AppLovin撕开巨头裂缝的1000天：AI审判、被做空与Underdog的“弱者之心”【硅谷101专访】

Meta and Google as the global advertising monopoly position was torn apart by a crack This crack made a company rise rapidly, the stock price increased 25 times in two years, added a standard 500 index to the 500 billion dollar club He is Apple loving. However, the controversy and the fake reports surrounding the company have never stopped. Got some fresh hate from Muddy Waters after this epic run.

never stopped. Got some fresh hate from Muddy Waters after this epic run.

I think this is at least the third that we've seen in the past month or so. These short sellers are questioning the integrity of their AI platform.

or so. These short sellers are questioning the integrity of their AI platform.

-

Good talk!

From 2013 to 2015, Ge Xiaochuan's promotion rate is similar to Apple's stock price. In three years, he quickly advanced from an engineer to a core technology

price. In three years, he quickly advanced from an engineer to a core technology leader of the company. Now he is the chief product and engineer of the company.

Although there are many rumors and interpretations about Apple Loving in the outside world, the company's high-level deep conversations with the outside world are still limited. Many information and conclusions are conflicted. When the job control report came out, we were very careful. We

went to our system to compare each one. Then we found that it was all wrong. So the point I'm interested in this company is that I'm curious about what this company did right behind the huge number of giants like Meta and Google, and what time it was right to tear apart the market crack of giants. We don't have as much resources as Google and

Meta. We can't use their methods to solve our problems. From the moment I

Meta. We can't use their methods to solve our problems. From the moment I entered Apple Loving, I just hoped that one day we could get out of the game. What do people think of Apple Loving's internal pursuit and questioning of it?

game. What do people think of Apple Loving's internal pursuit and questioning of it?

Ge Xiaochuan is also how to lead the team to play modern to recommend the algorithm this time to set up a challenge is the talent market for our recognition of the rich we basically very difficult to find people very difficult to find people with such curiosity we Gui 5101 is a unique interview Ge Xiaochuan this is also his first time to interview foreign media video in

Chinese for the past three years really feel like ten years after every day in the war Gui Gu this place I think on the dog here is the most appreciated a spirit We interviewed Ge Xiaochuan on December 21, 2025. The interview was recorded for more than two hours. But to be honest, our conversations were a bit repetitive. The reason was that I desperately wanted to find the answer. What did

repetitive. The reason was that I desperately wanted to find the answer. What did

Apploving do to get us up? But I found that this was not the way Ge Xiaochuan and Apploving looked at the problem. I found that when I was chatting with you, I found that you wanted to ask whether our success was caused by a specific decision. But it's not like that. The decision itself is very simple. The real success of the company is the success of 365

days after the decision. How does the day pass? The outside world always wants to find an explanation to make our success look simple. I think this kind of doubt and underestimate has been long-term. In the first quarter of 2026, Apple Loving continued to experience attacks from the air-conditioning institutions, the challenges of the advertising industry and the market cake struggle of competitors. After the US stock was sold

on February 11, Apple Loving released the fourth quarter of 2025. The profit and expectation of the revenue are beyond the expectations of the previous analysts. It

can be said that it is a very good financial report. However, the company's stock price has fallen sharply. This backslash shows to a certain extent the fear of the capital market to subvert the SaaS business by AI. At this time, we are online for this interview. We try to see through Ge Xiaochuan's deep sharing.

In his point of view, the past three years, about a thousand days, how he led uploving, tear apart, head-to-head, First, before we officially enter the interview of Ge Xiaochuan, let's take a little time to talk about the business and story of this company.

In 2018, Google's digital advertising rights have been simplified. Google has a search intent, and Meta controls social patterns. Uploading is just a small-scale mobile advertising network, Ad Networks. Simply put, it is to connect mobile app developers, that is, advertisers and advertisers' intermediary platforms. The biggest problem for this intermediary is that it doesn't

know what users are doing in the app. Think about it, if you can't How do you fight with a giant like Google and Meta? In order

to break this, Apple started to buy large-scale gaming studios. From the early 2021 IPO of 2018, Apple has expanded its game version with a series of strategic investment and studio optimization actions, and the total trading volume has reached the $1 billion level. and rely on its advertising and distribution platforms to reach millions

of mobile device users around the world. At that time, Apple's purpose of buying these studios was not to profit from the game, but to obtain the core of the transformation signal. These self-produced games became the first laboratory for computing engines. At the

transformation signal. These self-produced games became the first laboratory for computing engines. At the

same time, Apple has made a series of key acquisitions. First of all, the app-only technology company Max, which was acquired in 2018.

In 2018, Uploving was more like an advertising platform. But for developers, it was not a good idea to sell ads at a high price. At

that time, the most used in the industry was a method called "publiu". In

simple terms, developers first tried to sell ads to A advertising platform. If A

did not accept, they would sell ads to B. If not, they would sell ads to C. one by one. The order is pre-arranged based on the past ECPM, which is the effective display cost per thousand times, and the fill rate, etc. These historical data. But the problem is that the whole process is in line-up, the efficiency is low, and it is easy to sell at

a low price. The rules changed completely after Max came out. It introduced the actual price in the application. Once the advertiser is given to all advertisers, let everyone bid at the same time. Finally, whoever is able to bid high will sell to whom. The developer went straight to the highest price, and the income immediately

whom. The developer went straight to the highest price, and the income immediately came up. The advertiser did not suffer losses. Although he spent 1.5 yuan,

came up. The advertiser did not suffer losses. Although he spent 1.5 yuan, if he can earn 2 yuan back, it is still a good deal. The

result is that the developer made more money, and the advertiser got better users.

Both sides are satisfied. It is because Max Applovin has been promoted from a "buyer-only" role to a platform that serves advertisers and developers at the same time. The second

important acquisition is Appjust, a mobile app data monitoring and marketing company that cost $1 billion in 2021. In the

advertising industry, the most concerned about advertising is only one thing, that is, how much money can I make from the money I spend? But in mobile advertising, this question is actually very difficult to answer. Because an advertisement from display to click, to download, to use, and even to pay, there are many, many, many parts in the middle. And these data are often scattered in different places, and

even can't be seen at all. For advertisers, it's like a black box. Compared to

these platforms that can track data more clearly, such as Meta, it's easier to get advertisers' money. Apploving's growth has been restricted. And Adjust is the solution to this problem.

advertisers' money. Apploving's growth has been restricted. And Adjust is the solution to this problem.

Adjust is essentially a tracking and recoding tool for mobile ads. It can help advertisers to remove the effect of each dollar from the ad by tracking and recoding accurately. For example, where did you spend the money? How many users did

accurately. For example, where did you spend the money? How many users did you bring? How much did you earn in the end? More importantly, these data can

you bring? How much did you earn in the end? More importantly, these data can finally be reversed to the AI system of Adlovin to make the investment more and more accurate. Adjust can be said to be an upgrade from an ad-only platform that only takes care of advertising to a service platform that is responsible for the effect from a customer perspective. This is why Apploving emphasizes the

value of LTV, that is, the value of life cycle. We will talk about this in detail in the interview. It can be said that the layout of Max and Adjust is to make Apploving a channel for rapid growth. However, when Apploving completed its IPO in April 2021, Apple's IDFA policy cut off the ability of cross-app tracking users,

and the mobile advertising market fell into chaos. In 2022, Apploving's stock price fell into a deep abyss, falling 80% from its previous high point.

But in this dark and dark movie, Apple didn't search for the net through the source like its competitors. Instead, they made a key acquisition from Twitter.

This is the person who bought the entire cash transaction worth $10.51 million called Moopop. It was a mobile advertising platform under Twitter. Moopop

can be understood as a market for advertising. It gathers a large number of ads and sells them to advertisers. This acquisition brought three very direct changes. The first is that it can reach more people at once. Mopop was

changes. The first is that it can reach more people at once. Mopop was

serving tens of thousands of applications, including news tools and life apps, covering about 700 million users. After Apploving introduced these advertising sites, the number of

million users. After Apploving introduced these advertising sites, the number of people who directly expanded the number of ads and reached is also growing. Second, it's

about being able to vote for the right person for the ad. Mopop's advertising

position is not only from games, but also from news, fitness, life tools, etc. This means that female users, middle-aged users, non-games users are all in.

广告就不再是什么人都给同一条广告 而是不同的产品应对不同的人群。

Uploving也是从游戏广告平台 是直接升级成为了真正的全行业广告平台。

第三是整体更稳定,更赚钱。 之前Uploving的广告位主要是来自于自家游戏,

我们之前也说了 而且人群结构是偏向游戏玩家 而有了Mopop, Uploving就不再是一个游戏, no longer need to work hard to make their own games to make their own ads. A very

interesting ratio is that from growing vegetables to running a supermarket, the revenue comes from all sides, stable and controllable. To summarize, Max solves the problem of not being able to sell at a good price, Adjust solves the problem of not knowing the effect, and Mopop solves the problem of not having enough scale. Meanwhile, we

see that the acquisition of Mopop brought a very important data source to Applovin. The

increase in the number of users who regularly log in is to improve the ability of the algorithm to provide data bases. At this time, Applovin urgently needs a core technical staff who understands the algorithm. So after these three key acquisitions, Applovin is looking for another upgrade. This is the background of Applovin's

history. Ge Xiaochuan joined Apploving at this time. After joining

history. Ge Xiaochuan joined Apploving at this time. After joining

for three months, Apploving launched a new model called Appson. This

model is a very long way to the three-year company business. Here is a video interview with Ge Xiaochuan. You have been at Apploving for three years. How do

you see your three-year career?

Why did you

choose Flowing?

I think I was not a person who liked to follow the flow of things. I wanted to go to a freer and more flexible place. So at

that time, I felt like leaving Leaving the meta was quite in line with my personality. What was the most impressive moment for you? I think the most impressive moment was that everyone was very honest.

When I was interviewing Apple, I talked to the CTO, Bezos, and the CEO of Atom. They were very honest about the biggest problem they faced at the time and what kind of people they needed to help them solve it. Especially

when I stayed in the big factory for a long time, I got used to the kind of friendly people in the big factory. And

there were a lot of roundabouts in the chat. And then the first time I got to interact with this C-level person and have such a direct and frank conversation, I think that was quite impressive for me at that time. I saw

your previous sharing that you talked about Apploving and found that the company was going to further do advertising business. What they lack is the skill set you have. Can you tell us what they lacked at that time? What kind

you have. Can you tell us what they lacked at that time? What kind

of stage was their advertising business at that time? What kind of skill set do you have to make up for their shortcomings? From a business perspective, the company's business was already relatively mature at that time.

China

companies like Apple Loving and other small and medium-sized companies are still using the previous generation of recommendation algorithms. So they wanted me to help them transform the company into the most advanced generation of recommendation algorithms. What was the traditional recommendation algorithm ten years ago? And what is the current recommendation

algorithm? Ten years ago, many of the recommendation algorithms were based on boosting trees.

algorithm? Ten years ago, many of the recommendation algorithms were based on boosting trees.

Although the theory of deep learning was very mature at that time, it was not very successful in the application of the specialization algorithm. The turning point appeared in 2015 to 2017. Several

algorithm. The turning point appeared in 2015 to 2017. Several

important jobs appeared at that time. After that, the first-class companies quickly transformed into the current deep learning model. Which important jobs were there at that time? It's important

learning model. Which important jobs were there at that time? It's important

to note that there are many high-catenality sparse features in your recommendation. For example, you recommend people and The material in this case, a company like Meta or Google, the material can easily reach tens of millions of dollars. The

user you recommend is also easily hundreds of millions. Each

user has an ID, each material has an ID. Of course, there are also some less sophisticated IDs, such as national IDs, Zip Code IDs, etc. How do these IDs use and represent them in the model? How do they deal with the intersection between IDs? In fact, people didn't know how to deal with

these issues in deep learning before 2015. But in the three years I mentioned, there were many important papers that discussed how to put these IDs into the model and allow these IDs to make some cross-references. After these

works appeared, Deep Learning's framework was included in the recommendation algorithm. After you came to Apple Loving, you launched your

algorithm. After you came to Apple Loving, you launched your first model in three months. Why so soon? Because time is short. I think

life is short. That's what we do here. When you know a goal is right in front of you, we hope to make it very, very fast. Before I came to Apple Loving, my mindset was not like that.

fast. Before I came to Apple Loving, my mindset was not like that.

After all, I've been working at the factory for many years. When I first came here, I was planning to use a meta approach. For example, I would discuss the roadmap, and then make a plan. Then I would assign people and work. I would do it step by step. But soon after I came

work. I would do it step by step. But soon after I came here, less than a week later, I realized that this was not the way the company works. Just a few small things made me change. After I changed,

company works. Just a few small things made me change. After I changed, -

How many people were in the team at that time? There

were very few people in the team at that time.

The core work was included in the CTO. We added about five people. When you

launched the first model, how was the effect? Did it reach your heart? I know

you are more perfectionist. Did it reach your heart? Or did you want to push it out and see the effect first? I am a reasonable perfectionist. I definitely

won't ask the first day of the model to be perfect. But when it first came out, the effect was very, very good. I can't remember the exact number.

But from the financial report number, it should be very good after it came out.

It's obviously a difference. It must be a 10% change. Can you explain to us how Apple Loving did the advertising?

OK.

OK.

Apple Loving is very interesting. Apple Loving is one of the top advertising advertising platforms in the international market. We are the only company that has no free flow in this type of advertising advertising platform. Our target is more on the third-party trading platform. This is the biggest difference between us and Google Meta.

trading platform. This is the biggest difference between us and Google Meta.

This is a big difference, but it is also the advantage of you in terms of gaming. In terms of gaming, What do you think are our advantages in terms of gaming? Data. We don't have any gaming companies now.

But in 2022, you will feel that because you have acquired many gaming companies before, right? So you have a lot of gaming data and user data. The

before, right? So you have a lot of gaming data and user data. The

outside world will think that this is your once or once advantage. I don't know how much impact it will have on the specific amount of data. You know, in the recommendation algorithm, there is a thing called ablation test. When you guess the processing of a certain model, or a

ablation test. When you guess the processing of a certain model, or a specific data, or a specific feature, how much it affects your entire system, you will take an ablation test. You copy your system into two parts. and then erase the factors you want to measure. Then

parts. and then erase the factors you want to measure. Then

one is to keep the status quo and see the difference in performance between the two. We often do such a test, but we haven't done this test yet. So it's hard to say how much it works. Or how much it

yet. So it's hard to say how much it works. Or how much it plays a role in the path process. But in this year's perspective, we don't have this data. The advertising's effect, the hair's effect, the growth of the advertising, actually didn't have any impact on these game companies. So I

don't know if those data were really that important back then.

I think for

Apple Loving, 做产品的时候 我们不会一味着 去引导我们的 技术发展方向 去迎合这些最热门的话题 而是我们看当前

我们最需要解决的问题是什么 然后用这个来 引导我们技术发展 我觉得可能因为这个原因 我们在从产品角度来说的话 我们也算是 第一个解决了很多 在广告推荐产品里面的 一些最大的痛点 我举几个例子 第一个是 What does this so-called value-added mean? What does it mean? When many companies

launch ads, you will optimize the click rate, optimize the conversion rate, right?

But in fact, for most advertisers, what they care about is not the click rate, nor the conversion rate, but the value they bring after the conversion. Right? For

example, as an advertising company, you don't care whether you have bought 5 or 10 people after spending 100 yuan. What you care more is how much value they give you after buying 5 or 10 people. Many companies are doing this kind of which is value products. But I think we really put our focus on this. We may be the first company in the industry to

make value products bigger. For example, our Apple Albums value products are the absolute majority of advertising and advertising. This is because our model of value is the best in the industry. What is the value model? As an advertiser, you are concerned about the long-term value of the ad. Can I get

back my money within a year? How many times after I get back my money?

How many times after two years? If your ad can optimize a longer window, it can bring more value to advertisers. But before us, most of the optimization windows on the market only last one day. Occasionally, people will do it for seven days. But no one has ever used a very rigorous model to do it for more than seven days. We are the world's first company

to do it directly from seven days to 28 days. At that time, after this product was launched, it immediately became the most popular product in the industry. I think

compared to many other companies, we will really find the most advanced technology to solve the problem of the product, and then find and think about which technical innovations can help us achieve such a goal. Instead of finding a hot topic and then thinking about this hot topic, what kind of technological innovation can

we do? This is LTV, the core to surround it, to design products and

we do? This is LTV, the core to surround it, to design products and algorithms. Who proposed this? Or is it the company itself?

Why is Meta and Google I don't know. I'm not a Meta's CPO. I think what's more important is how you manage and use the resources

within the company. How to effectively help your team solve such a problem. I think this is the most important thing. I think everyone in the industry knows it's important. Can we solve

important thing. I think everyone in the industry knows it's important. Can we solve it? That's probably the real important question. Let's talk about other competitors. Besides the big

it? That's probably the real important question. Let's talk about other competitors. Besides the big ones like Minecraft and Google, are there any small ones, some potential competitors? I

see some people call them Unity, IronSource, and Charboost, and LiveOff. Why did Apple Loving run out of these companies? I think you've been using the word "competitor" to describe these. But from our point of view, these companies are not competitors. I

describe these. But from our point of view, these companies are not competitors. I

think these companies It's more like a partner. Because we ultimately serve the mobile developer and advertisers. Especially for mobile developers, why would our platform allow and welcome companies like Google, Meta, Amazon,

-

-

is actually serving the entire mobile app ecosystem, giving developers a platform where they can transform on that platform.

So for Apple, this platform itself is not Apple's main revenue. Or

rather, its revenue is a very small part of it. The other branch is advertising. When we have such an ecosystem, we also hope that one day we can

advertising. When we have such an ecosystem, we also hope that one day we can put our ads in Meta, or Google. But I don't think they will allow us to do that. We can only put our ads on the third party gathering platform. This is why, from my point of view, maintaining such an ecosystem is

platform. This is why, from my point of view, maintaining such an ecosystem is very important. When our ads are put on this part, right? Then you can

very important. When our ads are put on this part, right? Then you can say there are competitors, right? The competitors in this area are not Unity or Ion Source, but all modern marketers. There are Google, Meta, all people. In

my opinion, our competitors are more Google and Meta. Because from our purchase list, the sales ratio between their purchase and ours will be a little bit higher. In

2023, your model became stable. At the same time, you also started to prepare the Asian team. Yes. At the same time, you also started to join the senior

Asian team. Yes. At the same time, you also started to join the senior management team and start to take on more jobs. What do you think is the biggest challenge you encountered? After I came to the company, it was around the year of the small white year. It was around April or May of 2023. At

that time, the model of our Exxon 2 was first launched. And at that time, it was only a part of the model that was on the line, not all models. But that was a very important milestone. Before that, we weren't sure how much profit we could make after the model was on the line. I remember the team Before the first model was launched, we

line. I remember the team Before the first model was launched, we would often do some preparations and small-scale tests. At that time, our infrastructure was relatively weak. If the model was launched for testing, we would have to wait for about an hour. We would start joking around,

saying that we would need to buy tickets to Vegas for the party.

Then the model finally went online and failed. They said, "OK, let's postpone it and continue to wait." And then this is the state of the day and a month or two ago. When the model finally went online, I think everyone lost that feeling of joy. But for the company, I think it's a

very important milestone. After that, six months passed. The next six months were actually continuous. Because after the first model was launched, there was a

actually continuous. Because after the first model was launched, there was a very important question for us. The first model is a generation gap between the models. We took the Apple loving model from a model that is one or

models. We took the Apple loving model from a model that is one or two generations later than the most advanced model now and brought it to this more advanced model. The next question is, After the time of the time difference, were the models still able to continue to be used and

receive profits? In fact, this is one of the biggest questions the

receive profits? In fact, this is one of the biggest questions the outside world has about Apple Label. You can see that in 2023, our financial statements were far beyond our expectations. But after the financial statements, our stock price actually almost didn't move. The reason is that the industry always

thought that we will not have a delay after the return is one-off. For us, I have seen how this industry works, I am actually more confident in this delay, but it still needs time to prove it. So at the end of

2023, I think we proved the delay of the model. It

didn't stop after the model was launched.

In early 2024, Apple's ambition to be a game advertising company is no longer limited to a $10 billion game advertising. Its goal is to be a full-scale digital advertising company with a $1 billion level. The first to join the army is e-commerce. In

May 2024, Ge Xiaochuan formally proposed to expand its business field. This seems to be a bit crazy in the technology world. Because game advertising is based on virtual sense of achievement. while e-commerce is based on food logistics and resale. So, is

the advertising model of Apploving used in the e-commerce industry? I'm

very skeptical. In 2014, I think you really started to be generous and then slowly accepted by mainstream from underdog. Because this year, the stock price also began to rise, and there are also many questions. What are the three things you did right this year? I'll correct you first. You said we were

accepted by mainstream media from Unidog. I think we have always been Self-recognized underdog. And I feel proud of myself for being such an underdog. What is

underdog. And I feel proud of myself for being such an underdog. What is

self-recognized underdog? I don't know how the world views us. You think you are an underdog. Yes. Some people may think that Apple Loving has developed well. But some

underdog. Yes. Some people may think that Apple Loving has developed well. But some

people may think that Apple Loving is just a popular online company that has been popular for a short time. But I think this is not important to us.

我们自己把自己定位为一个underdog, 我觉得这个是,并且用underdog这样一个精神, 来驱动我们的一个文化的根基和我们很多做事情的方式, 这个在我看是最重要的。

你把underdog翻译成中文叫弱者之心嘛, 对,然后能不能再给我们解释一下你对这个词的定义是什么?

我觉得其实更好,可能更准确的一个翻译是不被看好的人吧,对吧?

热之心是为了凑一个四个字的成绩。

我觉得Underdog其实它有很多很多不同层次的一个意义。

首先我觉得Underdog就是代表了你不会骄傲自满,因为你没有骄傲自满的资格,别人不认为你很好。

Secondly, underdog doesn't have the burden of fear of failure.

Because you are an underdog anyway. No one thinks you will succeed anyway. On the

contrary, you are a natural child who has been successful all your life.

You may not dare to make a very risky choice. Because you have been successful for too long. You don't want to break this personality. Underdog

doesn't have such a problem. I think that underdogs have a lower ego.

They don't focus on themselves in a question, a matter, or a mission. They focus more on the mission. I think that underdogs also

mission. They focus more on the mission. I think that underdogs also have a desire to prove themselves. Because everyone thinks that not being looked up to, right? So this

desire to prove oneself is actually very powerful for a person. You

mentioned the three things to do right, right? I think it's really not a way we think in Abloving. I will never put 365 days of mood work to make it as simple as three correct decisions. I think this kind of statement is a kind of harm to the entire team. Because there are

dozens of people in the team, and they are all doing their part. And

these parts are gathered together. Uh, uh, uh, uh, uh, uh, uh, uh, uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh

uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh

uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh uh

uh, uh, uh, uh, uh, uh, uh, uh, uh, uh, First of all, when it comes to AI, I think everyone's definition of AI is vague. In

the mainstream, I think in the general narrative, AI is mainly about LIM models. But in fact, the word AI has been mentioned often before

LIM models. But in fact, the word AI has been mentioned often before the appearance of LIM models. In fact, the recommended idea is before the appearance of the original model, it was the most successful application in the entire AI application field. Apple Loving is a world-class company in the world of

advanced algorithms. It certainly uses all the AI technology related to advanced algorithms. As for the language model, it's not a secret that Apple is not training its own language model today. But we have indeed used the language model in our products to help us improve our products. And the product and algorithm will

use the language model in our algorithm. This is undoubted. As

for the outside world, how do we see Apple Loving? Is it an AI company?

Or is Apple Loving because of AI benefits? We really don't think about this problem at all. We don't care about how others will evaluate us.

Because in our opinion, is Apple Loving an AI company? It has no impact on our daily decisions. You said that AI will be involved in every part of the process. You said that AI refers to the large-scale model, and LIM will be

the process. You said that AI refers to the large-scale model, and LIM will be involved in every part of the decision-making process. I believe that other companies will have such a approach in the industry. Right. Manta will also have it. We saw that the stock price has risen so much recently, and they also

it. We saw that the stock price has risen so much recently, and they also said that their ads have been promoted. There are many differences. How to use AI?

What are your advantages? I think we have a big advantage in terms of the current situation. On the contrary, our company is not

current situation. On the contrary, our company is not self-proclaimed. Because we are not self-proclaimed, we have the

self-proclaimed. Because we are not self-proclaimed, we have the full autonomy to choose the best model on the market. Although many

companies are self-proclaimed, I think everyone will recognize that the best language model in the world is Gemini and OpenAI. Because

we don't have a self-proclaimed model, we can freely choose the best language model. Because of this reason, I think in many applications of

language model. Because of this reason, I think in many applications of language models, we are on the front line of the world.

For example, a year ago, we used the large-scale model to fight against some fraud and integrity issues in advertising investment. Recently, we are the first market to launch a

investment. Recently, we are the first market to launch a large-scale model to automatically produce advertising materials for advertising. We have already invested in this product to our advertisers. We are all leading the industry in this regard. The reason why

advertisers. We are all leading the industry in this regard. The reason why we can do this earlier than our competitors is because I have made a judgment a long time ago. For example, in the content production of advertisements, I think in a short time, only two

companies can do the best, Google and OpenAI. We don't have to do something they've already done well in this field. We focus on how to use their model and integrate it into our product. And this decision, you have to know that this decision was made a year ago when Google and

OpenAI's model was not ready yet. At that time, our decision was to believe in our partner. Google and OpenAI are actually our partners to us.

We believe in our partner. We do some things that they won't do. We wait for them at the other end of time. We know

do. We wait for them at the other end of time. We know

that one day they will make this model. So when Vio 3 and Solver 2 came out, that was for us, "Oh, the right moment has come."

So at that time, the flow of our product was ready. We just waited for their model. Once their model was ready, our product was ready to go. So

their model. Once their model was ready, our product was ready to go. So

it feels like a speed contest. Who can use this product well and use it to the state of art, the most advanced level, and push the product out. I

think it's not just speed. I think it's also an art of making choices.

A lot of things, even if it's the most popular, even if it's the most exciting, I don't think that will

happen.

Right.

- - I mean, we ask if our success is caused by a specific decision. But

it's not like that. The decision itself is actually very simple. What really makes the company successful is how the 365 days after the decision are spent every day. Right? So back to the beginning of 2024, there were a few, I think,

day. Right? So back to the beginning of 2024, there were a few, I think, important times. First of all, our technology has been confirmed in mobile app

important times. First of all, our technology has been confirmed in mobile app ads. At least it's not too different from the world's top companies. We are fully

ads. At least it's not too different from the world's top companies. We are fully capable of competing with them. Secondly, in the mobile phone application field, the ceiling is not particularly high. So the market also thinks that Apple AppLighting, even if it is very good in mobile applications, our online is not particularly high.

I think the possibility of the algorithm being able to do things in e-commerce is a good understanding.

I think our technology at that time could be transferred to e-commerce. So in this context, we decided to do

to e-commerce. So in this context, we decided to do this project without taking much time. This happened

around May of 2024. When you were working on a different industry, a track, was it difficult to adapt the model? It was difficult, but it was a very subjective

model? It was difficult, but it was a very subjective concept. What did you need to do to achieve this? First of

concept. What did you need to do to achieve this? First of

all, it was feasible. A group of people who are very good at solving problems, and a team that is very good at thinking with the first-person perspective. At this time, when you go from a familiar track to a field that

perspective. At this time, when you go from a familiar track to a field that you are not familiar with, right? Or a field that you are not so familiar with. I think the most terrible mistake is to use experientialism.

familiar with. I think the most terrible mistake is to use experientialism.

I think I was really reminded of the team's small decisions at that time. I

think the principle of first-person thinking was very important at that time. For example,

can you give us some examples? For example, how did it work in the game field? Then what are the difficulties in the e-commerce field? For example, I

field? Then what are the difficulties in the e-commerce field? For example, I don't know what kind of difficulties will be encountered in data or algorithm. Then

how do you solve this problem?

Right.

The first principle is that you have to understand which differences can be easily transformed. Which differences require different processing methods and

transformed. Which differences require different processing methods and changes in the model. When you have the first version of a simpler model and system, it will definitely not be as easy as it seems. Because you are a new system. When it is not as easy as it seems, you have to know to analyze which one you think is actually good. Then there may

be some problems in that piece, and then you can analyze what the problem is behind it. It's a data problem, it's a sound problem, it's a model problem,

behind it. It's a data problem, it's a sound problem, it's a model problem, or it's a feature problem, right? So it takes a lot of this, I think it's very good, very, very rigorous thinking to judge. And when we did this, our team members were very, very few.

I think you can take it as a re-making of a model.

When you have a team, and your team has just built the most successful mobile application recommendation algorithm and model, right? Then you can use a lot of infrastructure.

For example, the infrastructure of my training model. and the

infrastructure of the model. But how do you collect the data behind it?

What kind of structure does the model use? What characteristics does the model have? Because the characteristics of e-commerce and applications

have? Because the characteristics of e-commerce and applications may be very different. So you can almost think of it as a new startup, starting from scratch.

- So I think the difficulty is that you have a lot of things that can be copied and pasted, but in fact, there is none. If you

assume that all of these things are almost restarting, the only thing that comes out is the past year's experience in your mind. Can I understand that since you can be an e-commerce startup, you can also do other things in other industries? It's not a negative model anymore. It's about reintroducing

other industries? It's not a negative model anymore. It's about reintroducing different products. You will target different industries and different platforms in

different products. You will target different industries and different platforms in the future. You are right. I have always thought that Apple Loving is excellent

the future. You are right. I have always thought that Apple Loving is excellent because we are good at solving problems. This is what I have been encouraging the team to do. I have always told the team that Apple Loving is a method that allows us to solve any problem in the world that is worth solving.

For example, we don't have a large-scale training model for Apple. Because in my opinion, the best ROI is not to train for a large-scale model. But if

one day we feel that we need to do this, we can also become a world-class tier 1 player. I think the outside world has a deep understanding of you. Because I have read a lot of reports and reports that the

of you. Because I have read a lot of reports and reports that the outside world thinks that you have been playing games for so many years,

-

对,你看外界总是希望找到一个解释,让我们的成功看起来 It doesn't

bother us at all.

Apple is a very smart company. They can do everything successfully. They can do it.

If this day comes, I will feel uncomfortable. This is the spirit of Underdog. We are good at doing things when everyone is not optimistic

Underdog. We are good at doing things when everyone is not optimistic about us. You think it's your team. Your advantage is your team. Yes.

about us. You think it's your team. Your advantage is your team. Yes.

这一点其实还蛮颠覆人知的。 我问你,我反馈问你一下,

对吧? 你现在自己在做自媒体, 你以前也是在一个比较,

a big media platform. When you leave your own business, how come people around you are cheering for you, saying that you are doing great, that you have made a good decision. No, no. My mom asks me every day, why do you have to do it yourself? Right? Do you think this kind of doubt is actually a great deal of motivation for you in your life?

OK.

I can only guess. I wasn't the one who directly participated in this decision. After the incident happened, Adam told me that we

this decision. After the incident happened, Adam told me that we were going to sell the game. I said, "That's reasonable." Adam was the main participant and decision maker. The whole process of the whole drive was done by Adam. From my perspective, I think it's

reasonable. Before Exxon 2.0 was successful, the company had

reasonable. Before Exxon 2.0 was successful, the company had many different business lines. But when our advertising promotion strategy gradually became successful and the future of e-commerce became clear, you can see that the profit rate of advertising is very high. And

the density of talents is also very high. So in this context, I think it's a reasonable decision to get out of the game. As I said, the data of the game is not that important for our core algorithm. So I

think this is a reasonable decision. Although I didn't make this decision.

Did you have a rebound? Did anyone question you? No. Everyone thought that you would be in the OEM advertisement. Yes, I think this is also a very interesting point of Apple Live. I don't know, I didn't tell you before. I said

we rarely have meetings in the company.

对吧 我觉得很少开会一个一个前提条件其实就是大家在就是方向上 is very consistent. Even if you don't interact with people directly, you can ask them

consistent. Even if you don't interact with people directly, you can ask them and say, "What do you think the big picture is?" You will find that everyone's understanding of the big picture is extremely consistent. It's also a condition that when everyone's big picture is consistent, everyone decides how to go with the next step

according to their most reasonable judgment. You will find that everyone is actually not bad.

TikTok

Although TikTok is finally down, why did you want to buy TikTok? Because we wanted to...

When TikTok was discussed on the stage, we thought that for us, having such a platform

Yes.

Is this public news? It should be public news because we have publicly recruited people. Can you disclose what kind of route map or roadmap? We want to

people. Can you disclose what kind of route map or roadmap? We want to create a new next-generation social media platform. And then, such a social media platform can bring Apple fans a very valuable free flow. Will it follow the old path of Meta? I think we are the opposite of Meta. Meta started from

free flow. After free flow, it built its own advertising and algorithm team

free flow. After free flow, it built its own advertising and algorithm team based on free flow. Then it introduced its algorithm and advertising team technology to the third-party platform. We started from the third-party platform. After

the third-party platform was successful, we went to build this free flow. I think our path is not easy, but it has its advantages. I think building an advertising platform in terms of freedom of flow is harder than building a third-party platform. Or the competition is lighter than building a third-party platform. It's

platform. Or the competition is lighter than building a third-party platform. It's

very difficult to do it on a third-party platform because you are facing countless competitors. Your flow is not reserved for yourself. But I think

countless competitors. Your flow is not reserved for yourself. But I think this is like a plant that grows in a harsh environment. Although you

encounter a lot of difficulties, these difficulties itself make you more tough. I

think Apple Loving has been in this third-party platform for so many years. In

fact, our technology and our culture have become very tough. Now, we are building a free-flowing platform. I have a very optimistic vision. It's hard to do

free-flowing platform. I have a very optimistic vision. It's hard to do social media now because the traffic has been blocked by these giant companies. Do

you remember what was difficult to do three years ago? Advertising.

Yes, advertising. Maybe another opportunity will come. So many people will not look up to us. So many people will question how Apple's lobbying will succeed in such an environment. It doesn't matter. What are your next strategic goals? I know you are still recruiting teams in Asia. You will also

goals? I know you are still recruiting teams in Asia. You will also recruit top AI talents. What are your next strategic goals? I think there are several strategic goals. First, when we talk about strategic goals, we always think of new things. But now, I think we have The future development of

mobile applications and e-commerce is still the most important part of our strategy. Within many years, they will continue to define the baseline of our company. This is very important.

Another key strategy is Supply Expansion.

Supply Expansion is an expansion of advertising traffic. Adam mentioned it a year or two ago. It's an

traffic. Adam mentioned it a year or two ago. It's an

expansion of traffic in the CTV sector. Another

expansion of traffic is a very ambitious project.

to build our own social media platform. To summarize, why Apple is determined to sell its gaming business? There may be deeper business considerations behind this. First, the historical mission of the data lab has ended. CEO Adam said at the press conference that we started to acquire a gaming studio seven years ago to help us train the earliest machine learning models. This is a key step in creating the AI

that depends on our Axon platform. However, we are not game developers in essence.

Now, with the maturity of Accent 2.0, the algorithm can extract enough signals from the global price stream of Max platform. The so-called first party data, which is the data collected directly from consumers via free channels, has been greatly contributed. The second is the value-added reset. Although the game industry is stable, 19%

contributed. The second is the value-added reset. Although the game industry is stable, 19% of the EBITDA profit is in the eyes of Wall Street as a private asset. Compared to Apple's advertising business, the EBITDA profit of software platforms is up

asset. Compared to Apple's advertising business, the EBITDA profit of software platforms is up to 81%. This means that it has completely transformed from a game and advertising

to 81%. This means that it has completely transformed from a game and advertising company to a purely high-end AI software platform. Third, the limit of resources. E-commerce

platforms have more than 10 million potential advertisers, and the scale is several times that of games. In order to achieve the growth of the ladder function in the retail market, Apploving must remove the poor gaming team and smash every watt of computing power into the generation of e-commerce models. And then Apploving's advertising business growth proved the correctness of transformation. Apploving began to rise in stock prices.

More and more reports call it the first stock of AI profit.

But as soon as the company rose, the outside world never stopped questioning and making a profit. In the battle of $2 billion, Apple's three-game strike was also as expected. The first battle was the AI disguise of Copper and Fuzzy Panda. On February 26, 2025, the joint release of the report of Copper and

Fuzzy Panda. On February 26, 2025, the joint release of the report of Copper and Fuzzy Panda was a big hit. Apple's stock price was at a high point promoted by Accent 2.0. Copper said in the report that the so-called Axon 2.0 is not a high-tech AI, but a "multi-tool" that uses the bottom of the system. The report accused the company of misusing the installation of

the Android system. In the case of the user's lack of awareness, the user used the single-click download mechanism to enter the installation game in the background.

Fuzzy Panda further accused Aflowing of using illegal SDKs to track underage people's privacy and steal Meta's sound-signal model to increase advertising effects. Apple's policy of adjusting privacy is now without a doubt to accuse a bank of stealing and stealing passwords. The second battle is the real-life question of the increase of the

passwords. The second battle is the real-life question of the increase of the water supply in Muddy Waters. If the first battle is about technical means, then the flood of the flood of the flood led by the king of the flood, Carson Block, pointed the tip of the hat to the underlying logic of the business model.

The flood report criticizes the logic of the so-called increase and transformation of the supply.

Capital Watch

to the accusation of "fraud" and "political" for Apple's participation in Southeast Asia's money laundering.

However, soon, Capital Watch apologized for some of the facts and withdrew some of the accusations, but said that it will still question Apple's further. The series of twists and turns not only reflected Wall Street's doubts about the survival of this high growth and low transparency model, but also caused Apple's stock price to experience extreme fluctuation tests. In a long period of time, The high demand for the company's stock

tests. In a long period of time, The high demand for the company's stock price has caused a series of fluctuations in the market. You joined S&P 500 in September. Is this important to you?

-

We often discuss whether we will enter S&P 500. Sometimes we

even joke about betting 10 yuan. We didn't feel unexpected.

Although S&P 500 was decided by the board of directors, it is not like some indexes. It is completely determined by the rules. But

some indexes. It is completely determined by the rules. But

S&P 500 still follows a rule framework. Right? So the checkboxes in those rules, we actually checked them several months ago. So at that time, in our internal view, adding S&P is just a matter of time. Does this mean that you are finally recognized by the mainstream market? I think it's not important. After so many

years of underdog, we really don't care if it's recognized by the mainstream market. Another

con of underdog is that many people will question you. In 2015, of course, you encountered a lot of work. I saw your CEO's writing came out and clarified it many times. I still want to ask this question. Especially for work, some of

many times. I still want to ask this question. Especially for work, some of your data is consistent.

- - I can talk about the time I was working on the project. I played a role. When the project report came out, we all paid attention to

role. When the project report came out, we all paid attention to it. I also paid attention to it. So my main role at that

it. I also paid attention to it. So my main role at that time was to read the project report carefully and then confirm whether the technical claims in the project report were true. Then I went to compare each one with our system. Then I found that they were all wrong. After that, I stopped participating in the case. I handed over the information to Adam and the

other public officials. After that, I returned to my position and did my own thing. I stopped paying attention to this matter.

thing. I stopped paying attention to this matter.

Right.

After a while, you will find that the accusations against us were not enough. In

the year of 2025, you have continued to rise in the stock market. I

also think that the attention and the report on the stock market, what do you think has changed in this process? I think the whole team is relatively stable. When the report came out, I was actually more worried that the team

stable. When the report came out, I was actually more worried that the team members would be affected at this stage. Right. Right. Even the team members would not be questioned by their neighbors. I also sent some memos to the team members at that time. I told everyone that if you have any

- The expansion of the talent of Plouin is very restrained. The first feeling when you walk into the headquarters of Palo Alto is "so empty". There are very

few people in the whole building. The company now has more than 100 engineers.

Compared to the company's financial report, it is equivalent to the average profit of $40 million per engineer. But Ge Xiaochuan's biggest problem is still in recruitment. The difficulty

is that the company is facing rapid growth, while there are all kinds of questions about the space outside, and the culture of maintaining the underdog inside.

I think the biggest challenge is that the talent market is not enough for us.

It was basically very difficult for us to find people that year. Very, very difficult.

Because everyone wanted to go to a more famous big company. Yes, and it's not like that. In fact, I think the talent selection in North America is quite good.

like that. In fact, I think the talent selection in North America is quite good.

Everyone is willing to go to some small companies with a future. But I

think in the eyes of many talents, we are not a creative company with a future.

- So it was very difficult to recruit people at that time. How did you

solve this problem? I waited. I think there were some problems. I tried a lot of methods at that time. They were not very effective. I

think I recognized it later. I think the most important thing at that time was to do things first. I used some resources that the team had.

But I think Apple Loving is a good point. Apple Loving's technical leaders from top to bottom are actually very complicated. This is also a disadvantage for us, as it is not available in other companies, including large companies. When

we want to do something new and we can't find the right talent to do it, our technical leaders can do it themselves. Write the code themselves. Yes, write the code themselves. So at that time, this was our choice.

code themselves. So at that time, this was our choice.

- and we have improved our capital growth. So in 2024, we will make some

changes. One is that our capital growth is the highest in the entire

changes. One is that our capital growth is the highest in the entire shares. Of course, after the appearance of AI, those bubbles covered

shares. Of course, after the appearance of AI, those bubbles covered us again. But in the AI field, we are still the richest company

us again. But in the AI field, we are still the richest company in the entire shares.

Why did you make such a change?

- experience. This conclusion is very important. Why? How did you get this

experience. This conclusion is very important. Why? How did you get this conclusion? You find that if you manage a company, you want to solve a

conclusion? You find that if you manage a company, you want to solve a problem, and you think that the problem has been solved by other companies. The

most direct way to solve this problem is to find a director from a company that is very good at this problem. and then let him re-organize the team and then the problem can be solved. But in fact, if you look carefully from history, you will find that most of these cases have failed. There are many, many reasons

here. I don't want to go through them one by one. But I think that

here. I don't want to go through them one by one. But I think that the real excellent talent, I think his learning ability is the most important.

But all these experiences do not represent a person's learning ability. It only represents a person's knowledge. But I don't need to hire someone to do something he already knows how to do. What's more important is that you hire someone who can grow quickly and figure out things he didn't know before. This thing's value

is far higher than the knowledge in his brain. So, the newbies are very mature. And then, in our way of doing things, we want to break the

mature. And then, in our way of doing things, we want to break the norm. Many times, this kind of old knowledge is actually a hindrance. So,

norm. Many times, this kind of old knowledge is actually a hindrance. So,

from 2024, our recruitment philosophy has changed. I have a challenge for you. I think

the ad of Apploving is not a big reason because it's you who found it. Then, your skill set has brought this business up. Yes.

found it. Then, your skill set has brought this business up. Yes.

对,所以我经常自己自嘲嘛,我说我自己是绝不会害我自己的。 I think I was lucky enough to join this company before I came to Apple. Is it

because of the different stages? At that time, you needed a new skill set because you wanted to do something new. But now you have a method, so you need to scale. So it's okay to hire some smart young people.

Yes, I think if you want to build a team with no experience, your team really needs a seed. 一个种子选手 对吧 这种子选手可以是一个 有很多很多经验的人 他也可以是一个 没有很多经验 但是他很擅长学习 这样也可以 我觉得我当时的话 其实我来Apple Labing 很多东西

并不是说我来之前就知道的 我也是来了之后 就是在时间的过程中 自己去摸索 So I think the 10-year experience of Apple Loving is actually valuable. I

think it's not that valuable. If I just stayed in the industry for two years, I think I could do something similar. What do you think is your characteristic that makes you the right person to stay in Apple Loving for three years? I think doing things and challenging them is a regular thing.

years? I think doing things and challenging them is a regular thing.

Don't follow the rules. Don't imitate them because Google and Meta are the absolute masters in this field. I have always been telling my team that we don't have as much resources as Google and Meta. We can't use their methods to solve our problems. I have seen many companies in the industry, not

only in the advertising industry, but also in other fields. I think when these companies are doing things, they often make a mistake. They are imitating the way the big companies solve problems. but they didn't realize that they didn't have the resources of the big companies. If you don't have the resources of the big companies, and you still use the big companies to solve the

problem, then you are destined to fail. Then, Apple's living will gradually expand. But I

see that you are still very restrained on hiring people. Although you are already a company with a lot of market value, I believe you have a lot of cash in your hands. How do you go about expanding? How do you keep the word "continue" in the team? And how do you avoid this kind of big company? I think the word "continue" is very good. We are not deliberately

saying that the team must be small. We are also recruiting people. We are constantly recruiting people. But we are indeed restraining. We won't recruit people for the sake of

recruiting people. But we are indeed restraining. We won't recruit people for the sake of recruiting people. We won't destroy our resources because we have enough resources.

recruiting people. We won't destroy our resources because we have enough resources.

In my opinion, when a team becomes very poor, the extra pay is only a small part of all the losses. Once the team becomes poor, the loss of opportunity cost will increase in the future due to the impact of the impact of

the younger team and the future of the interoperability efficiency. So from this point of view, this is what we have to keep in

efficiency. So from this point of view, this is what we have to keep in mind to keep this kind of momentum. As for what you said, how to avoid being a mistake that other big companies make. Actually, I'm also very curious. Everyone knows that this is a mistake. Why do companies that are constantly

curious. Everyone knows that this is a mistake. Why do companies that are constantly making such mistakes? Because there are too many people, there will definitely be politics. Right? It's just that they might not sell their products

be politics. Right? It's just that they might not sell their products 卖出第一步 因为我觉得大部分 Fond还在的这种公司 他其实都会比较节制 但可能你不小心卖出了第一步 When you make a mistake in the first step, then the company becomes more and more people, once

the number of people becomes more, you lose control. Losing control will make it more and more. We may try our best not to make the first mistake. It's

and more. We may try our best not to make the first mistake. It's

okay for now. I think at present, our company has never taken the number of people in our team as a It's a matrix that we need to show. I don't want the team to be too dominant. My managers don't

show. I don't want the team to be too dominant. My managers don't want the team to be too dominant either. So in terms of culture, we have a better resistance to the team's dominance. Will you find that recruitment is better after 2025? It's better in terms of the number of recruits.

- 但对于最优秀人的判断方法,我觉得其实是不太一样的。

在这波AI人才的竞争当中, 你会发现很多公司真的是喜欢去寻找那些聚光灯下面的人才。

对对对,背景很好。

背景很好的,并且在媒体上有过一定的曝光, 发过一些比较有名的paper,对吧?

但这些聚光灯下的人才,我觉得他们多数是被, 因为... They might overprice. But

因为... They might overprice. But

another important thing is that the people under the spotlight don't have the spirit of an underdog like Apple and Apple. We actually focus more on the people at the edge of the spotlight. They are actually very close to the G-Wave. They are also very, very good. Even I think they, if you look at

G-Wave. They are also very, very good. Even I think they, if you look at it from a statistical point of view, they are not as bad as the people on the G-Wave. But they may be better using this kind of underdog spirit. Or they are not under the G-Wave because they never want to

spirit. Or they are not under the G-Wave because they never want to be under the G-Wave. These talents are what we pay the most attention to. And then we, due to the development of the past few years, we are

to. And then we, due to the development of the past few years, we are fortunate to be able to attract such people and persuade them to join us. How

do you judge? 这群人就是在聚光灯边缘的underdog,但是又非常的talented。

没有一个很高效的办法,我觉得就是一个个聊, 还有就是说我们也在努力打造我们作为雇主的一个品牌和形象, 我们要想办法精确精准地向我们潜台的候选人传达, 我们是什么样的公司,我们希望要什么样的候选人, 这样的话也会经常会吸引到一些候选人主动来找我们。

你在面试当中你觉得你最看重的一个问题是什么?

我最看重的问题就是候选人在人生里面是不是经常做一些与众不同的选择。

这个怎么理解呢? 就是...

比如说我有个大公司,我不去,我去自己创业。

就算是一种,对。

比如说你在中国高考这样环境里面,你在高中的时候不仅不仅

Right?

Right? These are very small details, but when you see it from the inside, when he was making choices, he actually participated in his own self-reflection.

I really value such talent. I'm very curious about one thing. After you graduated from the Chinese University, many people went to study in the United States. Because everyone thinks that studying in the US is the best. But you went to Italy. This is

very very...

-

做选择的时候特立独行之类的。 你觉得欧洲的留学,

做选择的时候特立独行之类的。 你觉得欧洲的留学, 还有生活经历,还有经验, 对你有改变吗? 嗯, I think

还有生活经历,还有经验, 对你有改变吗? 嗯, I think Europe's life is more diversified. In such an environment, I think it's inspiring me to accept more possibilities when I'm making choices and thinking about problems. I think Italy is a very magical

place. It has a very rich history, culture, and romance, but at

place. It has a very rich history, culture, and romance, but at the same time, it's a bit melancholic. What do you think Italy means to you? I think Italy is like my hometown. I spent a good period of my

hometown. I spent a good period of my life in Italy when I was 20. I learned a different perspective on life.

I think the word "stereotype" is not a common term in Italy.

It's very easy to be introduced as a confirmation bias. In fact,

there are many people with a stereotype all over the world. If you look at the overall distribution, Italy is indeed You came to Guigui and did Guigui change

you to some extent? Yes, I have. I

extent? Yes, I have. I

have lived in many different places, especially in Italy. When I talk to my friends in Italy, they

Italy. When I talk to my friends in Italy, they will think Their lives and the height of a world that can change are very, very different. So they will never think about it. I think the biggest inspiration that Gui Gu gave me is that every ordinary person is actually very,

very small in the gap between him and his ability to change the world.

What you need is to be able to sell that courage and work hard every day for 365 days after selling that courage. This is why you asked me a lot about whether the moment of the problem is important or not. I think this is not the main reason for the return of the stock.

not. I think this is not the main reason for the return of the stock.

Maybe after many things are successful, we will look back and hope to sum up one or two highlights. But the real success story of Guigu is the little things behind the scenes, not the high-light moments. Very well said. Why

did you start working on a motorcycle? I didn't start working on a motorcycle every day. My family lives in a company, and it usually takes 30

every day. My family lives in a company, and it usually takes 30 minutes to park a car. If I park a car, it will take 50 or even an hour. Sometimes if I get to the high-speed period in work, I might consider that I will be faster when I get on a motorcycle.

So you ride 101 high speed? 280. 280. And the scenery of 280 is very good. Yes, yes, yes. And the crystal spring. Yes, that lake. What does it

good. Yes, yes, yes. And the crystal spring. Yes, that lake. What does it feel like to ride a car on 280? There is no car. There

is no car. What does it feel like to ride a motorcycle? Yes, yes. I

usually ride a 70 mile. I feel very comfortable. The wind blows from both sides of you. Then you can feel the sound of the wind from the helmet. 280

of you. Then you can feel the sound of the wind from the helmet. 280

often makes a fog. 280 has a mountain on both sides. Then the fog floats in the mountain. Then you ride a motorcycle, the view will be better.

就能感觉到左边右边的那些东西 能感觉到,就非常非常好。

你自己是对速度有一点喜欢的人吗? 也还好,我是一个,

其实我是一个不是那么, I like to take risks. I usually take calculated risks. If I ride a motorcycle, I won't ride it too fast. Actually, you

risks. If I ride a motorcycle, I won't ride it too fast. Actually, you

can easily drive up to 80 miles per hour. You won't even feel that if you drive 80 miles per hour on an SUV, you won't feel very fast. But if you ride a motorcycle up to 70 miles per hour, you can

fast. But if you ride a motorcycle up to 70 miles per hour, you can feel that speed is very fast. So you feel the speed and you like this feeling. Yes, I do. I heard that your job is very boring. 几时半夜还会回消息。

feeling. Yes, I do. I heard that your job is very boring. 几时半夜还会回消息。

我一直跟别人说 我说我们工作不卷 阿帕拉雯工作一点也不卷 我觉得我们只是勤奋而已。

Okay.

They are just like me, they have a lot of expectations for what we are doing. So they will take the initiative to make the choice. You have shared

doing. So they will take the initiative to make the choice. You have shared this before. I know you may not like the word genius.

this before. I know you may not like the word genius.

But you are smarter than the average person. You

went to a very young college. You are young and went to a university. You shared that before you went to Meta, you

university. You shared that before you went to Meta, you were at a low level in life. - Yes, I did say that in some cases. - Yes, you think that you will become a poor person in this

some cases. - Yes, you think that you will become a poor person in this life, and you will be looked down upon by people around you when you were young, or you will think that you are different. Why do you think there is such a gap? I think that my first major turning point in my life was when I started university. I think that my biggest shortcoming at

that time was that after I started university, I did not have a good career guidance. I was very scared of my future career. I was learning physics. I was

guidance. I was very scared of my future career. I was learning physics. I was

very scared of the question of why I chose physics and what kind of career I could do after choosing physics. At that time, because of all kinds of subjective reasons and some restrictions on external conditions, I have not been able to get a good answer. And that impact has a very deep negative impact

on my life afterwards. Because of this reason, I have always hoped that I could, through my own life, through my own experience at this stage, I hope to be able to help these young people in schools now, to help them solve some of the difficulties I encountered

back then. I'm still

back then. I'm still facing a lot of confusion. So

I'm actually doing a lot of things to help them.

I founded a personal foundation. I see. Part of the focus of the foundation is to go to Chinese universities and help college students provide them with a lot of professional guidance.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Is it a matter of your heart? Or if you keep trying, you

will find it. Because I think there may be many geniuses in high school.

It's also very, very, very important. But later it may be a person. I

think there are many things in life. There is no way to have so many ifs. You don't know what will happen after the if. But the

objective fact is that before going to Meta, I did make several mistakes in my career. When I went to Meta, I actually got offers from other places.

career. When I went to Meta, I actually got offers from other places.

Then I thought about it again, if I didn't go to Meta and went to one of the offers, I think it's very likely that there will be a wrong choice again. So I think I was lucky to choose Meta at that time. How do you define a wrong choice? The wrong

choice is of course looking back now, you know it's a wrong choice, but you didn't know at that time. Yes, I didn't know at that time.

I didn't know about the mistakes I made before I joined Meta. You came

to Apple Loving for three years and you have a great career. It's rare to see this in your hometown. I believe many Chinese engineers have a strong background. When they want to sell to the leadership, What kind of advice do you have for them? For example, they want to

turn from a high-end engineer to leadership. What kind of characteristics and skillsets do they need to break through this ceiling? I

think the most important thing is the framework of recognition.

Many people in large companies and development industries are still pursuing a system that gives them recognition. Yes, the big company system. The last one is the system. For example, they will be extra concerned about what kind of rating they

the system. For example, they will be extra concerned about what kind of rating they will get in the performance review system. When will the system be able to promote it? What level do you think it is? They will be very, very concerned about this. But I think if you really want to be a leader, you should focus more on your own standards and your own recognition of yourself. I

think when you pick out such a thinking pattern, it's easier to become a leader.

When you become a leader, what you do is not the standards that the system sets for you, but the standards that you set for yourself. You just talked about growing up as a leader. I think there is a second point I want to share. In this world, it seems that the choice is very high. It is all well-priced. When you see a high-certainty

choice, even if it's a good choice, the cost you have to pay will be high. Or the fixed cost you pay will be limited. Because

high. Or the fixed cost you pay will be limited. Because

everyone knows the value of a high-certainty choice. So you will have more competition. So it's important to identify and appreciate these uncertain

competition. So it's important to identify and appreciate these uncertain choices.

Transaction.

Transaction.

They will think that they need something to join the company at this moment. They

don't want to appreciate it. In this process, it is an investment between the company and the individual. The company is willing to give them an important opportunity in an uncertain situation. They are willing to give the company some benefit without. This is also very important. This kind of job opportunity is an

benefit without. This is also very important. This kind of job opportunity is an investment. The return of investment is in the future, right? If you invest, you

investment. The return of investment is in the future, right? If you invest, you will lose money in the short term. If you buy 100 yuan today and invest in something, you will lose 100 yuan today. But if you only look at this opportunity as a transaction, if I give you 100 yuan today, I will get 100 yuan. When such a return is made, you actually put a

lot of very, very precious opportunities outside the door. Very good. The last

point is for you. For those who are considering applying to work at Aflavin, or for those who just graduated and are a little confused, do you have anything to say to them? Please call everyone to apply for the job at Aflavin. When young people are making choices, the easiest choice is to choose the most

Aflavin. When young people are making choices, the easiest choice is to choose the most popular ones. There is a reason why they are popular. Because they are generally

popular ones. There is a reason why they are popular. Because they are generally not the worst choice. They are generally relatively decent choices. But if you want to be an extremely good person, right? You want to be the best person in the world, do something that others can't do. You have to have the courage to make some choices that are different from others. You have to use

your own judgment to find what suits you best. I think in Apple Loving is a company culture. We encourage you to challenge the status quo. We encourage

you to make different choices. Although we have very few employees, we are a large company worth more than 200 billion dollars. But our company's culture and way of doing things is still like a first-class company. We

have less than 100 engineers so far. So I think any kind of I think Apple's engineers are very good at their own work and they are willing to create future candidates through their own efforts. I think Apple Loving is a very suitable stage for them. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own

work. I think Apple Loving's engineers are very suitable for their own work. I think

work. I think Apple Loving's engineers are very suitable for their own work. I think

Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think

work. I think Apple Loving's engineers are very suitable for their own work. I think

Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think

work. I think Apple Loving's engineers are very suitable for their own work. I think

Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think

work. I think Apple Loving's engineers are very suitable for their own work. I think

Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I think Apple Loving's engineers are very suitable for their own work. I

work. I think Apple Loving's engineers are very suitable for their own work. I

Humility.

Empathy is the empathy for your colleagues and our clients. I think this is very important. I found that there are many excellent engineers. When they lack

very important. I found that there are many excellent engineers. When they lack empathy, they are always obsessed with what is the most fancy way to solve this problem. Then I was in there,

problem. Then I was in there, ability ability resilience resilience I think that in this world, you have to do things from one

to one. There is nothing that is one-sided. You do things that others

to one. There is nothing that is one-sided. You do things that others have not done before. There is nothing that is one-sided. How you can continue to persist after encountering the wrong person is very, very important. And then the last point is tenacity. And these five are linked together. In fact, it is an English word, but it is heart. So I just summarized the most important five characteristics of Apple's engineers.

This is not the end of the story of Apple Loving. In the

midst of the ongoing ups and downs of the advertising industry, the market is facing unprecedented uncertainty. In the past two years, Apple Loving has been regarded as a

unprecedented uncertainty. In the past two years, Apple Loving has been regarded as a "cow and cow" in software stocks in the U.S. However, this feast, once defined by software, is facing an unprecedented AI trial. In February, the entire US stock software market was hit by a brutal sellout. The software was in a very bad state. From the 40% drop in the price of Unity to the collective downfall of

state. From the 40% drop in the price of Unity to the collective downfall of Adobe, SAP, ServiceNow, Snowflake, and others, panic was spreading. Investors began to reflect on the situation. When more advanced AI tools such as the Android-based automation suite and Google's

the situation. When more advanced AI tools such as the Android-based automation suite and Google's deep-sea world model begin to reproduce the production history, are the mutual funds of traditional software companies already in the same shape? In this industry-based long-term shock, Apple's share price has dropped once in a few days. Before the financial news broke out in early February, the share price was almost at a downturn in mid-December

when I talked to Ge Xiaochuan. There are three reasons behind Apple's fall.

The first is the new Google 8-part Project Genie. This simple-to-use 720P model is only able to interact with 3D world. It takes a few seconds to complete the traditional game engine. It takes weeks to complete the work. Despite some analysts' strong support, it does not have the balance and deterministic logic required for commercialized

games. But panic has spread in the market.

games. But panic has spread in the market.

对于Apploving而言,Project Genie的威胁在于对其广告供应链的潜在重塑。

市场担忧如果谷歌能够通过AI一键生成海量的轻量级互动内容,并将它垂直整合进自身的广告体系, 那么Apploving现有的游戏广告库存价值将会被稀释。

那么这个猜想是带崩了整个游戏股行业,包括Apploving,Unity,Roblox等等, 反映出投资者们对AI替代论的极端脆弱心理。 The second is

反映出投资者们对AI替代论的极端脆弱心理。 The second is that the former engineering has come out of the dark. On February 4, advertising technology company CloudX officially announced the full launch of the product. 并且高调地喊出了利用AI智能体AI

agents 重构移动广告技术站的口号 CloudX之所以会对Apploving产生影响 核心是因为它的创始人James Payne 他是Mopop的创始人 我们在开头提到过 他曾经亲手为Apploving搭建了帝国的基石 也就是MaxG和平台 如今这位最初的缔造者想要搭建新体系 The core profit

of applying is the price of the Max platform's algorithm. Payne's main CloudX claims that using the large language model can turn advertising stations into programmable infrastructure.

The core weapon is AI agents. These smarts no longer follow the default price rules of the death version, but are like real negotiators, using the credible implementation environment to implement real-time code generation, directly bypassing traditional platforms such as Max. If the first-hand developers find that using CloudX can jump over the gap

as Max. If the first-hand developers find that using CloudX can jump over the gap between the middleman and obtain higher transparency, the network effect of Applovin's hard-working management will face great challenges. The third is Meta's revenge. Now Applovin's

biggest competitor Meta has returned to where it once was. Since Apple's implementation of the IDFA privacy policy in 2021, Meta has been greatly improved in terms of recommendation accuracy on non-authorized traffic. This gives Apple's X3 engine a great opportunity to overtake.

However, with Meta investing billions of dollars in capital spending, its advantage and AI system have shown an amazing evolution. Meta has greatly improved the EZPM price density of iOS applications by the end of January this year. The third-party issuance of some parts even increased by 3-5 times. Meta is trying to prove that through huge

user behavior data and advanced prediction models, even without IDFA, it can still achieve accurate投資. This is undoubtedly a tough battle for Apploving, which is the basic platform

accurate投資. This is undoubtedly a tough battle for Apploving, which is the basic platform of the iOS market. Our interview has been re-run for the past 1000 days. How

is Apploving tearing apart the gap between Google and Meta, the two monopoly giants?

However, when AI technology tries to conquer all industries, this gap is just the beginning of the challenge of advertising. Will Apploving be the last one? We also

look forward to it. Apploving and Gao Xiaochuan, who are the leaders of the underdog, will bring these companies to where?

拜拜