Applovin CEO: The "Darkest Hour" of a 92% Stock Price Crash and "Self-Salvation"

Today, AppLovin is one of the highest-valued technology companies in the United States. It is expected to generate around $6 billion in cash this year, and annual ad spend on its ad platform has reached $20 billion. However, few people know that less than three years ago, the company's stock price had plunged 92% from its peak.

Recently, at the 2026 All-In Summit, AppLovin's founder and CEO Adam Foroughi systematically shared this chapter for the first time—from a collapsed IPO to an AI transformation, and ultimately rewriting their fate through buybacks.

Foroughi frankly said during the conversation: "Advertising is ML 1.0; it is the first commercial implementation of all the technologies now powering AI." AppLovin, following this very technology curve, made the leap from mobile game advertising to generating e-commerce shopping intent.

An "Advertising Empire Hidden in 100 Million Games"

AppLovin is not exactly a household name. Foroughi himself admits, "That silly name certainly didn't help much."

But its scale shouldn't be underestimated.

Foroughi explained during the conversation that AppLovin is essentially a mobile game advertising company—helping game developers monetize their traffic. He gave some figures:

• Over 1 billion people worldwide play mobile casual games every day

• At the start of 2024, annual ad spend on AppLovin’s owned platforms reached $11 billion. Since then, it has grown at an annual rate of about 60%, so today’s figure is roughly $20 billion

• He estimates the entire mobile game advertising ecosystem’s market size is around $50 billion

"Not long ago, the social media advertising market was just a $50 billion opportunity," Foroughi said. "This space is growing rapidly."

More importantly, AppLovin’s EBITDA profit margin now sits at 84%, ranking first among similar companies.

Darkest Moment: Profiting $1 Billion a Year, Market Cap Down to $3.8 Billion

The company’s current state stands in sharp contrast to where it was three years ago.

In April 2021, AppLovin went public at a valuation of around $28 billion and once surged as high as $40 billion.

Then it collapsed.

"In 2022, the stock price seemed to drop every single day," Foroughi recalls. "We eventually fell to a $3.8 billion market cap."

That same year, the company's EBITDA was $1 billion. The valuation multiple tumbled from IPO highs to less than 4x EBITDA.

As for the reason for the collapse, he is direct: "Your pricing in the market depends on the quality of your investors."

During the 2021 IPO boom, blue-chip investors didn’t have time to study a strangely-named company. Early private equity shareholders and employees were anxious to cash out, supply outstripped demand, and the stock price plummeted.

During that period, Foroughi got calls from family and friends. "They asked me if I was okay," he said. "I told them: We started with nothing, the stock is still at $10, so we’ve done well. But as a CEO, you quickly realize your team is getting the same calls—and they don’t have your confidence, or your equity stake."

Self-Rescue: Stopping Roadshows, Buying Our Own Stock

Facing abandonment by the market, Foroughi made an unconventional decision.

"I told the team, I don’t plan to meet with investors anymore. They’re not buying our stock, so meeting them is a waste of time. We generate a lot of cash—so let’s buy back our own stock and become our own best investor."

AppLovin then launched a massive share buyback program.

The results: About $6 billion in share repurchases, retiring about 20%-25% of outstanding shares. At the peak, those shares were worth over $50 billion.

At the same time, the company launched a performance stock program internally, covering key employees instead of just the CEO. Foroughi’s logic: "We know it’s a tough time, you thought you had a house, and now it’s gone. But if you stick it out and we rebound, you’ll gain a lot."

What backed his judgment was a key technology upgrade.

In April 2023, AppLovin upgraded its ad algorithms from regression models (ML 1.0) to deep learning models (ML 2.0). The effects were immediate—advertisers' ROI improved significantly, and the platform’s growth exploded.

But the company still refrained from making public statements.

It wasn’t until September 2023 that Foroughi went to New York to meet investors for the first time. By then, the stock had rebounded to about $80.

"That week, the stock price rose from $80 to $150. The market cap jumped from $28 billion to $55 billion."

"Just because I went to New York and told everyone: Our company is alive, we made it back."

Advertising as "ML 1.0": The Technical Logic

Foroughi has a clear framework for AppLovin’s technical positioning.

Advertising is ML 1.0. It’s the first commercial implementation of all these technologies, which now drive AI.

He distinguishes between two types of advertising logic:

Search Advertising (Bottom of the Funnel): Users already know what they want to buy and use Google to complete the transaction. For this need, large language models are directly replacing Google.

Discovery Advertising (Top of the Funnel): Users don’t know what they want, and advertising creates demand. "You show a recommendation, users say 'Wow, that looks great, I’ll go buy it!'—this is what drives Meta’s advertising business, and it’s what we want to do."

He believes discovery advertising creates true economic growth, while search ads merely accelerate a transaction that would have happened anyway.

On whether AI Agents will disrupt advertising, he is cautious:

Some people will use agents to optimize fixed consumption, like auto-renewing supplements every month. But discovery platforms don’t follow this logic. Ordinary shoppers aren’t the techies hanging out on Twitter chasing the latest innovation.

Our audience is more like New York Times readers. Today, a lot of people still use Yahoo products every day.

Why Can a Small Company Beat Meta and Google?

AppLovin’s EBITDA profit margin has held at 84% for years. Facing Meta and Google, the ad giants with the best engineers, why hasn’t it been crushed?

Foroughi’s answer is simple:

We never think we’ve won. Every morning we wake up, we feel today could be the day we fail, so we work like crazy.

His competitive logic: Focus, streamline, iterate quickly.

If you’re highly focused and lean, you can move faster than the giants.

On doubts about whether high margins will be squeezed by competition, he explains:

These technologies are extremely complex. If you keep innovating and have differentiated data, you build a moat. It’s like Anthropic—it “shouldn’t” be able to run so fast in LLMs, but when a model gains scale effects and widespread adoption, it builds a high barrier others can’t easily cross.

He also mentions, AppLovin has an engineering team in Beijing, China. "The Chinese are very hardworking, very smart. Working with them is one of the company’s strengths. When I’m in a meeting room, I often feel like the dumbest person in the room—which actually makes me excited to come to work every day."

Full Interview:

Adam Foroughi, AppLovin CEO: Surviving a 92% Stock Crash, Advertising as ML 1.0, and the $50 Billion Game Advertising Market

September 21, 2026 · All-In Summit 2026

Chapter One: Adam Foroughi Joins the Besties!

Host: Adam is arguably the most outstanding yet little-known founder. There’s an ad platform hidden in over a hundred thousand mobile games, quietly surpassing Facebook ads and becoming the top choice for e-commerce brands. Out of thousands of IPO companies, the highest market cap belongs to AppLovin. Founders’ mindsets must relentlessly pursue victory. This year alone, they’ll likely generate about $6 billion in cash flow. In an irrational world, people assume you’re cheating, unaware that you’ve built one of the world’s coolest technologies. Please welcome Adam Foroughi!

Host: Adam, thank you for joining us. We think chatting with you will be fascinating, because you rarely appear in media headlines and your company runs practically off the media grid—few interviews, little public discourse about the company. But this is an outstanding enterprise. Can you tell us what AppLovin does, and sketch the market landscape for us?

Adam Foroughi: Of course. I think we became such a large company without early VC investment precisely because we kept our heads down and operated quietly. And yes, the quirky company name didn’t help either.

We are essentially an advertising company, specializing in helping mobile game developers monetize in this sector. Many people don’t realize just how massive the mobile gaming world has become—over a billion people play mobile casual games daily; they’re adults, household decision-makers, and the market opportunity is enormous.

A little over two years ago, in the January before last, we disclosed $11 billion in yearly ad spend just on our own platform. Since then, we’ve grown at about 60% year-over-year, so that number is approaching $20 billion today.

We aren’t the only player in the market; many other advertising companies monetize in the space. When you add it all up, the entire mobile game system attracts around $50 billion in ad spend annually.

Not long ago, the social media ad market was likewise around $50 billion. This field is growing rapidly—and the audience willingly watches ads, often opting in for incentives, creating the possibility for “intent generation.”

For most of our company’s history, we’ve created intent in this way—guiding users naturally from one game experience to the next. What excites investors, and ourselves, is this: deep learning models are now powerful enough that we can use this same channel to guide adult users into commerce contexts. This allows us to enter a larger economic field and create greater value, which is why the team is so passionate about what we’re doing.

Chapter Two: Discovery Ads vs Search Ads, and Is Your Phone Really Listening?

Host: The first wave of Internet advertising lit the spark for countless key technologies that spread worldwide, such as Google’s AdWords, AdSense, application semantics, etc. Is a similar wave underway now? Are there foundational technologies being created here that could shape the internet’s future?

Adam Foroughi: Yes. Advertising is ML 1.0—it’s the initial implementation of all these technologies now powering AI. LLMs today generate much more economic value than ads, but advertising remains an incredibly profitable deep learning application.

Recommendation systems are structurally different from LLMs, but in many ways follow the same development path. Many advances in LLM research can be migrated to recommendation systems, and vice versa. Lots of today’s LLM researchers started their careers studying ad systems. The two fields are deeply interconnected.

The benefit of the ad business, and any ad business, is that when you build a model, you’re predicting a future outcome–maybe an ad, a social network post, or a content series—and you can instantly monetize that prediction.

Host: Is there a universal behavioral pattern in how humans react to ads in 2026 as compared to 2006? Can we forecast how this evolves?

Adam Foroughi: That’s a fun question. I got into the industry in 2005, and advertising then was a mess—full of spam, with crude tech. Later, Facebook did something smart—they realized that combining all their data with strong tech could make advertising ultra-precise. Now, if you ask most shoppers, the majority of their shopping inspiration comes from Instagram. Advertising has become much like the content itself.

It’s the same in our field—people enjoy our ads. You might not think so, but with mini games, we see high engagement—these "ads" show up in other games, and people actively try these previews because the tech is strong enough to surface genuinely relevant content.

Host: There’s concern about AI’s impact on ad networks, especially for Google. OpenAI now has its own ad product, and you’ve been monitoring and learning from this. When people start using chatbots for multiple rounds of queries, what happens to ads? Obviously, 95% of users won’t pay $20/month—they’ll expect free use. ChatGPT is going free. What’s their plan in advertising, and will this lead to lower per-ad effectiveness but higher total use, or will it be more effective than Google’s ad model?

Adam Foroughi: There are two dimensions to advertising. One is bottom-of-funnel advertising—consumers mostly know what they want to buy, they’re researching, completing the transaction. That’s Google’s search business. For example, if I want leather shoes, I used to search on Google, do some research, and get ads that direct me. Now you can complete that loop using an LLM. So this ad model directly competes with Google search.

What we do is show ads to users who don’t have explicit intent. We generate something from scratch—showing them a recommendation so they think: "Wow, this looks great, I want to buy one." This is also what drives Facebook’s ad business.

Transactions from search or LLM are usually inevitable—with or without Google ads, as long as search exists, the transaction will happen, just by a different path. So there’s limited opportunity for extra economic value there.

But when you show a user an ad for a product they didn’t know existed or needed—that’s pure discovery, a purely new experience. That’s why Meta’s ad business is so strong, and it’s our goal as well. You create that “discovery moment,” which is pleasant for the consumer—they’re anticipating their purchase, eager for the delivery—and you truly generate incremental economic value.

Host: What about the “arms race”? Sometimes at dinner, you mention something, then see ads for it on Meta or elsewhere. Are we overreacting, or does this really happen? Is advertising becoming increasingly invasive as companies try to sell more?

Host: Right, when this happens, it’s uncanny. What actually goes on—did I really say something out loud and see ads for it?

Adam Foroughi: I think you probably did some other trackable behavior, like a search or site visit, which you forgot, then talked about it and started seeing ads. It’s not that your mic is on or some app is eavesdropping.

Host: But some theorize that, for example, if we have dinner together, the apps know our locations and link us as a group. Say we all discuss buying a car or a watch, and then Friedberg searches for the watch, so the system shows ads for it to all four of us—is that the logic?

Adam Foroughi: I don’t think ad companies track location. We don’t track location at all. Tracking precise location for ad targeting is very heavy. Moreover, if an app really listened to your mic, parsed it, and converted that to ad directives, the data volume would be unrealistically large.

Host: What if we’re friends and have social ties?

Adam Foroughi: We don’t have that data. But on social networks, your relationships might certainly impact ads. If your friend searches for something, you may see related stuff. There’s nothing wrong with that.

There is indeed a "creepy factor" here, but with so many companies buying ads at scale, most collected data is already regulated.

People often overlook that, it’s because ads are so accurate today that huge economic value is created. You see an ad and can relate to it—20 years ago, you didn’t care about the ad at all. Today, a significant part of GDP comes from this digital ad economy. The more advanced the tech, the faster GDP grows.

Chapter Three: IPO Crash and Becoming Our Own Best Investor

Host: Adam, your company’s operations are fascinating. Are you based in Los Angeles now?

Adam Foroughi: I’m in Los Angeles. Originally, the company was founded in Silicon Valley, in Palo Alto.

Host: Palo Alto. And you have many developers in China too, right?

Adam Foroughi: Our engineering teams are in Palo Alto, Beijing, and Singapore.

Host: You didn’t raise much venture capital. The company went public in 2021, valued at about $20 billion?

Adam Foroughi: We listed via a joint IPO in April 2021, at about $28 billion.

Host: $28 billion. By 2023, what did the market cap hit?

Adam Foroughi: One interesting thing about public markets: At IPO in 2021, we had $600 million EBITDA, a $28 billion market cap (peaked at $40 billion). By 2022, the stock tumbled almost daily, falling to about $3.8 billion by the end. That year, we did $1 billion in EBITDA.

Host: That’s incredible. Let’s clarify—so for some reason, the market didn’t believe in your business. What did you do?

Adam Foroughi: You quickly realize (I have a finance background, so I understood): Your share price depends on the quality of your shareholders. Our early investors included private equity, cofounders, and other team members who sold after IPO. At the time, tons of companies were going public, and blue-chip investors had no time to research companies with weird names, so demand was low and supply was high, which tanked the price. Our valuation multiple slid from a relatively high level—I wouldn’t value the company at 50x EBITDA!—to a ridiculous low under 4x.

So from a finance perspective, when the crash happens, you must see the opportunity. I told our internal team: I’m done meeting investors; they aren’t buying our stock, so it’s a waste of time. We generate loads of cash flow, so let’s use it to buy our own shares—be our own best investor. So we launched a very aggressive buyback plan. Since then, we’ve repurchased about $6 billion worth of stock, retiring 20–25% of outstanding shares. At the peak, those $6 billion repurchases were worth over $50 billion.

So what looked like a terribly depressing moment can totally transform into a gigantic opportunity.

Host: Did you always see it that way, or did you have periods where you felt really low, like, “What is going on?”

Host: Exactly, how did you maintain the company’s internal culture when the stock dropped 92%?

Adam Foroughi: It was tough. Family and friends would call and ask, "Are you okay? Are you thinking about quitting?" I’d say, you know, we started with a penny, and the stock was still at $10, so it had gone up a lot in absolute terms. But as CEO, you quickly realize a tough truth: your team is getting the same calls, but they don’t have your conviction or your equity stake.

So our strategy was to create an “us versus the world” mentality—everyone’s given up on us, but we’re going to keep buying back stock. At the same time, we launched a performance stock plan. Normally that only goes to CEOs, but we gave it to core staff and said: we know it’s tough, you thought you had a house, and now it feels like you have nothing. But if you stick it out and we rebound, you’ll earn handsomely on the way up.

Then, as investors started to refocus on us, that was when we upgraded from ML 1.0 to ML 2.0—from regression models to deep learning. The results were massive. Our ad algorithms drive our business—the better the algorithm, the higher advertisers’ ROI, and everything is performance-based. So the company began to grow rapidly.

In April 2023 we launched the new model but hadn’t told the outside world, so no one knew yet. Around September 2023, I went to New York, by then the stock had recovered to about $80, getting a performance-driven bounce. I thought, time to talk to investors, since the market cap was now high enough that buybacks couldn’t stay as aggressive.

That week, the stock went from $80 to $150, and market cap doubled from about $28 billion to $55 billion—just because I went to New York.

Host: Just because you went to New York and said, "Hey, we’re alive, we made it through"?

Adam Foroughi: Exactly. I’d sit in meetings and it was obvious—you could see people texting outside, “Buy buy buy buy!” It felt amazing.

Host: Flip side—once they’re long, do they start asking: "Adam, how do we expand? Why just games? Why not e-commerce? Why not X, Y, or Z?"

Adam Foroughi: Totally. It’s a double-edged sword; you’re never right.

Most people aren’t contrarian investors—the drop is as wild as the rise. In two and a half years our stock went from $9 to $750 a share, market cap from $3.8 billion to $250 billion. Extremes on both ends.

Now we’re back at a relatively stable level, very bullish on growth. But I’ve found that public and private market investors aren’t that different—they all chase trends, they just move a bit slower than you’d hope. The best investors spot trends ahead of the market. That’s why there’s a gap between top VCs and the rest, and between top and average public investors.

Chapter Four: Privacy Regulation, Apple’s Controls, and How AI Agents Change Shopping Behavior

Host: Can you talk about privacy? Apple and the EU are scrutinizing companies like yours for aggressive data collection. Some game devs don’t want their user data shared, so things are tightening up. Zuckerberg faced this issue head-on back in the day. So what obstacles does this bring to your business? When Apple is—let’s be blunt—trying to challenge you, how do you handle privacy?

Adam Foroughi: In any such field, you want clear regulation. Once rules are defined, tech can adapt. Five years ago, you could target an iOS user precisely; today, if a user opts out, you put them in a cohort and serve less accurate ads.

Interestingly, after Apple made this change, many users complained: "Show me more relevant ads, these are junk!" Privacy regulation is needed—it lets tech companies do things right—but consumers need relevant ads to discover products too.

If you’re playing a game and watch a 30-second ad for an extra life, you’re getting real value. Would you rather see a 30-second junk ad, or 30 seconds of relevant content? Since new privacy rules rolled out, tech companies have adapted, and deep learning systems remain extremely powerful.

Host: Quick follow-up on game studio acquisitions. You tried to buy some studios; we’ve got Bending Spoons here—acquiring solid but slow-growth, VC-overlooked businesses. For you, is becoming a game studio sustainable? Would this create conflict?

Adam Foroughi: We’ve since divested those games. We acquired studios for data reasons—when building our first deep learning model, we needed training data, and studios rarely share it with third parties. So we bought studios to train our first models; once third-party partners came on, we spun the studios back out.

Host: In a world full of AI agents, what form will ads take? If interfaces move beyond keyboards or screens to Meta Glasses or something else, what’s the role of advertising in agent commerce? So many are building for this future.

Adam Foroughi: In reality, a part of the world will use agents to optimize routine shopping—like having an agent handle your monthly supplement orders. But discovery platforms aren’t like this, and the average consumer isn’t a power agent user lurking on Twitter for the latest tech.

Our audience is more like New York Times readers—many people still use Yahoo products daily. Regular shoppers want to find, compare, track delivery, and enjoy the process themselves.

If, after the fact, you say, "Hey, an agent could have done this and saved you 20%," for a $50 transaction, it’s irrelevant—the dopamine hit from buying yourself is what they want.

So, yes, a tech-forward minority will use agents, but I think we overestimate the representativeness of the Twitter crowd and forget that most consumers just aren’t like that.

Chapter Five: How Small Teams Outsmart Giants, Margin Moats, and Building Teams in China

Host: Let me understand how you win—Meta and Alphabet/Google are the world’s smartest companies, have the best engineers, and they’ve been at ads for a decade or two. How does a small company compete in this market and win? What sustains your edge?

Adam Foroughi: One thing got us here—we never think we’ve won. Each morning we think, today could be the end, so we hustle. This makes us a lean team of true specialists, deeply focused on mobile game experiences and how to translate that to the consumer side.

I believe if you’re focused and lean, you can move faster than the giants. That’s how you challenge incumbents.

Host: Where are the “leaky spots” in the business? I used to analyze Amazon’s P&L this way—you find soft spots, and over time, Amazon absorbs them into new lines, which is why you’d long Amazon. What are your weak spots—payments infrastructure? Or, differently, what’s the margin expansion potential that gets people buying this story?

Adam Foroughi: Our EBITDA margin is, I believe, highest in the market—84%. So I’m not sure how much "leakage" we really have.

Host: At that number, it’s hard to say there’s much leaking—but another angle: advertisers come to the platform and operate on a transaction model. If they’re selling lipstick, we offer arbitrage—they buy users here, get their margin from sales, and keep investing. It’s a highly replicable model.

Our weak spot is, we’re not the whole chain. We’re not the advertiser—we want to enable them to reach consumers. We run super lean, focus on algorithms and automation, so there’s not much leakage.

Host: But when your EBIT margin is 85%, people might say "This company is over-earning," and competitors will say, "I’ll bleed Adam’s margins, I’ll do it at 60% or 50%." But—that hasn’t happened, margins stay strong. Why?

Adam Foroughi: Because these technologies are extremely complex; if you keep innovating and have differentiated data, you have genuine advantage. Similarly, Anthropic “shouldn’t” have run so fast in LLMs, but when a model reaches scale and broad adoption, it builds a moat others can’t cross.

Host: Tell us about your China team—what is their advantage?

Adam Foroughi: Chinese people are very humble, extremely hardworking and very smart. Whether you’re working with engineers in China, the U.S., or anywhere else, you’re working with some of the smartest people on earth.

When I started this company, one of my goals was to work with talented people and figure things out together. Now, when I sit among my team, I know I’m probably the dumbest person in the room—and that motivates me every day to come to work.

Host: Honestly, how does that make you feel?

Adam Foroughi: It works for me.

Host: Let’s give it up for Adam! Adam, thank you.

Adam Foroughi: Thanks, guys, great to meet you.