. Um, TomTom is one [44:49] of Geonet's customers. TomTom is a [44:51] supplier to basically every every AV [44:53] program in the world, uh, excluding [44:55] maybe a couple. And today, TomTom is [44:57] using Geonet's data to update their maps [44:59] to get them more accurate and precise as [45:01] they need to cover every square inch [45:02] basically around the planet. [45:05] Um, one of my favorite use cases are [45:07] kind of the next wave of consumer [45:09] robotics, which are getting a lot of [45:10] hype these days. I think the most [45:11] obvious one are robotic lawnmowers. I [45:13] don't think anyone loves to mow their [45:15] lawn. Um, robotic lawnmowers are now [45:17] actually rolling out at pretty good [45:18] scale. They're estimated they're going [45:20] to sell 1 million robotic lawnmowers [45:21] this year, made by companies like Yarbo, [45:23] Sunseeker, and others. Um, all of those [45:26] these guys are all powered by Geonet. [45:28] Um, [45:30] next up, let's get to drones. Um the [45:32] world's largest drone manufacturer DJI [45:33] is a Geonet customer. Uh it's not in all [45:36] of their models, but it is in a lot of [45:37] their models. Uh and so obviously DJI is [45:39] sending a ton of traffic now over [45:40] Geonet. Uh in the coming uh you know [45:43] months and years as DJI wounds down in [45:45] the US and you have new a wave of [45:47] American drone manufacturers pop up. I'm [45:49] going to venture to guess that most if [45:50] not all of them are going to end up on [45:52] the GeoNet network as well. Uh the [45:54] Geonet team is based in the US has deep [45:55] roots here. What I love about Geonet is [45:57] it's a very obvious network effect [45:59] networks effect network effects [46:01] business. Um this thing looks like a [46:03] natural telecom, right? You have base [46:04] stations kind of all over the world. You [46:06] got to cover the whole planet. Um [46:08] telecoms naturally form monopolies [46:10] historically. I think the same is likely [46:11] to be true here. Today, Geonet is the [46:13] world's largest and fastest growing [46:15] network with also the lowest cost [46:16] structure by a very wide margin because [46:18] of this decentralized nature where [46:20] people just put these things on top of [46:22] their house. [46:24] In terms of where the business at, the [46:25] business just crossed about $1 million [46:27] in annualized run rate a few days ago. [46:29] Uh, and it's growing more than 3x [46:30] year-over-year. I think it's going to [46:32] probably more than triple over the next [46:33] 12 months. Um, what's really cool about [46:35] GeoNet is how capital efficient it is [46:37] and how they're actually returning [46:38] capital to token holders. So today, uh, [46:41] the Geonet network is taking of that 11 [46:43] million in revenue roughly, uh, excuse [46:46] me, 80% of it is being used to make open [46:48] market purchases of Geo tokens. Uh, and [46:50] this is all visible on the Salana [46:52] blockchain. they have all the addresses [46:53] are published and stuff. So, it's all [46:54] verifiable in real time. Uh that means [46:57] $8.8 million right now per year is going [46:59] into buying, you know, Geonet tokens on [47:01] the open market. Uh what's amazing is [47:04] that last 20% is they're they're [47:05] covering all their R&D costs and scaling [47:07] out now their business development team. [47:09] With a business like this, of course, [47:10] like it's a pretty small network of [47:12] customers. The guys who work at John [47:13] Deere know the guys who work at DJI, who [47:15] know the guys who work at TomTom. And [47:16] so, this thing is now growing virally [47:18] amongst this kind of core community of [47:20] of customers. Um, and as you can imagine [47:23] with customers who sign up for a service [47:24] like this, they tend to ramp up their [47:26] usage of that service over time. So once [47:28] someone starts rolling out GeoNet in the [47:30] first year, they're usually spending [47:31] about $60,000 per year. After two years [47:33] though, they're usually spending about [47:34] $170,000 per year. So the average Geonet [47:37] customer is growing their their revenue [47:38] with GeoNet about 3x in that second [47:40] year. Um, obviously then we look at [47:42] their just their customers. They've [47:44] signed up in the last two years. You can [47:45] see they 5x their customer base last [47:47] year. Those are that net new customers. [47:49] So applying some pretty simple math [47:50] here, you can see they have a very clear [47:52] path to more than 3x this year as this [47:54] thing ramps up. Just to wrap things up [47:56] summary, um Jonet is the world's largest [47:58] RTK network, growing the fastest. It has [48:00] really obvious network effects and is is [48:02] likely to be a very natural monopoly [48:04] growing 3x year over year with a bunch [48:05] of flagship customers and brands that [48:07] you all know. Obviously, we have this [48:08] huge physical AI tailwind behind us now, [48:10] robotics and all the other amazing stuff [48:12] happening. Uh and they're returning [48:14] capital to shareholders. Uh the token [48:16] does trade on the Salana blockchain. Uh [48:18] if you want to buy, it trades 247. The [48:20] ticker is geode geo d. Um so if you want [48:23] to actually get some geo tokens, I [48:25] encourage you to sign up for a crypto [48:27] wallet, a salon wallet, and you can go [48:28] ahead and buy geode tokens from there. [48:31] And with that, I think we are ready for [48:33] some Q&A. [48:36] >> Awesome. Um what's the market cap? [48:39] >> Oh, sorry. It's trading about 150 [48:40] million on a fully diluted basis. If you [48:42] were to go look at any of the crypto [48:44] price websites like coin.go require [48:45] market cap. They're going to show you [48:46] something like 60 or 70 million. That's [48:48] because not all of the tokens are [48:49] floating yet, but the fully diluted [48:51] number is about 1 million. [48:51] >> Is there a corporation behind it or is [48:54] this just like a project in the Cayman [48:56] Islands in Panama with a board that [48:59] nobody knows who's on it? Tell us about [49:02] governance. [49:03] >> Uh so the Geonet team is uh US-based [49:05] corporation. There four teams in San [49:07] Francisco. The CEO's name is Mike [49:08] Horton. Uh really, really good guy. He's [49:10] been building in this kind of IoT smart, [49:13] you know, device space for a while. the [49:15] relationship between the corporate [49:16] entity and the token and which one [49:19] should we own? [49:20] >> Uh you should own the token because I [49:22] own a lot of the token uh as as you [49:24] might imagine. I don't own any of the [49:25] equity. Um the relationship is geoet the [49:28] c you know the company is facing John [49:30] Deere DJI all these companies and they [49:33] have a contractual relationship with the [49:34] Geonet Foundation to use 80% of their [49:36] revenues to buy tokens off the open [49:38] market. And that's [49:39] >> that corporation raised venture capital [49:41] or anything? Uh yes uh my prior company [49:44] Multicoin actually led around in Geonet [49:46] previously. [49:47] >> Okay Kyle I have I have many questions [49:49] from the audience so bear with me. [49:50] Question one do you like Helium as much [49:53] which is Geonet for 5G signal. [49:56] >> Uh yes I actually led multicoins [49:58] investment in Helium six or seven years [49:59] ago uh and continue to be a very big [50:02] long-term believer. They actually had [50:03] big news go out this morning. Uh but [50:05] yeah I'm a big Helium fan. [50:06] >> Question two. There's a long list of [50:09] deepin projects that have failed because [50:11] people just don't value the token [50:13] rewards. Why is this any different? [50:15] >> I mean, they're returning capital to [50:16] shareholders. This thing is, you know, [50:18] returning $8.8 million to shareholders. [50:19] It's trading at $150 million valuation [50:22] and it's going to grow 3x this year. [50:24] It's an unbelievably cheap asset. It's [50:26] just people aren't paying attention [50:27] because it's crypto bare market right [50:28] now. [50:29] >> Okay. Uh from Sam, [50:30] >> sorry, let it's a securitized interest [50:33] in the cash flows from the customers [50:35] >> effectively. Yes. It is a revenue. It is [50:38] a [50:38] >> It's a revenue share token. [50:39] >> Correct. 80%. [50:41] >> Okay. So, the more John Deere pays [50:43] Geodet the company, the more you [50:46] basically deprecate the tokens, which [50:48] should cause the token, [50:50] >> they're buying tokens to open the [50:51] market. Correct. [50:52] >> Yes. [50:52] >> Okay. From Sam, what acres value the [50:54] equity or the token? Similar to [50:56] question, how does the value acrruel [50:58] mechanisms square or not with current [51:00] securities laws or what's contemplated [51:02] in the Clarity Act? [51:03] >> Uh, yeah. Um, so the one answer to your [51:05] question is the the tokens are the ones [51:07] acrewing value because they're taking [51:08] 80% and buying. The other 20% is [51:10] obviously funding operations. They have [51:11] engineers, salespeople, all that stuff. [51:13] Um, so that's all there and being [51:15] funded. In terms of securities laws, uh, [51:17] the Clarity Act, uh, passing is [51:19] certainly very good for Geonet. Uh, I'm [51:20] not a lawyer, so I'm not going to tell [51:22] you that, you know, it it passes the [51:23] bars set in the Clarity Act, but I can [51:25] tell you I'm an optimist and I've been [51:26] very involved in the Clarity Act, uh, [51:28] and I'm not too worried about it. [51:30] >> Okay. Can I can I ask about the business [51:32] just real quick? So John Deere, I know [51:34] this space somewhat well. I used to [51:36] manage a company called Precision [51:38] Planting in Agriculture and um there was [51:41] John Deere makes their own RTK systems. [51:43] So when you like run a a piece of [51:46] equipment that relies on RTK, you're [51:47] buying in the construction industry, [51:49] Topcon or Leica or Trimble or John Deere [51:52] and you install the RTK base stations [51:54] and you run your equipment. Why would [51:57] John Deere and others want to rely on [52:00] this system as a different like why is [52:02] it better than like the systems that [52:04] they're already using? It wasn't quite [52:06] clear to me. [52:07] >> I mean capex versus opex, right? Like [52:09] right these networks are all over the [52:10] world now. They're running at very low [52:12] cost. GeoNet is is probably a third to [52:14] half sorry a third to a quarter the [52:16] price um than buying up your own capex [52:19] and doing it. And it's just available [52:20] everywhere. So now it just reduces the [52:22] sales cycle time for John Deere when [52:24] they just say buy the tractor. It's [52:26] good. [52:26] >> There's another big push right now for [52:28] microats to be an alternative to GPS in [52:31] a way that they can actually provide [52:33] subcm resolution effectively replacing [52:36] both GPS and RTK using a a mesh network [52:40] from SpaceX launched or actually SpaceX. [52:43] I don't know if SpaceX has looked at [52:44] doing this, but um I know that there is [52:46] a very wellunded company that is trying [52:48] to put up microats to basically replace [52:52] GPS and RTK. Doesn't that ultimately [52:55] kind of wash out the need to have all [52:56] these earth-based base stations? [52:58] >> I there's no chance they can compete on [53:00] cost because just sending things to [53:02] space with satellites that's so I mean [53:03] these geoet uh base stations are a few [53:06] hundred bucks. Like you're just not [53:07] going to compete on cost with with [53:08] geoet. [53:10] >> Do you think that this is a viable [53:11] replacement at scale and saturation for [53:14] GPS itself? [53:15] >> No. GPS is definitely very different. Uh [53:18] and the SLA is different. [53:20] >> You've got to you've got to have [53:20] ubiquity for [53:21] >> Yeah. for GPS um for GPS alternative [53:25] which is why you have to have the [53:26] satellites everywhere. You got to have [53:27] enough but if you get enough satellites [53:29] you can actually get to RTK precision [53:31] and you don't need to have the big [53:32] expensive GPS. [53:33] >> You'd have a hybrid situation where you [53:35] have a bunch of GEO and LEO plus a bunch [53:37] of base stations all over the place. [53:40] That hybrid situation probably [53:41] >> you could actually the Leo the LEO alone [53:44] can replace all of the the geo stuff. [53:46] That's the goal. And then um if you get [53:48] enough of them which SpaceX unlocks. [53:51] Yeah. [53:51] >> And um Kyle, what what about like other [53:54] tokens when you think about other [53:57] compute tasks like work to be done for [53:59] example? There's a bunch of tokens that [54:01] have emerged in distributed training. [54:03] How did you hone in on this and exclude [54:05] the others? [54:06] >> I mean, not meaning prefer this over [54:08] that. [54:08] >> I mean, I met the GeoNet founder years [54:10] ago. He pitched us and I've gotten to [54:11] know him and followed it. The [54:12] distributed training stuff, there's a [54:14] whole bunch of people trying it. I I'm [54:16] pretty skeptical. I don't think any of [54:17] it's going to work. Um the distributed [54:19] inference stuff is is possible although [54:21] it has not worked as well as we would [54:22] have hoped. I did put some money behind [54:23] that a few years ago. It's it's working [54:25] but not not A+. Um one last thing [54:28] actually David on on your prior question [54:29] I want to highlight is also energy use. [54:31] Going to space just consumes way more [54:33] energy than going to a base station [54:35] that's you know on the ground. Um and so [54:38] yeah for a tractor maybe that doesn't [54:39] matter but for a drone or for any other [54:40] battery sensitive application ground is [54:43] always going to be uh the preferred [54:45] solution. [54:45] >> Super interesting. Well done. Thank you. [54:47] Thank you so much. [54:49] >> All right, guys. [54:51] >> Before we vote, Timoth, give your [54:53] feedback. [54:53] >> Here's what I like. I I I apply the Stan [54:56] Ducken Miller school of invest [54:58] investigate. I really believe in it. [54:59] >> Yes. [55:00] >> Um if you don't have any skin in the [55:01] game, you don't care. And this is the [55:03] kind of stuff that I love. I love [55:04] hearing ideas like this. [55:05] >> I love all four. My my difference is in [55:08] sizing. So, you know, there's there's [55:12] certain asymmetric alpha that each one [55:14] of these exhibits and then there's very [55:16] different downside risk for each of [55:18] them. [55:19] >> Uh, and then there's also liquidity [55:20] issues. So, for example, like I love [55:22] Kyle's idea. The problem is I I could [55:24] not get enough working for me where so [55:27] you I don't even think I could get a [55:28] million dollars in today. It would to [55:30] scale in it would move the market. Um, [55:33] so I would have to I'd have to probably [55:34] I'd be like 10 20 30,000 and then maybe [55:37] start to buy into it. talent. I think [55:39] they could absorb tens of millions and [55:41] people wouldn't bite an eyelash. Um [55:45] the biotech company the issue there is [55:47] that I think that there is as you said [55:49] Freberg this discontinuous [55:52] illi liquidity zero risk but then [55:54] there's the 10x upside. So there's just [55:56] like huge [55:58] >> Lily will bid for it [55:59] >> and then MGM I think is just so I think [56:00] MGM and Talon are the ones you could [56:02] have huge sizing in [56:03] >> and then the other ones I think you have [56:05] a piece because they're like lottery [56:06] tickets. I think your point on MGM. [56:09] >> Okay. Wait, hold on. Let me just review. [56:10] So, company number one [56:13] was MGM [56:14] >> and that was [56:15] >> Resorts. Okay. [56:17] >> Company number two, [56:19] >> Talon. [56:20] >> Talon Energy. [56:21] >> Company number three, [56:22] >> Actis [56:23] >> Therapeutics. Yeah. [56:24] >> And then [56:25] >> Geodet. [56:26] >> Geodet. Not company, but I guess token. [56:29] >> Yeah. [56:29] >> Company number four, Geodet. [56:31] >> And you're buying the token, not the [56:32] company. Do you think the um maybe for [56:36] you too, Gavin, like the [56:38] >> Gavin, you rank them. [56:39] >> Well, no, even before you rank, just [56:42] tell us what you think of the format and [56:44] then assess the companies. We'll do [56:46] ranking at the end. We're going to do 4 [56:47] 32. We're going to do 4 3 2 1 on stage, [56:50] but give me your general ideas about the [56:53] pitches, what you liked, what you did. [56:54] >> I thought the pitches were great. Um I [56:57] thought the format was amazing. I would [56:58] for sure expand it next year. There are [57:00] platforms that you guys could have a all [57:03] all-in basket or ETF that people could [57:07] trade in. So like maybe that's [57:08] something. [57:08] >> Will you do it next year? [57:10] >> Will I pitch next year? [57:11] >> Yeah. [57:12] >> Jake, I'll do anything. He's locked. [57:15] He's locked. [57:16] >> Actually, here's what I would ask Gavin [57:17] to put you on the spot. Next year I [57:19] would we I think we would all learn and [57:21] benefit if you would do [57:23] >> um silicon and memory super cycle. [57:28] >> Sure. Would you be willing to do that [57:29] for us? [57:29] >> I'll do it. [57:31] >> Sign me up. [57:32] >> Perfect. Great. Sign me up. Oh, thank [57:33] you. So, keep going. [57:34] >> Well, no. As far as the pitches, I do [57:36] think um I think it's important to [57:38] disagregate like what was a really great [57:41] entertaining pitch versus what I think [57:42] is is a really good riskreward. Um I [57:46] thought Oleg and Kyle did a great job [57:48] with the pitches, but I'm I'm not a [57:51] healthcare investor, nor am I a crypto [57:53] investor. I thoroughly enjoyed the [57:55] presentations. I actually thought GET [57:57] was very interesting. Um I'm happy to [57:59] learn from Oleg that I might lift well [58:01] into my 100s. That was good news for me [58:03] and everybody in the room. I enjoyed all [58:05] the um the military um terminology and [58:08] analogies. [58:09] >> Yeah, that was really great. Huh. [58:10] >> That was great. [58:11] >> Really great. [58:12] >> I do think um from a pure riskreward [58:15] perspective, I thought MGM was the best. [58:18] um your downside is really capped [58:19] because of the Barry Diller bid and then [58:21] you have uh Japan and Dubai as I think [58:24] very valuable future sources of value. [58:27] Uh and and I do think talent is also a [58:29] very compelling risk. I just think [58:32] everything in AI is going to need to [58:35] grapple with increasing regulatory risk [58:37] which we talked about last time um that [58:40] that I was on the pod with you guys and [58:42] I don't know how to dimensionalize that [58:45] and um you know I've been [58:48] >> like the big negative externality for [58:50] talent is nothing to do with talent. [58:52] >> Nothing to do with talent. [58:53] >> It's like something over the top from [58:54] the US government caps prices something [58:56] something. [58:57] >> Yeah. You have nationalizes the lab. [58:59] >> You have a change in administration. You [59:00] have a change in Congress. There's laws [59:02] that are passed that I think it [59:05] terrestrial computes the utility supply [59:08] demand. But I actually think outside of [59:10] that, Talon was super compelling. And [59:12] >> so you got MGM, you got Talent. Now the [59:14] other two. [59:15] >> I I I I thought they were both great [59:17] pitches. I can I tie them for third just [59:20] >> Well, no, don't even give the score. [59:21] just any feedback on those two ideas or [59:24] those are just a little bit lottery [59:26] ticket for you or [59:28] >> No, I thought Actis was was very [59:30] compelling. They're trying to trying to [59:32] do something different as Oleg said if [59:34] you ever get a biotech company that be [59:36] can become a platform and they have a [59:39] mechanism whether it's of drugging [59:42] whether it's targeting or if you have [59:45] something that is broadly applicable [59:47] that is when you can get these really [59:49] really big hundred billion dollar plus [59:51] outcomes in biotech which are rare so I [59:53] thought that part of um activis was [59:56] super compelling [59:57] >> and um [59:58] >> you don't play crypto [60:00] >> I don't play crypto, but I thought the [60:02] entire go Geonet discussion was [60:04] fascinating and I'm happy. [60:05] >> Is there anything that would get you off [60:06] the bench and make you jump into the [60:10] crypto game or it's just you're why are [60:12] you not playing the crypto game? [60:14] >> I feel about crypto exactly the way I do [60:17] about snowboarding. Okay. I'm I'm not a [60:21] very good athlete. I've spent a lifetime [60:24] learning how to ski and I'm okay. Um, [60:28] and just the idea of getting on a [60:29] snowboard, having, you know, thousands [60:31] of hours of ski instruction, [60:33] >> you don't want the pain for the game. [60:35] >> Yes. And I have 25 years of lessons, [60:39] learnings, pain, scars from from [60:43] investing in equities and public [60:44] securities and just crypto. It's a [60:47] little bit like snowboarding for me, but [60:48] like, you know, everybody who wants to [60:51] snowboard, that's great. [60:52] >> Everybody wants to do crypto, that's [60:54] great. Just please don't go sideways [60:55] down the mountain and ruin the powder. [60:57] David, [60:58] >> I think your assessment of MGM Talon, [61:01] >> I think I think MGM uh I look at the [61:03] kind of return upside, the downside and [61:07] the timeline. MGM's like probably a 3x. [61:10] I think it's also missing this point [61:12] that I've heard a lot about on you can [61:14] actually upgrade the monetization on [61:15] these Vegas properties. We were talking [61:16] to a friend of ours in Vegas. They're [61:19] they're making a million bucks a day in [61:20] incremental ibita every day that they [61:23] have a show at the sphere um at the uh [61:26] at the Venetian hotel which is an [61:28] unbelievable statistic which tells you [61:30] that when you have the entertainment [61:31] draw the gambling revenue just flies [61:34] >> flies [61:35] >> and so Barry Diller I have heard [61:37] separately has been spending a lot of [61:39] time on trying to reinvent the [61:40] entertainment at these properties and [61:42] thinks he has an idea on how to do it [61:43] which will cause the gambling revenue to [61:45] fly. So, I think even if you discount [61:47] the upside on these new locations, [61:49] there's probably a lot of work to be [61:50] done. And I do like the floor on the bid [61:53] and then you got call it 3x in 2 years [61:56] even if this bid goes nowhere and they [61:58] keep the thing running and they're like [61:59] we're going to reject the bid and keep [62:01] running independently. Talon is maybe 3x [62:05] upside, 5x upside, but it's eight years [62:07] out. And I think one of the other [62:08] challenges with Talon that I would kind [62:10] of use as a valuation metric is I think [62:12] it's more interest rate sensitive than [62:14] MGM is um because the power purchase [62:17] agreements really are where a lot of the [62:18] revenue comes from. So you're going to [62:20] get a discount rate that's a multip [62:21] that's a function of where interest [62:23] rates are sitting. So I think if [62:25] interest rates shoot up which some might [62:27] argue there's there's risk there you [62:29] actually get margin compression from [62:31] that 15x outlook that he has for Talon. [62:34] So that would be my kind of downside [62:35] scenario on Talon. um in the in the time [62:38] ahead and Actis I do worry because I'm [62:40] an investor in a company that's got a D [62:41] protein conjugate that shows really [62:44] strong efficacy in getting solid tumors. [62:46] I think that there are new modalities [62:47] for therapeutics for solid tumors that [62:49] are being discussed that that may kind [62:51] of put this at risk. I think the China [62:53] risk is legit because I've seen it [62:54] across the board in biotech. Everything [62:56] gets ripped off and people go to China, [62:58] but they could have a hit and Lily could [63:00] bid on it in 6 months if they actually [63:01] get a good readout. So there's certainly [63:03] upside, but the downside's probably 50 [63:05] 75% if they get a bad readout or China [63:08] or some new modality comes out. So I [63:11] think the ranking is probably MGM Talon [63:12] Actis and then the for me the geodet [63:15] piece uh I I just think the space thing [63:17] is is likely the path. It's going to [63:19] replace all RTK and all GPS in the next [63:22] decade. Um it's an inevitable piggyback [63:24] on systems that are already going up. [63:26] >> All right, great. So I think I've got [63:27] everybody. Uh for me, I put them into [63:29] two buckets. Um I think uh AKTS and GOD [63:35] those are like lottery tickets could be [63:37] crazy returns but you know there's a [63:38] there's a big probability of a zero [63:40] there if they don't you know uh actually [63:44] work and then MGM and Talon obviously [63:46] got the downside protection and those [63:49] feel like um people will always gamble [63:52] and leave the lights on. So I kind of [63:54] like both of those. I put 200k into each [63:56] in real time. [63:57] >> Gamble and leave the lights. Uh, so [63:59] that's just like my I don't have a [64:00] public vehicle. [64:00] >> Did you actually buy? [64:01] >> I'm just day trading. [64:02] >> I I bought half of his action [64:06] and I don't have a Robin Hood account. I [64:08] have to call my office. So I was like [64:10] just I'll take [64:10] >> you lose, buddy. [64:11] >> I'll take half your I didn't see you [64:14] with your thumbs. [64:14] >> I'm up 7% across the portfolio. So I [64:17] don't think I can include you here. [64:18] >> Steps to buy. [64:19] >> I did. I I waited the three in the order [64:21] I said on my So anyway, I I'll just give [64:24] mine really quick. I I will go MGM, [64:26] Talon, uh, G, AKTS. [64:31] >> Gavin's only gonna make trillion. [64:32] >> Let's bring the Let's bring our four [64:34] pictures out. [64:35] >> We have $1,000. [64:36] Please get the [64:38] >> two men hugging statue. Wait, wait, no. [64:40] Before you announce it, I need the [64:43] extremely alpha male heterosexual [64:45] trophy. The all-in heterosexual alpha [64:49] male trophy, please. And I need our four [64:51] pitchers to come on stage. It makes it [64:54] more exciting. It's like makes it [64:55] uncomfortable when like they show the [64:58] five people for best actor. [65:02] >> Yeah. [65:04] >> Yes. [65:09] >> You put those on the table. [65:11] >> But wait, where's my award doing? The [65:13] award. You guys have the award. [65:14] >> Please bring me the extremely [65:15] heterosexual alpha male award. [65:19] >> You'll see why when I show you the [65:20] award. Pass me this. [65:21] >> Over a little more. [65:22] >> All right. [65:23] >> Bring me that award. Let me show you how [65:25] we 3D modeled this. [65:26] >> No one wants to see this. [65:27] >> Look at this. This is two men [65:30] >> uncomfortably hugging. And the way we [65:32] did this, it's the best. Come here, [65:33] Freeberg. I'll show you. I'm not doing [65:34] it with you. [65:35] >> Come on, Freeberg. [65:36] >> You do it tomorrow. [65:36] >> Free. Okay, fine. He's extremely [65:38] comfortable. [65:39] >> That's David and I. [65:40] >> It's David and you. But let's show them [65:41] how we modeled this. We just did a long [65:43] >> This is uncomfortable. And we hold it [65:45] for five extra seconds. [65:47] >> At 2 minutes, you get the release of [65:48] oxytocin. There it is. [65:50] >> Okay. So, gentlemen, this is it. Do you [65:53] guys have the results? Go ahead. [65:54] Audience award. [65:55] >> Audience award. [65:57] >> So, [65:58] based on 150 votes from the audience, [66:02] uh, do I just go four to one? [66:03] >> Four to one. [66:04] >> Four to one is more exciting. [66:05] >> Okay. Fourth place, uh, with 5% of the [66:08] vote was Kyle Sani. [66:09] >> Okay. Well done on the board. [66:13] >> A very close [66:15] second place. [66:16] >> No, third. Third. [66:17] >> Third place with 21% of the vote. Oleg [66:21] rock [66:22] >> leg. Oh boy, we're closing in here, [66:23] guys. Very dramatic. [66:26] >> And with 50 50% [66:30] who's number two. [66:30] >> I'm going to go to you. No, you say [66:32] number one now. [66:33] >> Okay. Okay. [66:33] >> No. Okay. Well, sorry. Yeah, you're [66:35] right. With 24% of the vote [66:38] >> in second place, [66:42] >> Aaron Cowan for MGM. NUMBER ONE WITH 50% [66:45] of the vote. Dan Drafus. [66:47] >> Wow. Unbelievable. Give it up. Nicely [66:51] done. Now the bestie though. [66:52] >> Wait, hold on before we do the bestie. [66:54] How do you feel right now having won [66:56] this? Pass him the award. You guys look [66:58] so uncomfortable. [67:00] >> You guys are really It's very [67:03] >> But pass him his award for a second and [67:04] let him hold it. [67:05] >> Give it academy award. [67:07] >> Thank everybody how you got to this [67:09] place. [67:09] >> Say a few words. I got my award. I got [67:13] my tequila. [67:14] >> Thank you. [67:14] >> There you go. All right. Well done. [67:16] >> Okay. Now Okay. [67:18] >> That's the award. 4 3 2 1. It's uh [67:21] relatively similar here. Uh fourth place [67:24] was Kyle Simmani. Okay. [67:25] >> Third place was Oleg. [67:27] >> Second place Dan Drafus. First place [67:31] Aaron Cowen. Big upset. Flip the [67:33] audience vote. [67:34] >> There you go. [67:34] >> All right. So MGM wins. [67:36] >> All right. [67:37] >> Thanks guys. [67:37] >> This was amazing. [67:39] >> All right. Thank you all for [67:40] participating. [67:41] >> Thank you very much. [67:42] >> Thank you so much for coming and we'll [67:44] see you next [67:53] I'm going all in.