Bootstrapping a Business to $5 Billion in Free Cash Flow | AppLovin’s Adam Foroughi

以下是内容的中文翻译： AppLovin首席执行官亚当在演讲中分享了他的公司从上市（IPO）到股价暴跌、巨额股票回购，以及他们独特的经营理念的历程。 **上市与股价暴跌（2021-2022年）：** * AppLovin于2021年4月上市，估值达到280亿美元，并曾短暂触及400亿美元。 * 随后，在2022年全年股价暴跌，从每股115美元跌至9美元，市值最低达到38亿美元——跌幅高达92%。 * 尽管业务表现强劲，股价仍出现暴跌：2021年EBITDA（税息折旧及摊销前利润）为7亿美元，2022年增长至超过10亿美元。 * **暴跌原因：** * 广告和游戏领域是一个“艰难的领域”，公开市场难以理解。 * 疫情期间IPO过多；大型机构投资者（富达、贝莱德）无法区分，导致“缺乏支持的股权结构”。 * 私募投资者最初大量抛售股票，导致股价崩盘，看起来“一蹶不振”。 **战略性股票回购（2022-2023年）：** * 面对市场情绪，亚当的回应是：“如果没人买我们的股票，那我们为什么不自己买呢？” * 在股价触底时，公司市值38亿美元，年EBITDA超过10亿美元，这意味着他们每年产生相当于其市值五分之一的现金流。 * 他们启动了一次“非常成功的股票回购”，目标是那些持有约50%股份且很可能出售的私募股权投资者和前创始人。 * 在18个月内，AppLovin部署了约**60亿美元**用于回购，资金来源包括公司自有资金和杠杆。 * 亚当认为这次回购最终产生了“大约500亿至600亿美元的实际收益”，使其成为“公司史上最成功的股票回购之一”。 * 他为借钱进行回购的决定辩护说：“我从不相信要为‘不时之需’而存钱。我坚信我们正在建设的东西……如果我相信未来，而且我们是一个现金流非常充裕的企业，我们就应该一直回购我们的股票。”由于5倍现金流的吸引力，董事会也表示支持。 **创始故事与早期挑战（2012年以前至2015年）：** * **个人背景与动机：** 亚当四岁时，他的家庭从伊朗被迫迁徙；他的父亲失去了一项成功的房地产业务。这给他灌输了一种成功的责任感和“不服输”的精神。他也在员工身上寻找这种动力。 * **过往创业：** 在AppLovin之前，他曾拥有两家成功的广告公司（基于桌面），并于2010年之前将其出售。 * **AppLovin的创立（2010-2012年）：** 最初，他看到了向移动设备转移的趋势，试图开发消费者应用（约会、时尚、应用发现）。这些尝试都失败了，但应用推荐算法（用于应用发现）却展现了潜力。 * **转向广告网络：** 意识到推荐算法的力量后，他们放弃了消费者应用，并于2012年3月推出了AppLovin作为广告网络，为其他应用构建了SDK。 * **风险投资被拒：** 2012年，亚当最初寻求100万美元以换取公司25%的股份，但遭到顶级风投的拒绝。他们认为广告和游戏是一个由谷歌/Facebook主导的“艰难的领域”。这次拒绝激发了他的动力，尤其是当风投后来资助了拥有相同理念的竞争对手时。 * **白手起家与早期增长：** 公司在最初的1.5年里是白手起家的。从上线到8个月内，每月营收运行率迅速增长到100万美元，并实现盈利。 * **收购要约（2015年）：** AppLovin的EBITDA增长到5000多万美元。亚当拒绝了6亿美元的现金收购要约（例如来自Snapchat），他认为公司价值更高（接近10亿美元）。他认为，当时没有董事会使他能够在没有外部压力的情况下做出这一决定。 **资本市场“失误”与KKR投资（2016-2018年）：** * **中国交易：** 2015年末，为了给他的团队寻求流动性，亚当接触了中国投资者。他宣布东方弘泰资本将以14亿美元的估值对公司进行10亿美元的私募股权投资，意图将公司在中国上市，并让中国投资者控制70%的股份。 * **CFIUS审查：** 随着特朗普总统的上任以及中美关系的紧张，这项交易受到了CFIUS（美国外国投资委员会）的严格审查，该委员会对外国持有美国大型数据平台的国家安全担忧进行了调查。亚当，一个“波斯裔CEO”带着一个“俄罗斯裔CTO”将公司卖给一家中国公司，他意识到处境艰难。 * **解决方案：** 他将交易转变为10亿美元的可转换票据，这笔票据向股东支付股息，但只在转换后给予中国投资者10%的股权，从而避免了控制权问题。然而，这使AppLovin背负了巨额可转换债务。 * **KKR出手相助：** KKR于2018年投资以清偿这笔债务，带来了传统的美国资本，并建立了一个“清洁的股权结构”。这也促成了AppLovin第一个正式董事会的成立（亚当、Eduardo Vivas、KKR的Harold Chen）。亚当后悔没有早点设立董事会来指导他处理这些资本市场的复杂性。 **战略转型：利用游戏工作室获取数据与Axon平台演进（2018年至今）：** * **数据问题：** 为了建立一个更强大、更高效的机器学习模型（如Facebook的模型），AppLovin需要除游戏类型之外更多样化的广告主数据。然而，广告主不愿直接分享他们的数据。 * **垂直整合：** AppLovin决定收购游戏工作室（最终收购了14-15家），既是为了了解开发者需求，也是为了将他们有价值的游戏内购买数据输入到他们的广告模型中。 * **客户不信任：** 这一举动最初引起了游戏客户的不信任，他们担心AppLovin会利用他们的数据与他们竞争。亚当本人“讨厌游戏”，只关心数据，他进行了无数次透明的对话以重建信任。 * **Axon 1和Axon 2：** * **Axon 1（传统机器学习）：** 第一个机器学习模型，相对于基于规则的模型有了显著升级，提高了营收。 * **Axon 2（深度学习）：** 大约在2023年4月推出，这个深度学习模型显著提高了预测价值。它使广告主能够实现“套利”，即他们投入1000美元，并确定能赚回超过1000美元，从而显著扩大了他们的业务规模。 * **剥离游戏工作室：** 一旦Axon模型变得强大并被广泛采用，AppLovin便出售了游戏工作室（例如出售给Tripledot）。亚当意识到这些工作室“令人头疼”且“分散注意力”，它们雇佣了1500多名员工，而核心广告业务只有400人。 * **Axon 2发布后的增长：** 股价从2023年4月的9-11美元飙升至最高750美元，市值达到2500亿美元。 **团队、文化与AI驱动的效率：** * **精干的“A级人才”文化：** AppLovin运营团队规模惊人地小（350-400名员工却创造了数十亿美元的现金流）。 * **崩盘后的人才策略：** 在2022年股价暴跌期间，亚当调整了薪酬策略： * 工资较低的员工如果需要，则转为现金薪酬。 * 只有关键的工程师和产品人员（约100人）获得了股权，认可了他们长期的重要性。 * **追求极致的效率：** 亚当认为“A级人才不喜欢与B、C、D级人才共事”。在2023年，尽管业务增长了100%，他仍然解雇了大约40%的团队，以便“将其精炼成这些A级人才”。 * **乔瓦尼（CTO）的影响：** 新任CTO乔瓦尼（2022年11月加入）质疑了现有的流程和岗位，进一步推动了精简，并为适应AI时代对公司文化进行了重新评估。 * **不“补缺”政策：** 亚当亲自批准所有新员工的招聘，每年招聘数量从数百人减少到仅仅几十人，确保每个新岗位都真正必不可少。 * **精简的C级高管团队：** C级高管团队仅包括CEO、CFO、CTO和总法律顾问——没有首席营收官（CRO）或首席运营官（COO），这对于如此规模的广告公司来说实属罕见。 * **AI集成：** 超过80%的AppLovin代码在一年前就是由大型语言模型（LLM）编写的，现在这个比例甚至更高。AI工具使“10倍工程师”的效率提高了“100倍”，加速了研究和模型改进。 **未来愿景：** * 亚当设想AppLovin能实现“万亿美元的成果”，关键人员的薪酬计划与这一宏伟目标挂钩。 * 他认为AI带来了巨大的顺风，使规模更小、能力更强的团队能够取得超乎寻常的成果。 * **超越游戏的扩张：** 平台正在从移动游戏扩展到电子商务，最终目标是瞄准*任何*业务（从小型Shopify商店到“当地洗衣店”）。 * 他相信他们庞大的受众（超过10亿观看35秒广告的成年手游玩家）与不断改进的AI驱动定向技术相结合，将使广告体验更像“内容”，为消费者和广告主带来更大的价值。

The speaker, Adam, the CEO of AppLovin, shares his company's journey from IPO to stock collapse, a massive stock buyback, and their unique business philosophy. **The IPO and Stock Collapse (2021-2022):** * AppLovin went public in April 2021 at a $28 billion valuation, briefly hitting $40 billion. * The stock then plummeted throughout 2022, dropping from $115 to $9 a share, reaching a market cap floor of $3.8 billion – a 92% decrease. * This happened despite strong business performance: $700 million EBITDA in 2021, growing to over $1 billion EBITDA in 2022. * **Reasons for the collapse:** * The "tough space" of advertising and gaming, which public markets find hard to understand. * Too many IPOs during COVID; large institutional investors (Fidelity, BlackRock) couldn't differentiate, leading to a "cap table that didn't have support." * An initial flood of shares from private company investors selling, causing the stock to crumble and appear "broken." **The Strategic Stock Buyback (2022-2023):** * Adam's response to the market sentiment: "If no one's going to buy our shares, why don't we just start buying our own shares?" * At the bottom, the company had a $3.8 billion market cap and over $1 billion in annual EBITDA, meaning they were generating 1/5th of their market cap in cash flow annually. * They kicked off a "really successful buyback" by targeting private equity investors and former founders who owned about 50% of the shares and were likely sellers. * Over 18 months, AppLovin deployed approximately **$6 billion** in buybacks, using company capital and leverage. * Adam believes this buyback ultimately generated "somewhere in the neighborhood of $50-60 billion of actual proceeds," making it "one of the more successful buybacks in the history of companies." * He defends the decision to borrow money for buybacks: "I never believed in saving cash for a rainy day. I feel like I'm a big believer in what we're building... if I believe in the future and we're a really high cash-generated business, we should always be buying back our shares." The board was supportive due to the attractive 5x cash flow multiple. **Founding Story & Early Challenges (Pre-2012 to 2015):** * **Personal Background & Motivation:** Adam's family was uprooted from Iran when he was four; his father lost a successful real estate business. This instilled an obligation to succeed and a "chip on his shoulder." He looks for this drive in his employees. * **Previous Ventures:** He had two successful advertising businesses (desktop-based) before AppLovin, selling them by 2010. * **AppLovin's Inception (2010-2012):** Initially sought to build consumer apps (dating, fashion, app discovery) seeing the shift to mobile. These failed, but the app recommendation algorithm (for app discovery) showed potential. * **Pivot to Ad Network:** Realizing the power of the recommendation algorithm, they scrapped the consumer app and launched AppLovin as an ad network in March 2012, building an SDK for other apps. * **VC Rejection:** Adam initially sought $1 million for 25% of the company in 2012 but was rejected by top VCs. They saw advertising and gaming as a "tough pitch" dominated by Google/Facebook. This rejection fueled his motivation, especially when VCs later funded competitors with the same idea. * **Bootstrapping & Early Growth:** The company was bootstrapped for the first 1.5 years. It quickly grew from launch to a $1 million monthly run rate in 8 months, becoming profitable. * **Acquisition Offer (2015):** AppLovin grew to over $50 million EBITDA. Adam walked away from a $600 million cash acquisition offer (e.g., from Snapchat), believing the company was worth more (closer to $1 billion). He credits not having a board at the time for allowing him to make this decision without external pressure. **Capital Market "Mistakes" & KKR Investment (2016-2018):** * **The Chinese Deal:** In late 2015, seeking liquidity for his team, Adam pursued Chinese investors. He announced a $1 billion private equity investment at a $1.4 billion valuation from Orient Hontai Capital, intending to take the company public in China and give Chinese investors 70% control. * **CFIUS Scrutiny:** With the election of President Trump and rising US-China tensions, the deal faced intense scrutiny from CFIUS (Committee on Foreign Investment in the United States), which investigated national security concerns over foreign ownership of a large US data platform. Adam, a "Persian CEO" with a "Russian CTO" selling to a Chinese company, realized the difficulty of the situation. * **Resolution:** He pivoted the deal to a $1 billion convertible note, which paid a dividend to shareholders but gave the Chinese investors only 10% equity upon conversion, avoiding control issues. This, however, left AppLovin with massive convertible debt. * **KKR to the Rescue:** KKR invested in 2018 to clear this debt, bringing in traditional US capital and creating a "clean cap table." This also led to the formation of AppLovin's first formal board (Adam, Eduardo Vivas, Harold Chen from KKR). Adam regrets not having a board earlier to guide him through these capital market complexities. **Strategic Shift: Gaming Studios for Data & Axon Platform Evolution (2018-Present):** * **The Data Problem:** To build a more robust, powerful machine learning model (like Facebook's), AppLovin needed diverse advertiser data beyond just game types. Advertisers, however, wouldn't share their data directly. * **Vertical Integration:** AppLovin decided to acquire gaming studios (eventually 14-15 of them) to both understand developer needs and feed their valuable in-game purchase data into their advertising model. * **Client Distrust:** This move initially caused distrust among gaming clients who feared AppLovin would compete with them using their data. Adam, who "hates games" himself and only cared about the data, held numerous transparent conversations to rebuild trust. * **Axon 1 & Axon 2:** * **Axon 1 (Traditional ML):** The first ML model, a significant upgrade from rules-based, boosted revenue. * **Axon 2 (Deep Learning):** Rolled out around April 2023, this deep learning model dramatically improved predictive value. It enabled advertisers to achieve "arbitrage," where they could spend $1,000 and be certain to make back more than $1,000, allowing them to scale their businesses significantly. * **Divesting Gaming Studios:** Once the Axon model was robust and widely adopted, AppLovin sold off the gaming studios (e.g., to Tripledot). Adam realized they were a "headache" and a "distraction" from their core advertising platform, employing 1,500+ people compared to the 400 in the core ad business. * **Post-Axon 2 Growth:** The stock soared from $9-11 in April 2023 to a peak of $750, with the market cap reaching $250 billion. **Team, Culture, and AI-Driven Efficiency:** * **Lean, A-Player Culture:** AppLovin operates with a remarkably small team (350-400 employees for a multi-billion dollar cash flow business). * **Post-Collapse Talent Strategy:** During the 2022 stock crash, Adam refined his compensation strategy: * Lower-paid employees were shifted to cash compensation if needed. * Only critical engineers and product people (around 100 individuals) received equity, recognizing their long-term importance. * **Ruthless Efficiency:** Adam believes "A players don't like to work with B, C's, D's." In 2023, while the business grew 100%, he fired approximately 40% of the team to "distill it down to these A players." * **Giovanni's Impact (CTO):** The new CTO, Giovanni (joined Nov 2022), questioned existing processes and roles, leading to further streamlining and a re-evaluation of the company's culture for the age of AI. * **No "Backfill" Policy:** Adam personally approves all new hires, moving from hundreds of annual hires to mere tens, ensuring every new role is truly essential. * **Lean C-Suite:** The C-suite consists only of the CEO, CFO, CTO, and General Counsel – no CRO or COO, a rarity for an advertising company of its scale. * **AI Integration:** Over 80% of AppLovin's code was LLM-written over a year ago, now even higher. AI tools allow "10x engineers" to become "100x more efficient," accelerating research and model improvements. **Future Vision:** * Adam envisions a "trillion-dollar outcome" for AppLovin, with compensation plans for key personnel tied to this ambitious goal. * He sees massive tailwinds from AI, enabling smaller, highly competent teams to achieve disproportionate results. * **Expansion Beyond Gaming:** The platform is expanding beyond mobile gaming to e-commerce, eventually targeting *any* business (from small shopify stores to "the local laundromat"). * He believes their strong audience (1 billion+ adult mobile gamers watching 35-second ads) combined with improving AI-driven targeting will make the ad experience more like "content," increasing value for both consumers and advertisers.