[BidClub_]
Sohn Conference Foundation · · 36 分钟

深入挖掘:基本面投资对谈

Andrew BellasEduardo MarquesDavid RosenJonathan LennonMitch Golden

YouTube
TL;DR
  • David Rosen(Rubric Capital)的核心判断是:随着成交量不断流向那些都依赖动量的量化基金、宏观基金和 pod 基金,一只下跌的股票“根本没有买家”——“我可以出手买下成交量的50%,股价仍然会跌10%。” 他的选择是 Viatris(VTRS):按当前盈利7倍、远期盈利6倍交易,现金转化率100%;6-9个月后管线数据读出,未来12个月股价将涨逾1倍。衍生交易则是 Idorsia(IDIA SW)——若 QUVIVIQ 取消受控药物分类、拿到清醒适应症,且 ADHD/自闭症儿科数据全部兑现,股价可从4瑞郎升至42瑞郎,成为潜在10倍股。Rubric在3周内买入了该公司10%的股份。
  • Andrew Ballou 的多头是 Carvana:它是二手车行业最大、盈利能力最强的公司,但在一个规模达1万亿美元、由4万家参与者构成的市场中仅占1%份额——1家公司同时具备这些特征极其罕见。 7年的购车周期叠加口碑转介绍,意味着2026年的收入在2021-23年就已埋下种子,令30%以上的增速异常容易外推;按明年税前自由现金流约20倍估值,“我不知道它2年内能否成为10倍股,但5年或10年内可能做到。”
  • Eduardo Marques(Pretensul Partners)的难题是:如今标普500指数约23%的权重在半导体和光学股上,很难不把 AI 机会当作周期股;所以他的主仓是“韩国价值投资的可卡因”——SK Square 相当于以47%折价买入 SK Hynix,Samsung Life 则是李氏家族持有 Samsung Electronics 股权的载体,交易于0.48倍账面价值、约4倍盈利。 催化剂包括外资流入正在扭转韩国散户连续10年买入 Mag7 的趋势、政府的 Value Up 计划复制日本治理改革,以及 Interactive Brokers 上月刚向美国散户开放韩国市场。
  • John Lennon(Pleasant Lake)的逆向多头是 Reddit:股价从高点下跌约50%,却被错误地定价为 AI 输家;互动仍在增长(帖子和评论从240亿增至250亿),Google 和 OpenAI 的授权协议将在下半年按“约5-6倍”重定价,而 Anthropic 诉讼“要么对 Reddit 很好,要么对 Reddit 极好”。 Reddit按他对明年预测数字的14倍估值(按2028年约10倍)交易;Meta 的每用户收入(RPU)约70,而 Reddit 只有20出头;若 Reddit 被重新归入 AI 赢家篮子,股价可涨至现价的2-3倍。
  • John Lennon 给出的全场最锋利的做空逻辑是:如今最好的做空标的是面临“服务层面通缩”的高质量复利公司,这种压力在短期业绩中不可见;首先就是澳大利亚分类信息公司 REA,其永远把价格推高5%的模式会在 AI 智能体介入购买流程后失效。 如今在曼哈顿搜索一套3居室公寓,“在 OpenAI 上,结果已经足够好”,而房源直接来自 Compass,不是 Zillow。
  • 反向的 AI 赢家判断是:“真正的 AI 赢家会是一家平庸的企业”,只需把低 EBITDA 利润率提高2个百分点。 John 的算式是:利润率从50%升至52%只增加4%,但从50个基点升至2.5%就是增加400%;这正推动 Pleasant Lake 通过上市转私有交易,以每1美元20-50美分的价格买入“看起来就是垃圾公司”的零售商。
  • 主持人对市场结构的观察也构成一张机会地图:伊朗战争推高油价后,消费公司业绩超预期、指引也很好,“却没有一家上涨”——所有非 AI 资产都在充当资金来源,留下一个已经去风险、可以部署大量资本的标的池。
摘要 · 为研究而整理的核心内容

1. 深度价值却无人接盘:从 Viatris 到 Idorsia 的10倍股

  • Rosen 开场说,“做深度价值投资的人很孤独”,这也是他的优势:依赖动量的量化基金、宏观基金和 pod 基金主导成交量,因此股价下跌时“根本没有买家”。“我可以一直把所有抛售的股票都买下来。”
  • 核心仓位是 Viatris(VTRS):35%是仿制药,65%是品牌仿制药和专科药,增速处在低至中个位数;按当前盈利7倍、远期盈利6倍估值,利润100%转成现金,其中一半通过分红和回购返还,另一半用于并购——包括从流动性承压的 Idorsia 手里买来的管线:selatogrel,“基本就是心脏病发作患者的 EpiPen”,以及一款狼疮药(现场听成“cenegermod”,很可能是 cenerimod)。两款药都有潜力成为重磅药物,将在6-9个月内读出数据;这正是 Rubric 认为 VTRS 未来12个月会涨逾1倍的原因。
  • 继续追查卖方,带出了更大的交易。Idorsia 的 QUVIVIQ 是一种不引起嗜睡的食欲素拮抗剂类失眠药,“在中国和欧洲卖得火热”,但在2023年美国获批时被列为受控药物,原因是 FDA 不确定这一新类别是否会成瘾。受控药物分类给其成为重磅药物制造了2道重大障碍:医生会避开任何有滥用风险的药物,续方必须线下就诊。如今已有数十万患者的无滥用信号数据,Idorsia 正寻求取消 QUVIVIQ 的受控药物分类。
  • 重估催化剂来自一笔交易:Contessa(很可能是 Centessa)以63亿美元出售,交易逻辑是食欲素失调与 ADHD 和自闭症相关;Idorsia 已表示,其儿科 ADHD/自闭症试验达到统计显著性,并已向 FDA 提交无滥用信号的数据。Rosen 的账是:取消受控药物分类加上清醒适应症,增加每股22美元,或者说26块钱;自闭症/ADHD/儿科数据加上 SPV 资产再增加16美元——从今天的4瑞郎到42瑞郎,“如果这些事情全部兑现,就是一只10倍股”。

2. Carvana:罕见的 AI 受益者

  • Ballou 的电商筛选逻辑基于 AI 席卷电商之后的一个判断:“有人把孩子和洗澡水一起倒掉了。”拥有自有产品、自有数据和庞大实体基础设施的公司,可能是 AI 受益者,而不是受害者。
  • Carvana 的罕见之处,是1家公司同时具备3个特征:行业最大、盈利能力最强,却只占1%的市场份额——这之所以可能,是因为可服务市场规模达1万亿美元,由4万家高度分散的参与者组成。
  • 增长的可预测性异常高:消费者会把购车推荐给“另外10个朋友”,但美国人平均每7年才买一次二手车——所以2026年的收入在2021-23年就已埋下种子,今天的转介绍则在为2029-31年的收入奠基。
  • 贷款端有结构性优势:Carvana 年复一年把贷款卖给同一批投资者,这会激励更保守的承保,而不是像只需把车卖出去的销售员那样;按每年100万辆的规模做车辆整备,使得车辆被收回时,对贷款机构而言每辆车的价值高出500-1,000美元。长期保持30%以上增速时,估值约为明年税前自由现金流的20倍。

3. 韩国是价值交易主线:“我们仿佛仍处在亚洲危机时期”

  • Eduardo 的困境代表了整个讨论:既要相信“这场新的工业革命”,又要坚持安全边际;但当标普500指数约23%的权重落在半导体和光学股上,就很难不把 AI 机会当作周期股,而一旦按周期股看,估值就显得偏满。已知的对比是:Micron 按2026年盈利10倍交易,SK Hynix 约6倍;两者至少未来2-3年都是“巨额现金流机器”。
  • 利用韩国折价有2种方式:无投票权优先股和控股公司。SK Square 本质上相当于以47%折价买入 SK Hynix,控股公司承诺通过回购和分红收窄这一折价。“这种机会在世界其他地方不存在。”
  • 最曲折、也最令人兴奋的是 Samsung Life:名义上是保险公司,实际上是李氏家族用于掌握 Samsung Electronics 约10%股权的载体;Samsung Electronics 是1万亿美元级 DRAM/NAND 巨头,按当前盈利7倍估值,旗下零利润率晶圆代工业务一旦修复,可能带来类似 Intel 的繁荣。按市值重估这笔持股后,买入 Samsung Life 相当于按0.48倍账面价值、约4倍盈利交易,且盈利按明年股息流入的传导计算。
  • Eduardo 认为现在的催化剂包括:贸易收支向好;外资流入正在替代此前把资金搬进 Mag7 的“韩国疯狂散户赌徒”;政府的 Value Up 计划复制日本的治理改革;以及 IBKR 上月刚向美国散户开放韩国市场。John 的韩国之行见闻则让这笔交易更有说服力:Hynix 将10%的经营利润以员工奖金形式再分配,Samsung 也在照做——相当于约360-370亿美元的利好,约占 GDP 的中个位数百分比;按明年一致预期,Samsung 是全球利润最高的公司,Hynix 第二,Nvidia 第三。

4. Reddit,以及其他所有“资金来源”篮子里的标的

  • John 用来筛选多头的行为测试是“我们最害怕做空的东西”,这让 Reddit “位列我清单最前面”。它在 AI 输家因子篮子里从高点下跌近50%,但帖子和评论量从四季度的240亿增长到本季度的250亿。
  • 他预计下半年出现拐点:Google 和 OpenAI 的授权协议会带来估值重估——按可比口径,他的研究认为约为5-6倍——而 Anthropic 诉讼的结果“要么对 Reddit 很好,要么对 Reddit 极好,你只能二选一”。变现差距也很大:Meta 的 RPU 约70,Reddit 只有20出头;他预计收入增长50%以上、盈利增长75%以上,按明年预测数字14倍,按2028年更接近10倍——若重新归入 AI 赢家篮子,股价可涨至现价的2-3倍。空头占比处于高个位数至低双位数。
  • 主持人把视角拉开:伊朗战争推高油价后,消费公司公布了不错的业绩和指引,“尽管如此,没有一家上涨”——因为资金只想要 AI。他的结论是:市场里有一整片已经去风险、却被忽视的标的,足以部署大量资本。

5. 最好的空头是优秀企业;最好的 AI 赢家却是平庸企业

  • John Lennon 回答做空问题时发表核心观点:“识别输家比识别赢家容易”(这句话他归功于 Third Point 的一封信);真正的危险是短期业绩看不见的“服务层面通缩”,所以优秀公司仍在交出漂亮数字,股价却在下跌。永续价值的疑问,正从软件、BPO 和呼叫中心扩散到整个知识经济。
  • 他的样本空头是 REA:澳大利亚房地产分类信息公司,也是该领域估值倍数最高的股票。眼下还没有 AI 威胁,但2-3年后,AI 智能体介入购房流程,将打破其在受众不增长的情况下永远每年提价5%的模式。概念验证已经出现:今天在曼哈顿搜索一套3居室公寓,“在 OpenAI 上,结果已经足够好”,而房源也会直接来自 Compass 等公司,不只是 Zillow。对这些复利公司来说,转成价值股要跨过一道鸿沟——到那时,它们可能是我们应该做多的高现金流收益率标的,但中间有个大坑。
  • 主持人用一个问题反问:今年年初的共识是,重资产企业会在 AI 时代胜出,轻资产企业会被去中介化;但市场如今已经收窄到半导体和 neoclouds。“为什么现在没人买其他东西?”John 的回答成了全场最佳金句:“真正的 AI 赢家会是一家平庸的企业”(“The true AI winner will be a mediocre business”),只要能把低 EBITDA 利润率提高2个百分点,就会带来一场巨大的繁荣。
  • John 正按这套算术行动:利润率从50%升至52%只是增加4%,但从50个基点升至2.5%就是增加400%;他通过公开市场转私有交易,把更好的管理层与 AI 工具结合起来,以每1美元20-50美分的价格买入那些看上去就是垃圾公司的零售商。这不是传统的激进股东策略,他也承认其中的孤独:“买入看起来就是垃圾公司的零售商要逆向得多……我甚至比你还孤独。”

6. 职业建议:从街头斗殴到“富足心态”

  • John 对在场的25岁年轻人坦白:他20多岁时觉得这个行业“极其令人反感”,差点离开——“我宁愿在街上和行业里大多数人打一架,也不愿分享我的投资观点。”让全场发笑的那句挖苦是:“如果你不觉得这个行业充满了混蛋,那你最好照照镜子。”
  • 转折来自2条建议:East Rock 那群人的“富足心态”,以及很可能是 Oz(Ballou 的合伙人,现场听起来像“Os Duwan”)说的一句话:“等着看看和朋友一起赚钱是什么感觉。”10年后,他把这场圆桌也视为答案:Eduardo 是“我认识的最好的做空者”,Rosen 是“我心目中在世投资人的总统山成员”;最后的教训是:“你可以极度执着、极具竞争性,但也可以拥有美好、彼此成就的良性循环关系;我真希望自己在职业生涯早些时候就知道这一点。”

Host

All right. Thanks, everyone. In the interest of time, we're going to go very quickly with introductions and then dive right into the conversation. I first want to thank all the panelists. I've known this group for a long time—a decade-plus, I think, for all of you—and it's really a privilege to be up here with some of the best investors in the industry today who have volunteered their time to share their ideas with all of us. So, maybe real quick, Andrew, and then we'll go down the line: quick introduction, please.

Andrew Ballou

I'm Andrew Ballou. I manage General Equity Holdings, founded in 2017. We're a long-short, concentrated investor based in New York.

Eduardo Marques

Eduardo Marques from Pretensul Partners. We manage about $2 billion in a long-short strategy. We're not so much concentrated and are very much mid-cap oriented.

David Rosen

David Rosen, Rubric Capital. We are a deep-value fund, and we also dabble in distressed. This is our 10th year.

John Lennon

John Lennon, Pleasant Lake Partners, also in our 10th year. We have an equity long-short hedge fund, a private equity drawdown structure, and then a multi-manager investment platform as well.

Host

All right. I'm going to start with Dave, because you said you're a deep-value investor. It doesn't feel like that type of market. Can you just talk about what it's like, what you're doing, and share an idea with all of us?

1. Deep Value Finds Viatris

David Rosen

Sure. It is a lonely thing to be a deep-value guy. And I'll tell you what: it also does create opportunity. If you just think about the world that we live in right now, most of the volume and the capital has gone to quants, macro funds, and pods. Almost all of those strategies require momentum as a component.

When you see stocks that go down meaningfully, that's otherwise known as visual volatility. There are literally no buyers. If I want to buy a stock, I can go out and be 50% of the volume, and the stock will still go down 10%.

Host

That must be fun.

David Rosen

Yes, I can literally buy all the stock all the time. So, within that context, I'm going to quickly pitch an idea that meets that criterion. Just by way of background, we're big fans of a company called Viatris; the ticker symbol is VTRS. It is a legacy generic drug company. Thirty-five percent of the business is generic drugs. The remaining 65% are branded generics and specialty pharmaceuticals.

The company is going to grow low to mid-single digits. It trades at 7 times current-year earnings and 6 times forward earnings. One hundred percent of those earnings is converted to cash flow. Fifty percent of it gets returned to shareholders in the form of dividends or buybacks, and they use the remaining 50% on M&A.

What's fascinating about Viatris is that they went and acquired a pipeline. That pipeline was coming from a company called Idorsia. The ticker symbol is IDIA SW. Idorsia had some liquidity issues, and they were forced to sell these assets. The 2 assets they sold were a product called selatogrel and likely cenerimod.

Selatogrel is literally like an EpiPen for somebody who has a heart attack, and it would effectively stop the heart attack from happening. Cenerimod is a lupus drug. These are both prospectively blockbuster indications. They were sold, and it's why we think Viatris is going to be more than a double in the next 12 months. These products are going to read out in the next 6 to 9 months.

As we dug into Viatris, we spent a lot of time trying to understand Idorsia. What Idorsia did is take these drugs, along with another drug that they own called Tryvio, which is an approved product that they have actually not launched yet, and put them into an SPV. They raised debt against it, basically taking debt off their balance sheet and putting it into this SPV vehicle. If any of these things hit—if either one of these drugs hits, or if they're able to sell that other drug—they pay off the debt, and the remaining value will flow to Idorsia.

That was interesting, but the more we dug into Idorsia, the more we were surprised by the incremental asset they have. They have a product called QUVIVIQ. QUVIVIQ is a drug for insomnia, but unlike normal drugs for insomnia that cause drowsiness, this one has no drowsy side effects. You actually are very wakeful on it. The drug sells like hotcakes in China and Europe, but it has a problem in the United States because, when it was approved in 2023, it was scheduled.

The reason why is that this is called an orexin antagonist, and it was something that hadn't been approved before. The FDA basically said, “Listen, since we don't know whether it's addictive or not, we're going to schedule it, and we'll find out. Come back to us.”

Now that they've had hundreds of thousands of patients on it, they want to come back and get it descheduled. The problem with the scheduling is that if you're a doctor and you're going to give a drug to a patient, if there's any risk of it being abused, that's a big problem. Another problem is that if you actually want to get it prescribed and get an updated prescription, you have to go into the doctor each time. It can't just be manually done. Again, there are huge thresholds for this drug to actually get to the blockbuster level that it should be.

What recently happened is that another company called Contessa got sold for $6.3 billion. Contessa has a very similar drug, except it's an orexin agonist, so it handles narcolepsy. What was really exciting to people was that they realized that dysregulation of these orexins is often an issue for people who have ADHD and autism. They're buying it because they think that this drug will work for those indications.

It turns out Idorsia just ran a trial for QUVIVIQ in pediatrics, ADHD, and autism. They've already come out and said that the trial was successful. It hit statistical significance. But oftentimes with these companies, they don't want to give you all the data. They want to do it at a conference or in a journal, so nobody knows the details behind it, and as a result, nobody is valuing this.

To the extent that the studies come out and are favorable for all these indications, all of a sudden, people will value this. On top of that, remember, they say they gave this to pediatrics, and the safety data came out and was great. There were no issues with abuse. They've given that data to the FDA.

The bottom line is that the stock is a 4-Swiss-franc stock. They're also running a trial for the wakefulness indication that they have overseas. To the extent that it gets rescheduled at a lower level or descheduled, and they get that wakefulness in the label, we think that's an incremental $22 a share, or 26 bucks.

To the extent that this data, which we know is good for autism, ADHD, and pediatrics, works out, and any of these things in the SPV work out, that's an incremental 16 Swiss francs a share. And again, within 3 weeks, we bought 10% of the company. We actually think that the upside is 42 Swiss francs versus 4 today, or a 10-bagger if all these things hit.

Host

Love 10-baggers. I didn't understand most of that except that it makes you drowsy. But I love 10x. You should start with, “I have a $4 stock that's going to $40.” Then people might have listened to the deep-value pharma talk. And so, Andrew, I'm sure you've got a 10-bagger.

2. Carvana Scales Used Cars

Andrew Ballou

Or something close to it. Yeah. I've been spending time on e-commerce in the last 6 or 12 months. As the AI onslaught has come through, I think some babies have been thrown out with the bathwater. If you're an e-commerce company with proprietary product, proprietary data, and vast physical infrastructure, AI may mean you are a significant beneficiary rather than simply surviving.

In that vein, I'll pitch Carvana to the group.

Host

Long or short?

Andrew Ballou

Long.

Host

Okay, just to clarify. I want to put it out there and make sure we know what side we're getting.

Andrew Ballou

Carvana is the largest online retailer of used cars in the country. It buys and sells cars from consumers and delivers them to their doorstep in a day or 2. No haggle, low price, great value proposition to the consumer.

The thesis is that it's a big market, a great team, big moats, wide moats, and growing really fast at a reasonable price. I think the stock is going to go up a lot. I think a number of people in the audience may be familiar with the story, so I'll just mention a couple of things that make it unique, in my opinion.

The first is the rarity of what Carvana is. There are 3 characteristics that Carvana has, which I think are incredibly rare to find in a single company. One is that it's the biggest company in the industry. It's the most profitable company in the industry. And yet it still only has 1% market share.

Those 3 things in a single company are incredibly rare to find. The reason is that the addressable market is so vast. It's $1 trillion, and there are 40,000 players in it. It's very fragmented.

The second observation I would make is that usually it's very difficult to project high, fast growth over the long term for any company with precision. Carvana makes it a little easier because the biggest driver of its purchases is word of mouth. Consumers love buying from Carvana. They generally refer their purchase to 10 other friends and say what a great purchase it was.

Elena

Typically, a company will see referrals hit its revenue in the year in which the referral happened. But the average consumer in the U.S. buys a used car every 7 years. So, the seeds for 2026 revenue were sown in 2021, 2022, and 2023. Similarly, the seeds are being sown now for Carvana's revenue in 2029, 2030, and 2031.

The last interesting point on Carvana that I would make is that Carvana underwrites loans to consumers who purchase its cars, like the rest of the industry, for about 90% of its consumers. Structurally, Carvana's loans are advantaged relative to competitor loans. There are two reasons. One is behavioral: Carvana sells its loans to investors—the same investors year after year—so they're structurally incentivized to underwrite conservatively relative to a used-car salesman whose main job is just to sell the car, regardless of whether or not you can pay the loan off.

The second reason is that, apples to apples, the car that Carvana sells is worth $500 to $1,000 more to the lender than its competitors' cars. The reason is its scale. It processes about 1 million cars a year, relative to a used-car lot that has 200 cars in the back. When it reconditions a car, it usually means that car is worth $500 or $1,000 more. So, if there's a credit event and the consumer can't pay back the loan, the lender repossesses the car, and it's worth that much more.

Host

How much, Elena? We doubled?

Elena

It'll triple.

Host

Is it? How do you get to that number, Elena?

Elena

I think the company will grow north of 30% for a long time. They have certain markets that are mid- to high-single-digit penetrated, and today the company as a whole is only 1.5% penetrated. The stock trades at around 20 times pre-tax next year's free cash flow, so this is going to compound for a long time. I don't know if it's a 10-bagger in 2 years, but it could be a 10-bagger in 5 or 10.

Host

All right. Eduardo, we were chatting about a bunch of stuff. I know you have a variety of things you've been looking at. I'm curious to hear what's grabbing your attention today.

3. Korea Offers Deep Value

Eduardo Marques

Yeah, this is a really odd time to actually be a fundamental investor because we're trying to reconcile the fact that we believe there is something large—this new industrial revolution, this AI paradigm—that is changing the way that we should approach valuations for stocks, but at the same time we're trying to stay truthful to principles of margin of safety, et cetera. And it's been hard.

In the S&P, we were just running the numbers the other day, and about 23% of the S&P today is semiconductor companies and optical companies, this sort of hardware-centric world where every AI play is either hardware or capital-goods companies. It's hard for you not to treat the AI opportunity as a cyclical. And as a cyclical, even if you believe in AI, valuations look really full. So, what we've been doing is sort of rummaging in odd bits and ends of the world and trying to find value opportunities in the space. I think Korea really stands out as the place where we find most value.

The example that I think is most widely known in the Western world is the valuation discrepancy between Micron and SK Hynix, whereby you can buy Micron at 10 times earnings or SK Hynix at 6-ish. These are 2026 earnings, and this is actually converting into free cash flow. These companies will be massive cash gushers for at least the next 2 or 3 years. There are all sorts of convoluted ways to reach out for the crack cocaine of Korean value investing, and I think we're there. We're sort of abusing it at the moment.

There are two ways to take advantage of Korean valuation discrepancies. One is to go for preferred shares. These are nonvoting shares that trade at massive discounts to common shares. Another way is to look at holding companies. I'll give you a few examples of what we're involved with.

Obviously, I think John is also familiar with SK Square, which is the parent company of SK Hynix. That's a very clear setup whereby you can effectively buy SK Hynix at a 47% discount, and the holding company is actually promising to close that discount through buybacks and dividend payments. This is something that doesn't exist elsewhere in the world.

The one we're most excited about recently is a company called Samsung Life. This is especially convoluted. Samsung Life is, like the name says at first glance, just a regular insurance company that does life and health insurance in Korea. But it turns out that it's actually the holding company through which the Lee family controls about 10% of Samsung Electronics.

Obviously, Samsung Electronics is the company that just reached a trillion-dollar market cap. It's a DRAM and NAND powerhouse. It makes the phones we see and the televisions, et cetera. That business today—Samsung Electronics—trades for about 7 times current-year earnings. And that's actually excluding about half of their microprocessor business, which is, in fact, a foundry that today is operating at zero margin. So, if they can fix that, we can have an Intel-like boom in Samsung Electronics.

My point is that owning Samsung Electronics through the life insurance company gets you to a massive discount. And how massive? Obviously, this is a bit of a Berkshire Hathaway question: you're owning an insurance company and you're trading the underlying parts. But in this case, the easiest way to see this is that if you just mark to market the Samsung Electronics shares that the life company owns, you're buying Samsung Life at 0.48, a little under 50% of its book value. And if you just take the dividends that Samsung is going to produce next year and run them through the net income of the life insurance company, you're buying it at about 4 times earnings.

This is how powerful these discounts are, and they've existed forever. Our thought process is that there's now finally this alignment of stars whereby the trade balance is favorable for Korea, the investor inflow is actually arriving from foreigners, as opposed to, for the past decade, Korean crazy retail degenerate traders actually moving money to buy Magnificent Seven stocks here. We're seeing support from government entities through programs such as the Value Up program, which is actually sort of forcing large Korean companies to review their governance protocols and effectively allow for buybacks and dividends, just trying to mirror the experience that happened in Japan.

I think this is happening very fast, and it was only last month that Interactive Brokers opened Korea for U.S. retail investors to trade. So, we're still very early in this opportunity of just formalizing Korea. It's a market that is today larger than the Canadian stock exchange in terms of market capitalization. The Korean GDP is equivalent to the Japanese GDP, and yet we treat it as if it were still in the Asian crisis. I think that as the market is maturing very quickly, this is creating the opportunity.

Host

All right. Thanks, Eduardo. I'm going to turn to John. Before you, I know we were just catching up—you just spent several weeks in Korea and Hong Kong, in Asia. Do you have a similar thought? Are you finding stuff that's of interest in the region? Then I would love to hear your idea.

John Lennon

Yeah, it's wild, right? We were just in Korea, and it's not that it's about the size, or just that it's larger than Canada; it's that—no offense to the Canadians in the audience—it just eclipsed Canada, and the world is almost coming to Korea. They're leaders in defense, shipbuilding, and memory. If you believe in a multipolar world, where there's going to be regional rearmament, and certainly if you believe in AI, it's just wild. It's amazing.

I think we met with a bunch of former Samsung employees when we were there, and they were jealous of the Hynix employees because Hynix redistributes 10% of operating profit in employee bonuses. Samsung is now doing the same thing, so there's going to be, on our math, a $40 billion—our math was $36 billion to $37 billion—tailwind in bonuses to Korea this year, which is like a mid-single-digit percentage of the GDP or something, just based on the bonuses for those 2 businesses.

If you look at the actual operating profit, on consensus numbers and certainly even more so on ours, for next year, Nvidia is the third most profitable company in the world. Hynix is number 2. Samsung is number 1, and obviously people don't think about it. So, we're involved in all these things. It's hard not to be.

You're not going to get Dave Rosen interested in charts that look like these ones, but if they're single-digit earnings multiples, growing hundreds of percent, and buying back stock, it's hard not to be involved. I think the spillover effects to the rest of Korea—and where we spend a lot of time, the consumer—could be massive and unprecedented.

Host

We'll see. Maybe, as we're saying this, it's the very top, and you guys get to turn around and flip and find some things to be short, but it was a very powerful research trip.

Yeah.

Host

How about—we've focused the whole panel on contrarian, out-of-favor ideas, but we were chatting. Give us a really out-of-consensus, non-obvious name that you like at the moment.

4. Reddit Becomes An AI Winner

Yeah, so I jotted down a few names. The one I'm going to mention isn't maybe so out of consensus, and you can press me for more and I'll say them quickly. I'm going to mention this one because the founder was just speaking. I didn't get to do primary research backstage, but everything that is perceived not to benefit from AI is in a factor basket that's getting destroyed.

Reddit shares are down almost 50% from the highs. Sometimes, when we approach investing on the long side, we think about the thing we'd be the scaredest to be short, just behaviorally, and it would be at the very top of my list. I think the total posts and comments on Reddit have grown from around 24 billion in 4Q to 25 billion this quarter.

There is probably a behavioral inflection that will occur in the back half of this year, where the licensing agreements that they have with Google and, at least, with OpenAI will be repriced. Our work suggests they could reprice on an apples-to-apples basis at around 5 or 6x, and then there will be a massive shift in perception at that moment. At the same time, Reddit is in a lawsuit with Anthropic. The outcome of that will be either great for Reddit or amazing for Reddit. You can only choose one.

At the same time, they're monetizing. Meta's ARPU is around 70; Reddit's is in the low 20s. We have them growing revenues over 50% and earnings over 75%, and it's around 14 times our next-year earnings number, closer to 10 times looking out to 2028. If all of a sudden this gets put in an AI-winning basket, it's between a double and triple from here. So that one's pretty interesting to us—ostensibly not so contrarian historically, but more recently a bit more contrarian.

Mason

Is there high short interest? Is it still a battleground stock?

John Lennon

High single-digit to low double-digit short interest. But just look at the chart and what happens on the days when, you know, pick a meme-ing AI stock—anything that is perceived as an AI loser in a software or internet basket does. It's definitely starting to be prime.

Mason

You got me thinking. One of the things that we've seen—and it's not just AI losers—is that when the AI stuff is going up or retail investors are engaged in the meme, everything else in the market becomes a source of funds.

One of the questions that we had recently was: after this Iran war, as oil prices went up, how would companies respond? Would they be able to pass it through? How would earnings be affected? There was a lot of nervousness around performance, and a lot of these stocks went down fairly meaningfully because there was fear.

Many consumer-facing companies actually reported good numbers and provided good guidance. Despite that, none of them went up. It was a function of the fact that people don't really care. They wanted to deploy capital in AI. So I think there's this universe of ideas that have more recently gotten de-risked, where you can deploy a significant amount of capital.

All right, I'm going to ask: is anyone brave enough to talk about a short idea? We don't really get that in the afternoon. For those of you courageous enough to come in the morning, we have a short panel where we force the issue.

John Lennon

I'm happy to go again.

Mason

We have a volunteer.

John Lennon

I'm happy to say, “Hey, I'm trying to sell my wares here, man.”

Mason

All right, let's hear it.

John Lennon

Come on. I've got 80 different short ideas, so I'm happy—

Mason

Take one. Okay, I'll give you 3 minutes for as many as you want.

John Lennon

Let me just take the one. I've got my soapbox here, so this is my one opportunity to actually speak my mind.

5. Quality Businesses Become AI Shorts

Let's just say that I think that, so far, the market has been very concentrated in trying to find these AI winners. Basically, everything today has the valence of an AI loser or a winner. What's clear in my mind—and this has been repeated; I'm not the first to come up with this line—is that it's easier to spot the losers than the winners. I think this was in the Third Point letter.

What is attractive right now, and also just really, really hard, is that the best short ideas are actually high-quality, compounding, great businesses. This is my contention: if you believe in a future of AI, then you really have to be watching out for service-level deflation, something that is not visible today in near-term earnings.

I think a lot of this is reflected in the performance of some of these stocks over the first quarter. People are scratching their heads because great companies are actually printing good numbers and their stocks are actually going down. What I think is happening is that this question of perpetuity value is slowly spreading away from just software and the business-process-outsourcing companies and call centers. This is now spreading to all of the knowledge economy.

With that in mind, I think there are a couple of amazing businesses that stand out as shorts for me. One is the whole online-classifieds space, in particular the one that we're short, a company called REA Group in Australia. This is real-estate online classifieds. Do I see today a clear threat of AI to the business of online classifieds? The answer is no.

Going back to your point about e-commerce, there are clearly e-commerce winners in the space. But if I fast-forward the models that we see today to a world 2 or 3 years down the road, is it possible that agentic AI will be good enough—and is it even in the roadmap of companies like OpenAI—to try to monetize their models by actually figuring out a way to interject in the process of buying and replace what we consider to be this amazing supply-and-demand marketplace moat? I think the answer is yes.

If you run a search today in Manhattan for a 3-bedroom apartment, you'll find that if you run it on OpenAI, the results are good enough, and the listings are not all coming from Zillow. They're actually coming directly from Compass and everyone else. So the ability for the growth algorithm of these online classifieds to just push price 5% a year forever, even if there's no audience growth—I think that business is broken.

REA stands out because it's actually the highest-multiple stock. I think what we'll find across these knowledge companies is that there will be a chasm whereby they're going to go from high-multiple, forever-compounding stories to value opportunities. Once they arrive there, then they're probably great for cash-flow-yield stories that we should be long, but there's a big hole.

Mason

Okay, so one of the things earlier this year—I think there was a recognition that the AI winners would be companies that were heavily asset-intensive, with a very high asset intensity.

John Hempton

Well, there was an acronym for it. I just forgot it.

Mason

Right. I forgot. But the concept was that anything that had low capital intensity could get disintermediated. The ones with high capital intensity—you use AI and you actually improve your margins, right? But all of a sudden now, when everyone talks about AI, no one's thinking about that. It's just the semiconductor companies. It's just—you have to be a neocloud, right?

The question is: when does that come back? Do we really want to buy the ones that are asset-light, that may trade at a really good multiple but literally can get completely disintermediated?

John Hempton

Yes.

Mason

Probably the answer is maybe even not, right? Maybe those aren't the buys. But why is nobody buying the other stuff now?

John Lennon

Yes. The true AI winner will be a mediocre business that will just be able to expand a low EBITDA margin by 2 percentage points. That will make a massive boom.

Mason

Right. And, John, I suppose that this is part of your thesis too around some of the, let's call it, value plays that you're involved with in the retail and consumer names, right? The margin opportunity from AI.

John Lennon

Yeah, if something has a 50% margin that goes to 52%, that's a 4% increase. But if it's 50 bps and goes to 2.5%, that's 400%, so there are some interesting things. There are a lot of public-to-private transitions that we're slowly attempting to instantiate, where enacting better management with good old-fashioned strategic and operational thoughtfulness and AI tools can just lead to mitigating the downside risk. In some cases, like you guys, you can make your 10-bagger too. So, yeah, there's a lot that's very interesting there, for sure.

Mason

And that doesn't seem to have happened yet, right? The public market—I mean, you just talked about doing it in a private setting.

How come the public markets haven't done that?

John Lennon

Sorry, Mason. I know you're the moderator.

Mason

I'm here to watch like everyone else. I just roll you guys over.

John Lennon

Well, you don't know the answer, which is why, over the past few years, we've slowly tried to engage without, to be sure, being traditional activists. We only engage in conversation with management when we're genuinely strategically interested, but because we don't know the answer, we've just said, “Hey, if there's an opportunity to buy a 20- to 50-cent dollar where the dollar is going to grow a lot, let's get after it.”

The point on Reddit was that human content can be more valuable because of the amplifying impact. But obviously, buying ostensible shitco retailers is a lot more contrarian than that. I'm sure I'm the only one—I'm even lonelier than you in that.

Mason

Yeah, I'll play that.

And on that note, in a separate direction, but related to AI and human capital, we've chatted about it. I think a lot of people in the audience are starting their careers in the industry. All of you have been doing this for a really long time, though you all look awesome, young, and wonderful. But it's a different world. It is an AI world for our business. I think most people here are involved in investing one way or another.

What's your advice for a 25-year-old getting started in our industry in this crazy AI world? John, maybe you want to start with your thoughts. Andrew, we'd love to hear yours. Anyone else?

6. Human Judgment Still Matters

John Lennon

I have a derivation of the same idea, which is, in a somewhat contrarian way, that I think the human element will be all the more important. When I was in my 20s—some people may be offended by this—I found the industry to be extremely distasteful. I almost left it, and I would rather have had a fistfight in the street with most of the people who participate in our industry than share my investment ideas with them.

Mason

How do you really feel?

John Lennon

If you don't think the industry is full of douchebags, then you should probably look in the mirror—all of you in the crowd.

Mason

Ooh, wow.

John Lennon

Thank you, guys.

Mason

Damn. Over.

John Lennon

No, no. But around that time, the East Rock guys were involved with Sohn, and someone there said to me, “You should view things more with the spirit of abundance.” I was like, “What the hell is that?” I've had to claw with my fingernails for every inch of this opportunity.

Around the time I launched my firm, someone I admire for his wisdom—he might not like me referencing him—who was, interestingly, Andrew's partner, said, “I just remember this moment where he said, ‘Man, wait to see what it feels like to make money with your friends.’”

Ten years later, I'm sitting on the stage. Andrew's like a Zen Buddha—he even looks like a Zen Buddha—but he's calm in the face of the storm. I don't know if he's helped me make money; he's definitely prevented me from losing money with his balance and kind of voice of reason.

Speaking of looking young, Eduardo had no gray hair when he launched his firm, even 5 years ago, pretty much. There's a lot of silver there now, but he's the best short seller I know. Any time I want to look at something on the long side, I pose it to him to get a counter to it.

Mason

Dave had dreadlocks.

John Lennon

Yeah, yeah, yeah. But Dave is on my Mount Rushmore of living investors. He's done it in a lot of different ways over the course of his career, but when we launched a multi-tiered business model, he was the person who gave me the most valuable advice for being the best fiduciary I could be for my LPs.

I think you can be maniacal and competitive but also have wonderful virtuous-cycle relationships. I wish I knew that earlier in my career.

Mason

That's beautiful. All right, and on that, thank you, guys. Thank you.