Anker / Steven Yang: Consumer Electronics Death & birth, The Third Category, Product Philosophy

(English subtitles were generated by AI and are for reference only.) Hello everyone I am Xiaojun In the previous episodes Together with you I met many very young faces They are in the AI ​​era Native entrepreneurs and technology leaders But AI’s penetration into society is all-round It’s not just about new products and new technologies Our guest today is Yang Meng, the founder and CEO of Anker Innovation He was born in 1982 Started his own business in 2011

Now he is at the helm of a technology company with a market value of more than 60 billion yuan In the face of new AI paradigm changes What would Yang Meng do?

Especially in consumer electronics On the battlefield of Infinite War Next is my interview with Yang Meng This world is particularly fair What I just said We actually didn’t have any difficulties in financing before right? Then everything went smoothly.

right? Then everything went smoothly.

In fact, our life has been quite painful these past few years.

I think DJI is a player from day one I chose a difficult mode It’s even a hell difficulty mode company This is the first category I chose It’s actually a very, very difficult category.

So the first few years must have been very painful Becoming a third-category company Generally a CEO is needed What kind of character and personality is he?

hello Steven Say hello to the viewers.

hello Xiaojun I am the founder of Anke Innovation and my name is Yang Meng I’m glad to be on your show Yes I saw you said you want to do business from a company Transform into a deep technology company Has the world changed or have you changed Everyone’s first reaction when they heard Anker It seems that I first got to know this brand because of the power bank Then when I heard that Anker was making chips

to make embodied intelligence, right?

When doing these complicated things then will be very curious Actually we think this is a company it evolves naturally Step by step Stepping up A natural process Right Let’s talk about it from the beginning We also want the audience and listeners to know Who are Anke Who are you? From what I saw about your growth experience It seems that you have always been a good student in the eyes of others.

Including Changsha Yali Middle School Then Peking University Department of Computer Science Then go to the United States to study for a Ph.D.

Although I am a graduate I am working at Google again I have worked for 5 years And then won Google’s highest award, the Founders Award Did you grow up as an out-of-the-ordinary person Or are you an out-of-the-ordinary person?

Am I a good student I must be an above-average person I must be the kind of good student that everyone sees And in my heart it seems that I also use good students as the standard to demand myself right by the way The Founders Award is actually a collective award.

It is not an award for individuals Yes Tell me about my experience Actually, I majored in computer science Undergraduate Then what?

After graduation, I went to the United States to study with everyone Why I followed everyone Because I was in the department at that time The top one-third of the students may have applied to overseas schools and then went to the United States to study for Ph.D.

Yes This is not necessarily the case now But we were in 2003 That was almost 23 years ago It’s still the same trend Yes Then I went to UT Austin At that time, the CS ranking in North America was pretty good My professor’s name was Raymond Mooney He was a pioneer in the machine learning world Then he served as the chairman of ICML for several terms Machine Learning

is to use machines to do AI It has always been something that I particularly like and am very excited about But after I went there, Not feeling particularly good Because we were doing something at that time very small data set It's hard for you to imagine It’s an entire data set There may be more than 20,000 Reuters articles Then the adjustment of your various algorithms is based on Among the more than 20,000 articles

20% of the test set is more than 4,000 articles The difference between one point and two points above Right, then you feel It seems that you can’t feel its actual meaning Yes, so in 2005 I went to Google for an internship I jumped to it immediately It was to do algorithms from 4 billion web pages Right then Then you know that Google At that time, it was hundreds of millions every day

It’s the search volume You can change even one thousandth of the search It’s actually a big number So I felt that it was special and of practical value It was special and really useful Yes, so I didn’t go back to study for a Ph.D.

Just stay at Google Just dropped out of school, right?

In fact, dropping out of school was an accident It was not a well-thought-out choice In fact, it should be said that it was a It’s the inner calling It’s when you do something very practical today that creates value Then go back to an ivory tower and do something that is as sharp as a hornet's nest as if you don't know what the value is You feel a little reluctant Yes Did anyone object at the time?

Is this the first rebellious decision you have made It must be a big decision Because my family wanted to say hey It seems like you have studied for a Ph.D.

You finish it and then talk But I feel like I really can’t continue reading.

I think my choice is actually very consistent It’s just that I think I must create special and real value Yes and is not that special.

seems to be illusory.

It’s something that’s very far away I definitely hope to be able to stand where I am today can see There is a real value that we can achieve You didn’t want to do scientific research, right At that time In fact, the purpose of studying for a Ph.D. is to do scientific research I think doing scientific research may be divided into two types of students One kind is the ability to invent a completely new architecture today

Or the framework opens up a new direction What about another category?

Yes In fact, it is to improve in the existing direction or to combine it with specific problems There are very, very few people in the former category In fact, maybe every five years It will take ten years for a big star to come out Right, then it will really open up a huge direction Then most people will be like me In fact, they are all in the back direction.

Doing a lot of additions That What about our time The big vein was called Support Vector Machine SVM In fact, it was after the decision tree A big breakthrough in the machine learning world It was probably invented in 1997 or 1998.

It was very popular at that time because it was very effective and could solve many practical problems But by the time I got my Ph.D.

In 03 and 2004, basically this mine I’ve dug almost as much as you can see.

Right And then I feel like I haven’t found it either Just something that seems to be particularly capable of creating actual value Then go back to Google You think wow He can really change the lives of many people So I went How did you face the people around you Opposition On that decision Actually, because I am an I person So this decision was made without consulting many, many people.

And even though my parents objected But after all, I have grown up Oh Yes, then the rest is fine on Google Started internship in 2005 I started working officially in 2006 Yes Summer internship in 2005 Then after the internship, they gave me an offer So I went to work at Google at the beginning of 2006 What did you learn at Google?

Me I feel that I have learned too many things at Google No matter from the corporate culture Right, don't do evil Then work hard to create Or go to the specific methods Just today you found him thinking about the problem There are many ways to solve the problem In fact, it was all subtle It was buried at that time And then there is the atmosphere of the company today.

In fact, you know that Google has always been a very egalitarian Open and transparent atmosphere Yes, so when I started my own business later and then when I started a company I also had this kind of atmosphere Bring it to your own company For example, I don’t have my own office Sit with everyone Then we are also very equal when we have meetings And all our meeting rooms are not walls

It’s glass, you can see through it at a glance You can see what’s going on inside So equality and transparency bring openness Open There is an environment for collision and creation today Yes So I think the first company in my life actually had a huge impact on me I’m very grateful for it I heard you were with Jeff Dean?

There are also some direct contacts I submitted a changelist to Jeff Dean I modified his code After the modification, he helped me do a code review If this is considered contact Then I think it actually means that may not necessarily be able to come into contact with or may have come into contact with quite a few people in the company.

Because after you joined Google you found it particularly magical because of all the backbone codes From the very beginning Google File System to MapReduce to the file headers of the main types of files in Bigtable The first author is either Jeff Dean or Sanjay Ghemawat so he is the great one There you feel like the whole world was created by him

Then we are in the world he created Do a lot of improvement work So it’s the same feeling as in academia.

That is, someone can be like create a world Then we go inside Make the world a better place Yes But do you want to be the one who creates the world That point I don’t think I have that strong desire It’s the feeling that I can create a world Yes What kind of engineer are you at Google?

He is very stand out He is still a very decent engineer.

What kind of person were you at that time I actually don’t write too much code Because I am biased It’s abstract thinking Architectural thinking I am willing to discuss with everyone It seems like the definition of the problem The dismantling of the problem Right and so Our manager quickly promoted me to a We are called Tech lead which means we lead four or five people

and then take on the specific tasks of a module Right So I’ll do it with you later It’s just like communicating issues externally Then do the system architecture internally To be honest, Google engineers don’t have to worry about it Right?

Everyone In fact, everyone is very willing to do a lot of very creative things But how to clarify a direction with our customers and come back to our team Today we all reached a consensus Then of course we write part of the code ourselves right And this actually has to be done by someone Right How did you start doing Anke’s early work?

Were you still at Google at that time My girlfriend at that time Later wife She is in Silicon Valley We know each other That’s because she is a liberal arts student So I haven’t been working for a while Then we are thinking Find something that she can do together In fact, when I went to the United States in 2003, Just start buying things online First words Just buy things on many different websites

By 2008 and 2009 you will find Basically you shop on Amazon because of its experience its richness Its price including its logistics are the best Yes Then we discovered Actually there are a lot of things on Amazon Its selection is not the best Then So she was my girlfriend at that time Later, my wife Then come with me and we just want to say Then

Can you start selling some things on Amazon And then actually help her as a liberal arts student A classmate who has difficulty finding a job in the United States There can be one thing to do to and so the very beginning is Selling something In fact, you soon found out said Selling in e-commerce itself is not the difficult problem Where to get the best products The difficult thing is Let's put it that way

In fact, I quit Google in 2011 zipped up At that time, the person in charge of sales in Google China was named Zhao Dongping Dongping and the two of us I make products He is responsible for global sales We founded the Anke brand together Anker was originally an e-commerce company In fact, at that time our name was also called Haiyi E-commerce Just bring an e-commerce company by name Then we started looking for it from China

Soon you will find that you can’t find good ones So you have to go to the factory to customize to customize the best products Then sell it to the United States Then of course it won’t be that expensive Right then So in the beginning because there were no relevant e-commerce talents anywhere in the country We just want to say Then build the company in Changsha We will train e-commerce talents ourselves So after so many years Indeed

our e-commerce team in Changsha is very stable That is, the school admissions we have trained ourselves Hey, at that time, you would be an engineer at Google and then return to China to start a business Do you understand the idea of ​​making a power bank Actually, to be honest It will happen many years later When introducing it to everyone It seems that I don’t even carry a power bank on purpose.

Right Including today We definitely want to label the power bank Being able to tear it off because what we actually want to do It’s still a matter of creating real customer value In other words It can really satisfy people and bring value It’s just that point I just don’t have anything that sounds fancy.

Then create such great value It’s just that point It’s in the channel The most clear things we see that can bring value to customers are actually charging products Yes Because there were smartphones at that time Actually, you went back to 2011 Smartphones are developing rapidly At that time, the electricity would not be used for a day So the power bank is a thing It's something that everyone needs very much so we want to

Then let’s quickly leave this value blank Make up for it You quit at that time Did your family say anything?

In our family, my father is a teacher and my mother doesn’t really care about my studies So my father still expressed his objections Do you think you should talk about it first In fact, when I quit my PhD before My family actually didn’t say much.

Oh on the other hand Actually my mother Because my mother In fact, she is the person in charge of the sales of a state-owned enterprise Then she asked me if you need money to start a business I said I still need some start-up capital.

I said there were investors outside who were willing to invest My mother was very happy at the time My mother said, is this true?

I said I thought it was pretty reliable I said it could definitely be done My mother said Let’s do that reliable thing with our own money So She gave me 3 million Then she is our angel investor Oh, your mother is your angel investor Yes But in the end she asked for the money back but not the shares So everyone didn’t see this Hey, why did she want the money back Actually, from her perspective

she just wanted to support me Just get this thing done Actually, it’s not like I’m going back.

that is In fact, she wanted to support me in doing this Then she said that in the long run how can I do this better?

That's right The seeds for starting your own business At what moment was it planted?

Was it when you were a student or at Google Or where In fact, I have always been It’s about truly bringing value to customers.

This has always been a special one for me an exciting point Just visible Satisfying others Yes So actually I have always had this motivation to start a business.

In fact, we first started with electricity E-commerce started to start Then so Because China doesn’t do anything overseas today.

So e-commerce talents So we returned to Changsha Dongping also moved his family from Beijing to Changsha It was incubated in Changsha I brought out an e-commerce team and then soon I moved to Shenzhen at the beginning of 2012 Because we found that selling is not a difficult thing A good product is the most difficult thing And if you really want to have good products

then you have to come The place to create products is Shenzhen What kind of person is Dongping?

Why did you start your business?

Choose to pull him Dongping is fun It’s just my personality It’s just that I am an I person INTP Just thinking in a special logical framework Right, very rational and very naughty?

Yes Dongping is the complete opposite of me He is an ESFJ Oh the complete opposite To both of us they are like the poles of the earth Antarctica and North Pole The difference in style is huge And you find We have been together for 15 years Do you think our personalities are closer or more in opposite directions You ask me that It probably went in both directions We definitely found out

In fact, after 15 years Because there is someone who can look at your back safely right When there is a person who can see your not so good side In fact You are more bold and go towards what you are good at Oh, so how did you meet At that time, I returned to Google to start a business Our colleagues at Google think Should be doing overseas business My classmates all wanted to buy Google ads

and then they introduced me to who was Google at the time The advertising sales team in China Then they thought This classmate seems to be more interesting He didn’t just want to sell things He wanted to make good products Then he introduced me to Their team leader at the time Dongping Then Dongping In fact, he is very good at You know sales and selling things Then he thought hey

Then you can find someone to start a business together Then the engineer is responsible for doing it Then he is responsible for doing it Right, sales is actually a good combination So we started a business together This is what happened after you returned to China What happened after we returned to China Someone mentioned Dongping to me That means he should be very important in your company Because he is responsible for overseas Construction of all channels

That's our company Just kidding Yes, I’ll be in charge of product-related matters Then Dongping will be in charge of all the things I don’t care about In fact, we are a very good entrepreneurial partner We have been together for so many years Yes And we actually sit side by side all the time right in the company Sitting in the company Sitting for many years Side by side means Side by side means

It’s because we don’t have our own offices Right, we all sit outside This is the workstation we are talking about and his workstation and mine are which are right next to each other.

Yes The first product you created was Anker's charger?

In fact, the first product we made was a power bank Yes Then it is a It was actually very advanced back then It is a Stick on the back of the phone Very thin It also comes with a power bank with a built-in cable Basically after 10 years You just saw the whole market heading in that direction Could you please talk about the development process of your product This should be your first product

It is also the first hit product This is the first product It should be the first product developed by ourselves But objectively speaking It is not particularly popular Because of that The form of this product is quite advanced.

Back then, all power banks were cordless.

Well, although we have made it But our marketing capability back then was to bring the real of a product to users.

The ability for users to really understand it and like it is relatively weak So even if the wine is good, the alley is too deep.

That product was not particularly popular in the end Later, of course we built a R&D team We continued to launch one product after another Each product is made Yes Which is the first hot product in your eyes?

The first one was particularly explosive.

It’s actually a power bank we call lipstick.

It's made of a cylindrical battery core The size is about the same as a lipstick That product is very convenient to put in the bag and then it can replenish about half of the battery of the mobile phone so it has a long life cycle It should have sold millions of units That must have been 2013 Two years after starting the business in 2013 Two years after starting the business

Your company’s revenue and profits in those two years Are they pretty good?

In fact, when starting from scratch The growth rate must be very fast I don’t actually have much I don’t remember the numbers that clearly In my impression It should definitely be over 100 million in revenue in 2012 Oh Dongping and I have chosen this path.

In fact, the certainty is quite strong You can see the demand on the left You can find it with certainty on the right Or include customized good supplies this one in the middle this bridge There is no difficulty in setting it up by yourself Right So the certainty should be quite high So you are a team that can easily find business opportunities right?

I think it should be said that we are a very pragmatic team We stand on our own abilities every time Move forward to achieve that You should be able to achieve that step In fact, we always say in the company that we are the ones who have stepped firmly on each step.

I will definitely work hard A company that can reach the next level So back to your original question It seems that we have gone from power banks to today’s chips robots This is actually a level you can play steadily After stepping firmly I will definitely reach the next level It is a state that it will definitely reach At the beginning of your business Have you considered financing?

In fact, our cash flow was positive from the beginning Because in fact, every cycle can actually produce good profits So we actually have no real need for financing Our first financing was in 2017 Then there will be financing It’s the same as DeepSeek’s financing today It doesn’t necessarily require money Because we started with e-commerce Every cycle is the process from procurement to sales In fact, both have pretty good gross profits

So the company’s entire cash flow has always been positive We actually don’t need financing By 2017, we raised money for the first time In fact, it was not because we needed funds but because we needed it Because a lot of shares were issued to employees at that time How much do you think this thing is worth If you don’t know, let’s just say hey Find the leading VC at the time to invest some money in us

It's equivalent to setting a price In this case The shares in your hands seem to have a clear value What are the benefits to you?

The advantage of pricing is that students who get the shares can truly believe it It seems that this share is worth so much money because you told him that this share is worth so much money He always feels Do you think it's reliable?

right But today when I am a VC the head VC was IDG at that time Real money When you spend so much money to buy this share Then tell him again You see the shares I gave you are worth so much money It’s easier for him to believe it This is in line with DeepSeek’s financing strategy today In fact, it’s the same I don’t think so Because does DeepSeek need money?

Actually, I don’t know much Because large model companies Logically, they need to raise a lot of money Yes So I just felt that what I said was not quite right Oh, it’s not just I think DeepSeek, a large model company should need a lot of money for training Because what if we go to this time last year?

That pricing may actually be necessary because there is no complete pricing on the market right?

But at the beginning of the year, after they were all launched like Zhipu, Then I think DeepSeek’s classmates You can also refer to the price of Zhipu Let’s price the shares in our hands So I don’t think the demand is that strong But At that time, you were that was completely when you didn’t have such a high demand for funds.

came out for financing and pricing There was no comparable frame of reference for our No. 1 company at that time There was no mention of this company that seemed to be of the same type as us This company has raised so much money In fact, we were serious about making products in e-commerce There is no such company Because the company doing e-commerce at that time Everyone keeps thinking about buying and selling right As for product companies

In fact, there were very few at that time Yes, so we really need It seems that you have set a price and then you also launched the listing Was this already decided when you raised funds?

In fact, our listing is very clear We just hope that employees can realize their shares Yes Then I actually chose China as the place of listing I also thought that in China Maybe employees can cash out Their stocks can earn more If you take your company’s 15-year development history If divided into certain nodes How would you divide it I think the first ten years are a stage The first ten years

We went from one category to many categories From one channel to many channels From one country to many countries It is actually a very natural An extended process First we will divide all user groups into four groups we call them a series of groups Three series people, five series people and seven series people The first series people are in this category They are only willing to spend the minimum amount of money Buy something that basically works

Just the cheapest thing What about the third series?

He wants to buy one with the best value for money As for the fifth series, he hopes good and stable performance Then the quality is very good And I am willing to pay a certain premium What are the last words?

Seven Series is all about the most expensive and best things Let me give you an example For example, let’s talk about milk tea Right The cheapest milk tea is Mixue Bingcheng for 6 yuan a cup This is a series right and the one with the best value for money Maybe a little bit of 10 yuan or 12 yuan Then to what we call very stable For example, fifteen or six yuan Like this aunt in Shanghai

Finally, I went to the Heytea with the best performance Maybe 20 yuan Let’s take another category as an example.

For example, let's talk about cars The cheapest four-wheeled vehicle is the QQ that costs 40,000 yuan Right So The first series of cars costs 40,000 yuan The money is OK Then the most expensive and high-volume cars Basically Mercedes-Benz, BMW and Audi The average price is about 400,000 So you find that the first series is 40,000 yuan The seventh series is 400,000 Right The middle one Back to the era of oil trucks You found

BYD was around 100,000 Our Volkswagen Toyota basically costs 150,000 to 60,000 yuan Many times they use the same engine The same engine from the joint venture factory Installed in BYD and sold for 100,000 Sold for 160,000 to 70,000 yuan in the mass market I would like to ask if anyone is buying it Right?

Of course, you want to buy BYD?

What do you think of people who buy Volkswagen?

Right and the other way around What do people who want to buy Volkswagen think of people who want to buy BYD So you will find it very interesting.

In fact when looking at the abstract across categories the crowd is depending on his consumption habits can be divided into these four groups of people right 1357 So this is our company in the first ten years You will think that what we are doing is a series of brands or products I think many of your products are on the high-end side Mid-to-high-end It was only in recent years that I gave you such an impression.

What about the beginning Actually, the first ten years Many people would say Anke is an overseas Xiaomi Right If that’s the case Then we are one Because everyone must understand Xiaomi It is famous for its cost-effectiveness So it is actually a three-series brand At least that was the case before Mr. Lei wanted to go high-end Right That So many people will say When we were Xiaomi overseas Let’s just say that Anker is actually a three-series brand

But students who really buy Anker things overseas will know Anker’s stuff is actually not cheap right It’s actually often It may be 20% more expensive than the average price on the market or even 30% There is a very good quality behind this.

Just look at the star rating of Anker's products online On a scale of one to five stars We usually have 4.5 stars Even higher So In fact, we were a five-series company in the first ten years We make our products of very good quality very stable and then have a certain premium but not too much Expand to many categories Created a company like this This is also related to the channel through which we were first born.

is an inevitable result Because Amazon You will know when you buy things on Amazon In fact You will definitely choose the thing with the best star rating to buy.

Right, let’s just say for a brand Of course, one option you have is to swipe the screen and swipe orders.

Brush yourself to Then it’s not just about brushing to 4.5 You may have reached five stars Right But Anke actually Dongping and I have always been one At once We feel that we want to do things for a long time Otherwise we might as well go back to the company to work So we have to do things for a long time We feel that we can’t do those things that are cutting corners are illegal things

Things that cross the line So we never swipe the screen or place orders When you don’t swipe the screen or place orders If you want this product to achieve 4.5 stars In fact you are forcing your product manager and R&D team Make the quality of this product extremely good So you can see that Anker’s products are of very good quality.

There is a slight premium This is Anker in the first ten years This is our first stage I have a question here Because when people used to talk about Anker they would say that it is not well-known in China But it is a company with a high reputation overseas Is this a subjective choice on your part In which year was this a choice?

This is definitely a subjective choice because when we first chose Amazon as a channel So countries like the United States, Japan European countries are actually the most developed Amazon in China should have shut down right?

So when we grow from this brand from this channel In fact, our countries will grow faster Have you ever considered going to China In fact, we were the first We haven’t done it for a long time Until 18 or 19 years At that time, many people purchased our products from overseas.

Go back to China to sell on Taobao We feel like this shouldn’t happen right?

So we only started working on the Chinese market in 1920 Why so late Many people may think Do you think the Chinese market is too complicated So I won’t do it I’ll do it overseas I just do it overseas quietly In fact, China’s consumption upgrade did not start until 17 or 18 years ago Because it’s very simple Then you go to Taobao and buy a power bank Maybe at the earliest

The average price may be thirty or fifty yuan Until 17 or 18 years later You see everyone’s The pursuit of better products will become more and more emerge more and more Then at this time, a relatively five-series product The market has only just begun to exist in China So you actually made a product Positioned as a five-series product But it may not match the Chinese market at that time

The first step is to match the European, American and Japanese markets So maybe this is also a subjective choice In fact, these positionings are analyzed after the fact They are put forward when summarizing At that time, you only know that you want to make the product Do it well Then you find a good product on Amazon It’s actually a five-series product This is what you said

You have a lot to do with the channels you grew up in Your first choice was Amazon.

Hey why This looks like a blue ocean Does it count?

In that era Why did you discover this business opportunity Others did not discover it That point must be a very blue blue ocean Yes Then I buy things on Amazon all day long Because I am a special person who is too lazy to go to offline stores so In fact, I basically buy everything online When I first went to the United States to study for my Ph.D.

It was 2003 You basically go to various websites to buy things But after a few years You will find that you will basically only buy things on Amazon because of the richness of choices Price quality guarantee and the best timeliness So Then when you buy something you will find It seems that you can't buy the best thing for you Choose what you want to buy the most

Then I do it myself This is looking back You said we thought it over carefully Then there is the vision of today’s business There is today Strategic thinking Then I chose this category Then I think I have today’s understanding I will most likely not choose this category Why I don’t like it It’s not true Because you find it fun This world is particularly fair That’s what I just said

We actually didn’t have any financing difficulties before Right, and things went smoothly In fact, we have had quite a painful life in the past few years.

How many years have you been smooth?

Until 2022 I have been smooth for 11 years Wow It’s 11 years and I’m not short of money And the product is pretty smooth too Then it was also launched Targeted to be launched in 2020 Then I always felt that it would be possible Explore and do better During those 10 years when you were very successful How do you choose a category or product Do you want to enter or not

In fact, the extension of our categories is quite natural.

In the earliest days, you made charging-related products Then after you made power banks you found that everyone also lacked chargers.

Lack of charging cable So start this line Then follow the smartphone You discovered In fact, everyone still lacks For example, headphones Speakers lacking Bluetooth At that time, there were actually protective cases But we were not very successful in that category.

Yes And then in 2016 you will find In fact, at that time, everyone said it was the Internet of Things That is, all products today should be equipped with IoT So it is natural You are in the category of IoT Have you ever thought about making mobile phones or cars?

In fact, we subconsciously feel that Those particularly large categories We shouldn't touch it By 2020 The first time we really do strategy When we think systematically We realized that behind this intuition is actually supported by a logical system.

We call them super categories and small and medium categories They are also called deep seas and shallow seas So actually open up an industry and look at it You will find that very large category inside we call it a super category Right In consumer electronics Mobile phones are worth 500 billion US dollars PC 200 billion TV 100 billion This is the first, second and third place The fourth place is the tablet Tablets are only $60 billion

So you see From 500 billion for the first place to 60 billion for the fourth place In fact, the scale has deteriorated 10 times more This is the deep sea we are talking about Then the shallow sea Today, there are actually under 50 billion US dollars You can count one to two hundred categories That's what we're talking about today Just like power banks are a 5 billion dollar category Charging cables are 5 billion dollars

Then step out For example, headphones may be 20 billion to 30 billion and then to today For example, consumer drones What category do you think it is It should also belong to Qianhai Consumer drones Global annual sales Probably less than 5 billion US dollars Right, so it's actually included in our definition It's a shallow sea Many people have misunderstandings about this point of view People may think that shallow means simplicity

In fact, our shallowness represents its small size Because the market size determines the category In fact, its saturation investment In other words, you have to make good products in this category Approximately how many people will need to invest Then in such a relatively shallow relatively small category You find that actually its saturation investment will not be too large.

Vice versa Today you find that whether you are making a tablet a TV, a mobile phone or a car Then at least you At least you are a team of several thousand people R&D team Only by doing this category well Yes Actually we have an analogy It’s like a Texas Hold’em poker table It’s like every category If it’s a Texas Hold’em table That so-called shallow sea It's the blind

In categories with millions of bets In that case Just play the whole game Your win or loss may be in the hundreds of millions It looks like billions This is what we call the shallow sea It’s a small category Then the super category we just talked about It's the kind where a blind bet might be 500 million So you go to the table with 5 billion to fight Then maybe you play three games of cards

There will be no more five-game cards This is what we call the deep-sea category Yes In fact, in the past ten years the earliest ten years, you have been moving forward based on your intuition but you have never reviewed it.

Standing until 2020 starts There is one for the past Periodic review and summary Then I summarized the Qianhai strategy In fact, the first ten years This is what I think when I look back as a CEO What I didn’t do well I didn’t say it very seriously Systematically involving people in the company to think together Let’s talk about how many categories we do Why we do so many categories Then which categories we do

Why do I do which categories And then there are more It seems that I am moving forward based on my intuition.

Our colleagues are also moving forward based on intuition This is 2020 What kind of company do you think Anker was at that time?

Let’s go back to the logical framework we have built over the past 20 years.

There are four super categories In consumer electronics Then there are many small and medium-sized categories Then you see 20 years In fact, we have done about a dozen small and medium-sized categories OK Then we have What we do is a five-series It should be said that in this category we have Quite a few were making fifth-series products And a very small number were making seven-series products

Our security products were very expensive in North America at that time.

And the sales are pretty good So the security is the seventh series Then there are some categories In fact, we are essentially a three-series brand So we actually make more than a dozen categories and then focus on the five-series One category is seven and several categories are three So it is actually a relatively divergent company.

It is not focused and does not have a clear idea of ​​where it is going and why it is going there Do you think the company did this based on intuition in the first 10 years?

What does it have to do with your personality?

Let’s go back to the CEO personally Actually, it’s back to value creation And actually in the first 10 years it was still On the one hand, I am happier I feel In fact, we are actually creating value for our customers.

But on the other hand Actually Starting in 2016 Because I went to Hupan to study business school in 2016 Second term students I was asked at that time and said Why Will your category become a 100 billion category Then my answer is actually very scary That’s what I said about my category If you only look at the category of power banks Not only will it definitely not become a category worth hundreds of billions

It will most likely be dead in a few years.

Because from that time you start to look at things abstractly and you discover Consumer electronics is actually a product that comes very quickly An industry that is disappearing very quickly Live and die quickly, you say?

For example you have bought MP3 right?

And if you are like me you have probably bought a tape player or a CD player these products start with you Until you stop buying it The span in between is only ten years Yes So we thought that power bank thing We started buying it in 2012 We thought at that time that might only have a life cycle of 10 to 15 years which would be between 22 or 27 years No one would buy it

So it seems a bit over-thought afterwards The power bank hasn’t arrived yet that Just disappear dead node But we believe that the power bank category will definitely disappear Yes So this creates some kind of fear in you So for when you start asking today When it comes to strategy you will be afraid of your own category Then if In fact, there were a few years in between

and you would think Then how can we do something In fact, it is a category that will never die.

Is there any I actually saw a category just now Actually, it’s electronic cigarettes Oh yes Because actually you will understand It’s nicotine I’m big I think there is a high probability that it is a need that human beings will never disappear right And then the hundreds of years before Nicotine is ingested through combustion Then Switch from combustion to electric heating to ingest nicotine We feel that this change is likely to happen and after it happens

There is a high probability that it will not change again So from a logical perspective Do you think this category will grow and stabilize?

It shouldn’t change anymore It won’t disappear either But later I found out This category is actually a heavily regulated category It’s actually not really suitable It’s just for startups to do it right It should be said There are still categories that won’t disappear.

For example, today you said that air conditioners will disappear?

Do you think washing machines will disappear?

But these categories are actually still or that point in the long run. In fact, that’s right, someone is already doing it.

We actually find the answer to this question Actually it was in 2019 I saw it at that time Books about IPD and Huawei That is to say, talk about How to improve your ability to succeed in a category can put it Systematize and through each organization Able to do it systematically Good More categories So in other words

It is to precipitate people’s abilities into the organization Then this organization can produce more categories Start promoting it in the company We are organizational change But at that point we The company has several business divisions In fact, the head of the business department Make product decisions Then we are under the business department An organization that has developed a product line Then go back to the sales area The organization that has grown a business line In short

It is to transform this company from a small number of people making decisions to today's more clearly divided decision-making units one by one Then give these leaders Greater decision-making authority It should be said this way Let’s break down the company into more detailed ones That is, we talk about product lines and specific business lines Then let these leaders make 98% or more of the decisions themselves

Before that, you alone were in charge of the product?

No In fact, we have been divided into business units very early In fact, we are the heads of several business units In charge of products You are not in charge of products in the first line, right I am not in charge of products in front line We will start making changes in 2020 It’s not just about starting to produce It’s not just about product lines In fact, it is to let the company's power

or let the company's business really run in a system As far as you understand After there is a business department it seems to be the person in charge of the business department say how to do it It seems to be the way to do it The biggest change here is actually not that It seems that a layer of product lines has been added under the business department Instead, let the entire company operate in a

What we call a systematic way Yes Started to propose in 20 years shallow sea strategy and made changes to the company's organization This is a very critical time for you But you just mentioned It’s just that the first 11 years were very smooth but then it became unsatisfactory in 22 years So you think The set of strategies and organizations proposed at that time Is it correct? Looking at it today It seems to be correct today

But I was very doubtful in 2022 Or in other words, very desperate Why did we encounter those crises in 2022?

Because from the perspective of the phenomenon is the product line we built at that time So in real business competition It was you who discovered Can't compete with other industries Those unicorn companies Yes So at that time, people would ask Why is it like a product line of the company?

Being able to defeat an outside unicorn did encounter a very systematic failure at that time so it created a very depressing Even a feeling of despair But then you understand In fact It’s not that you have a process-based system today Then you have a set of formations It seems that your product line can win this battle In fact If your company does not have a clear mission

Vision and values No These are what they are Why this company is what it is Then how to make it like this today and how these are reflected on people Then it is your best process method In the end, there is no way to win the battle.

With process method With values Can you win?

This is a question of the company’s combat effectiveness How to improve the company's combat effectiveness In fact, we say What is a successful company today?

I should put it this way Companies that deal in a few categories And companies that deal in many categories Si's management methods are naturally different Yes Then A company with few categories You find that he may only have a few products a year It doesn’t require too many decisions At this time In fact, you should believe it more A group of very capable people make these decisions without having to say design a system and process

Then let a group of people share these decisions And then the other way around For a company with many categories Then if you make decisions alone or with a few people you will not be able to make this company sustainable in the long term.

So you need to design a system to allow people at different levels to make the decisions they should make.

so i think First, how the company operates It has something to do with the complexity of its categories Yes You are an organization with many categories We are a multi-category organization Because you said at the time that you had 27 product lines You can’t beat others with even 20.

Is this because the product line has been expanded too much?

It must have something to do with Duo But if you say less you will surely win Actually, I found the biggest difference It’s not about more and less I think if it goes down Or do you have core competitiveness today?

So if we look at it today Is there actually any truly leading core technology?

Then if you want to go up, let’s talk about it Does your organization have any specific The mission this group of people want to achieve together And then and the way this group of people behave together.

Can we really achieve this mission?

So you have now solved 22 years of difficulties?

How to solve it It looks like today Let’s look at the results It’s definitely much better than that time What have you done We have been working hard in recent years That’s what we say from the company’s mission and vision Values ​​make changes right How to change Substitution That’s the easiest one But usually it's not That is the purpose So when we talk about the purpose, let’s go back to Make it clear

What exactly does this company want to do?

This company Because back to what we said The first ten years in this company We are actually a five-series company.

Just make it with good quality Then bring some micro-innovative products Back to the second decade When we started doing our own strategic discussions We feel that in the long run we do not want to be a five-series company We just go back to what we said before Let's stand on one level We just want to take the next step We feel that we really need to do a good job in user value

We really need to do things beautifully In fact, you need to do extreme innovation And you have the ability to do extreme innovation on that basis So you have to do several series So we want to be a seven-series company We want to be a seven-series company So we have hope Take another step Yes, it feels like yours It’s like playing a game Maybe I chose it at the beginning It’s not a hard mode game

I chose the easy mode at the beginning Then I played it It’s so good that you want to play the medium difficulty mode And now the medium difficulty is pretty good In fact, we want to play the hard difficulty You also raised a few questions just now It’s you as a CEO How many categories do you think you want to enter Why should you enter this category?

Which categories Which categories to enter Not to enter How did you make these choices today?

To be honest, most categories today They should all have the ability to choose Or before you have a sense of choice It's already been done So it was inherited Let’s talk about a very interesting one This is a true thing Because you know we make UV printers Just on any object Can make a beautiful one A printer with a little bit of 3D A pattern like this

Its nozzle is provided by Epson So in February this year, Epson’s CEO Come and visit our company That’s when the product is introduced He will first introduce the technology He will say You see our Epson has made Romax technology We have done a lot with this technology and then developed it from this technology and derived 1 2 3 categories Yes Then what about when it comes to me

In fact, I will tell him You see we did charging we did audio and video and then we did home automation Then under this direction We probably made these categories So he asked me sincerely He said Yang Sang You are so awesome You do so many categories Hahaha Yes Then what was your reaction at that time I said at that time I said, I said I am very After listening to you, I said

Including combining my understanding today I think I must be very ashamed Right Then if I look back and look at it again I actually think it’s better Or rather, what should I say?

A more ideal way would be Today you have a good underlying technology Then relying on this underlying technology Today You seem to have expanded to one, two or three categories Yes Jack Ma once made such an analogy He said that when you start a business It's a bit like you are on the first floor of a building Your vision What you see is what you see on the first floor Then you slowly start a business

If you do well you should keep improving From the 10th floor 20th floor to the 30th floor Then after reaching a higher floor Your understanding of the entire world Your understanding of what to do should be improved Right What floor are you on now?

I think today is the day when the difficulty of the game has reached One is when playing games For example, I think hard is on the rise But I guess it’s not nightmare yet Yes Which level do you think Tuozhu is on Actually, I think their OK Actually, I think Tuozhu is taking a completely different route from us.

Yes If we choose easy and then go up layer by layer Then they might come up I chose a very difficult mode It's just that they are very difficult in front But the front is very difficult After you get through it it will actually be much easier later These two paths Which path do you think is better in the long run?

I think it has a lot to do with personal character and the characteristics of the times Yes In terms of personal character Just like what I just said You play a game It’s the hardest thing to hit you right from the start.

Or just You may choose to start with the relatively easy ones today Then the other thing is this era In fact, today I think everyone’s understanding Including the tolerance of the entire era Possibility Today we are all going to play a particularly difficult game It will also become easier Yes, so I think The combination of personal character and times In the end, people will produce Different options come out

What kind of person is Tao Ye (Founder and CEO of Tuozhu)?

Actually, I only dealt with Tao Bo (Tao Ye) once.

Not mature enough to express an opinion Yes You have reviewed Tuozhu before as an example You had 27 products at that time 20 of them cannot beat others They are an example Then your conclusion at the time was I think your No. 1 position at the time was more suitable for innovation You just didn’t do enough in terms of execution But I’m thinking Tuozhu is the whole company all in one product

They may also be CEO Engineering Then you have so many product lines Is it true that no matter how you select people you may not be able to beat such a category a company It's not that You have a particularly good No. 1 spot today You will definitely be able to defeat the unicorns outside Yes Finally, let’s go back to today’s topic Your company is in this field Is there any accumulation of good underlying technology?

Then Of course go forward Your commercial capabilities and channel brands It can definitely be reused Yes But it’s right If you are in this category today and you have no technical accumulation at all it will definitely be more difficult.

In 2022, you encountered a series of crises What do you think about the future now?

These explorations of deep technology Are they a solution to it Is there a causal relationship here?

In fact, setbacks in 22 years have contributed to our biggest change In fact, of course we are It is to shrink to fewer categories first.

Then go down Today, dig deep into deep technology and extremely innovative technology Right Another aspect that I just talked about is that we mentioned that the entire company needs to make it clear What kind of products do we want to create and what kind of values ​​do we have Being able to create such a product You said that what changes have occurred in 22 years The one I just talked about is about technology.

right In fact, we are in 2023 A laboratory called 2023 was established Just to do in-depth technology Then you just saw the storage and computing integrated chip we released This is the laboratory One of the achievements of the past few years Right and then another It is the change of mission, vision and values.

It was we who came up with the concept of 1357 It’s 2020 It’s when we seriously start thinking about strategy Then after forming some abstract framework Then I think Okay You see consumers can be divided into these four groups Then what did we do in the past Then what do we want to do in the future?

Then here’s how to do it In fact, you find that from a five-series company to a seven-series company today It’s a very difficult process Yes Then At the most basic level it’s a five-series company everyone can do a great job It’s the first thing today to improve the quality Then the second thing Make some improvements and micro-innovations in the product Yes So what were we doing at that time?

In fact, the company had a our values ​​at that time called being reasonable Pursue excellence and grow together If you seek excellence We said that we had three transcendences at that time It was called surpassing yesterday’s self Surpassing the best peers Go beyond the most essential needs of consumers Then you seem to feel You are on these three steps every day Stepping up Then it feels like you

In fact, you can see that you have reached the second step Then in the next stage, you may reach the third level But in fact, you will find that a company is not essentially a three-level ladder.

It’s the first two levels It’s what we call surpassing yesterday’s self Surpassing the best peers It is actually on a deterministic road moving forward And today When you want to say to go beyond the essential needs of consumers it needs Often requires you to break away from existing tracks and paths Taking a completely different path There may be no one walking this road Some people even say they died on the road

Then you have to dare to walk this road Or you have to see this road first Second Today your entire organization must dare to take this road So this For someone who has been walking for the past ten years Very certain path for companies In fact, such a change in the entire mind It’s not like the strategy has changed Your mind will become that simple In other words, the entire organization everyone’s atmosphere

what everyone believes and what everyone feels Hey The things we can do are all in the previous strategy And today your new strategy requires everyone to embark on a completely different path How could these thousands of people pass by?

In fact, it’s something we’ve been exploring over the past few years.

It's painful but I think I have actually gained some good results today.

one thing What things have you done and what are your standards for the seven series In fact, the seven series must have just been discussed today It’s the best performance and the most innovative thing There are actually some categories It’s actually relatively easy to make the 7-series For example, charging In fact, we were in 20 years The most expensive charger on the market A power bank may cost only fifty or sixty dollars

Then we have to do it When we launched the 7 series, we thought hey Then why today make this product really special and extreme?

Then we are in this field Indeed, I think it is the one with the strongest R&D capabilities.

So when you make such a request to the team the team will make some progress every year Make some progress every year It actually makes The most extreme product in the industry So now our Anker charging we call it Prime This product line is the seven-series product line Then its average price has reached more than 100 US dollars So actually I think it should have been accounted for

A seven-series one is in a stable position for but for many other categories For example, like headphones Then the seventh series has been occupied In other words, it’s a category like charging In fact, there was no one in the Seven Series before Then today we go to the first one to become the seventh series Create seven series Then we have done it in the past few years But there are other categories For example, headphones

In fact, there are already very successful seven series which is Apple’s Airpods series.

Then at this time you have to do to be as good as him or even better than him In fact, it takes a very long time and a very difficult process So this category really has seven series It is to test your first nature and your pursuit of perfection We actually spent several years on this I think by this year by the time this show airs the product we have released

I think I believe it should be done Yes That's right In fact, looking back, I just said I just talked about it In fact, for a company that makes seven-series products Well, we think that in other words For a company that is doing extreme innovation In fact, you have to answer a few questions The first one is you how can you do it

Whatever your company or your people do Only then can we make the ultimate innovation today What about this matter In fact, if you try to abstract it it’s when you look at those who are doing extreme innovation successful companies Like Steve Jobs Like Musk And come back to our company Those who do a good job in extreme innovation Look at their characteristics You will find

In fact, it boils down to the last three points It’s how you think about the problem how do you do things Finally, how do you deal with it?

So how do you think about things In fact, the first thing is that you are not willing to follow what others do Don’t start thinking about what you did in the past But you go back to what is best User value What kind of user is happiest Then you go back to To study the essence of physics is to dismantle such a problem

and then often you just talk about it Starting from the basic principles When building this framework layer by layer Then when solving the problem of user value You will find that there is a missing piece here then this missing piece In fact, what we mean is that in the first nature the direction you found And what about this direction?

Generally speaking, no one has done it Then this time will test you Do you dare to do it Can you take the huge risk?

Try your best to get this thing done This is how you do things What about the last words?

Actually thinking and doing this This is probably related to things You will find that there is actually another layer at the bottom It’s in this difficult process of doing things How you deal with yourself It’s how you deal with yourself Your various emotions, your various difficulties So the fear of the unknown Right, so what's behind this?

We say that actually this layer at this level You need several qualities The first one is long-termism That is, you can see the future situation And then the second one you can bear it Said it was for long-term interests and give up immediate short-term interests So when we add this up, we call it long-termism.

What about other words?

We said there is still one for you today Self-purification In terms of self-awareness and self-purification In our definition, self-awareness is called First see your own part of the problem Self-evolution means solving problems by changing yourself This sounds like a lot of nonsense But you will find that it is actually very difficult to do That’s what happened today After something goes wrong Should I look for the cause within myself first

Or should I look for the cause from outside first?

Then I changed myself to solve this problem Or should I try to change others?

So behind this we feel that we have just talked about self-awareness self-evolution So long-termism Self-awareness and self-evolution And continuous learning This is what we put together We call it mutual growth This is actually a quality that we feel comfortable with Yes In fact, after you read the abstract you will find We feel that this framework is quite stable as far as it can be used to describe as far as creators around the world are concerned Including

Just describe the only way to create Imagine if this is the case That is, if you want to create a something particularly new a new value You must have first-class thinking You must require extreme execution You must be able to be long-term in the final process self-awareness Self-improvement Right, so after we found this framework I felt that was quite essential.

As mentioned before 1357 Talking about super categories Same as the subcategories We tell the truth This is also the company’s own style As we say today, we must be the first priority in doing things When striving for the ultimate Let’s go back to today’s company Then we also hope to be able to achieve excellence based on first principles Explain this company clearly How did you find this framework It’s actually abstract induction That is, you understand today

Then what you saw today I just talked about the successful cases outside and your internal success stories You write down these behaviors Then you abstract it You will find that it is actually like this Then you give it a frame You find that it is these three things How to think about things How to do things and how to behave Because you are not a day one company that does this

you are a company that has been doing this for 11 years Then the company transformed On this transformation road How many people can make the switch?

How many people can’t switch It’s true When you want to start from a company that makes five-series products When turning into a company that makes seven-series products You first just said Explain clearly how to make seven-series products I just talked about the first nature of striving for perfection and growing together.

Then this behavior actually makes it very clear.

Then you can look back and say among your colleagues who work on the fifth series How are they?

Their behavior can be adjusted to Go up here In fact, objectively speaking This is a very difficult thing You think everyone has their own path to success Yes, and everyone thinks they are doing pretty well.

But this is back to Dongping I have my characteristics That is what we have been saying We say that we should step on one step firmly and we will work hard to reach the next step We think that our group of people Including our senior management team Everyone feels We really shouldn’t be satisfied with Only a five-series company we are in this era We are on a level we have reached

When we have the ability to do extreme innovation we want to do it So this intention In the end, I actually think it will lead us In fact, it means experiencing a lot In fact, it is a very difficult process Look at this game Let’s start from easy Hit medium Then hit again I think In fact, innovation We think it is at least a hard model In some cases just mentioned

It’s a category where the strong players of the Seven Series are particularly strong.

It may be a nightmare mode But we feel that that medium is pretty good we should go to hard Instead of just staying in medium and being comfortable Do you usually play games by yourself What is your style?

I actually don’t play games very much Because you think that in fact the entire business world the creation of the entire enterprise It’s the most interesting game Dongping actually always said Talk about your life In fact, your job is actually the game where you seem to have the strongest sense of accomplishment.

I have done a lot of things for you recently I feel very uneasy For example, you have released a chip How was this decision made This should have been made three years ago Let’s get back to the basics This value guided us to make the decision to make chips We are in the audio and video field In fact, we need a good acoustic algorithm for So in 21 We established our acoustic algorithm team Then

we started in the field of sound Actually, at the beginning They are using some small models to solve problems. Because you actually know the sound It’s in the noise reduction process Usually we divide it into several steps The first one is called beamforming Just through a few microphones Then figure out where this sound should come from.

Then the second step is to reduce the background noise The third step may be to do such as echo cancellation and then the fourth step is to do vocal enhancement.

Ten years ago these four steps might have been done which were some mathematical models It is done through some programs Then in 21 years When we started we had already converted it into some deep learning models There are some sub-problems I started to use deep learning models to solve them By 23 years We actually saw ChatGPT

and then you especially believed said The future should be an era of end-to-end problem solving This place is also an abstraction As far as we are concerned In fact, the first 80 years of computers was an era of solving problems using the divide-and-conquer method.

It is to divide a big problem into many small problems and then divide it down and then divide it down For example, when I was searching The search can be divided into many sub-sections In fact, search itself is natural language processing A sub-problem in the middle Then search for the question at the bottom divided into many sub-problems Divide further Divide further Then every question today A group of people went in to solve the problem

So what you find is that actually the entire search There may be a few hundred people inside A team of thousands of people came to solve these problems Yes But that ChatGPT comes out In fact, you can see the process very clearly The future In fact, we should no longer use the divide-and-conquer method to solve problems In fact, we should We talk about moving from the divide-and-conquer method to end-to-end In terms of end-to-end

in essence, you will find that it is actually abstractly viewed which is to first learn from the data.

The second step is to learn this basic model and then strengthen it in the real world which is data learning and reinforcement learning.

We feel that it constitutes the paradigm of future learning So this is something you can see A very clear thing that will happen That is When applied in the field of acoustics means We need to throw away all the program codes and small models of the previous divide-and-conquer method and replace them with a large end-to-end model Of course, this large

may be on a completely different scale than the large of today's language models.

That when this model is going to run When used in wearable devices such as headphones In fact, you will find that our hardware conditions were the best at the time It is possible to run a model with hundreds of K parameters and there will be huge power consumption which means that the model is actually in the headset I can’t run anymore Then you look down and you find out why

In fact, it is essentially because of the sound Such an application It needs to be divided into many time slices per second Then continue to process the sound of each time slice So what about processing each time slice?

You need to take out this frame and encode it into dozens of tokens and then pass it through the model completely All parameters of the type The challenge then becomes to say In fact, every second every frame you need to put all the parameters of the model Move it from the memory to your NPU once And this move itself accounts for the largest power consumption

So our challenge at that time was we can make Several megabytes of performance is actually a significantly better end-to-end model But we found that is inside the chip of the headset but it actually cannot run.

Because of this model hundreds of inference processes per second The parameters in the middle need to be moved constantly After moving too much, the power consumption increases.

So in the end we can only downgrade to one For example, a few hundred K is a 300K or 500K model at this time Its performance will be greatly affected So back to values At this time you know that you want to make a bigger model today Solve this problem end-to-end Then you know Today’s chips cannot run this large model

The reason why it can't run is that it has to keep moving data.

So what is your very natural method?

Or can I stop moving data?

As long as I don’t move this data today In fact, the power consumption of this move is eliminated So it’s very fun It's If you go back and look you will find that it is the architecture of computers in the first 80 years It's called von Neumann architecture Yes Von Neumann architecture is just one sentence It’s called separation of storage and calculation It means you put the program and code in the memory

and do the calculation in the CPU Then you find In fact, it is the von Neumann architecture Ways to Resolve Problems called divide and conquer There are also software files produced by the divide and conquer method called programs and data these three In fact, it forms a particularly stable triangle.

That's what you do when you pass divide and conquer to solve a problem What you write is a piece of code and program and then its data At any point in time you actually only need to run this program a small part in the middle So you will do it when calculating when constructing Separate the calculator and the memory Then put the program and code in the memory Then just put every second

Or that small part to be counted Put it in the CPU to calculate But you find that in the end-to-end era such a stable triangle In fact, it should be broken because the method of solving the problem in the first place is no longer the divide-and-conquer method.

For it has become data learning reinforcement learning end-to-end Then the you produce is actually not hundreds of thousands millions of lines of code.

And it is a neural network And this neural network Of course you are talking about MoE (Mixed Expert Model) it may only activate part of But even if the activated part It also represents millions or even billions of parameters At this time You then use a von Neumann architecture When you find you store your parameters in your memory Then every time you calculate you need to pour all the parameters into your CPU for calculation.

This natural efficiency will be very low To use an analogy it is a bit like For example You store all your knowledge in your left brain And what about the previous words When I just talked about the divide and conquer method Solve the problem every time Probably only a small part of the knowledge code will be used Put that part Just move it to your right brain and calculate it.

It seemed reasonable at the time.

But what about today Because Today your knowledge is no longer Or your method of solving the problem no longer exists It was broken into very fine pieces rather it is a whole It is an end-to-end model When you solve each problem at this time In fact, you need to use the entire left brain Or a large part of the left brain is moved to the right brain to calculate at this time Do you think

it’s starting to become unreasonable a bit inefficient?

So, look at what the human brain does The human brain is one In fact, the human brain is your storage today is all your model parameters in your neurons Then the same neurons actually play the role of calculation So in this case Your brain does not carry any parameters When calculating, it does not carry any parameters So in this case Naturally, it is more efficient So that is not what it is today

We separated from a von Neumann of storage and calculation Convert it into an architecture that integrates storage and calculation An architecture like a brain In fact, the power consumption we are talking about will be significantly reduced.

Then if this is the case We can use it on small devices like headphones Run a model with large parameters How big are the parameters of your model?

Today we actually have less than 2 million parameters.

But for a model that can run inside a headset This is already huge Why put it on the headset?

Is this the first product Or back to today Starting from the pain points of consumers So today we tell you to use the best headphones today Make a call in a noisy place, for example Or for example, if you go to the airport and make a phone call the person on the other side will hear your voice in the airport background Is it broadcast Is it still audible Or when you are in a video conference

Is it often the case that someone comes up?

You think his background noise is noisy Then in the end, his voice was actually unclear So today you will find In essence Picking up sound That's what we said Separate sound from background noise This matter Today’s headphones and then reach the level In fact, it is far from an ideal state There is still a big gap in this category Seven series products defined by you What is its standard

The best 7-series headphones today It should be Airpods Pro Then some people will say Sony or Bose You buy them all the best headphones If you go to test the call effect in a noisy environment you will find that it is actually very challenging.

Just mentioned Background noise Many times it will not be filtered out Then if it must be filtered out you find Then his vocals will become very incomplete.

Not even natural So today How about it It can effectively filter out background noise and at the same time make the voice very clear and natural You must need a large model Let’s do this Right, let’s go back to our own experience That’s what you want us to do since we were first in 23 years We realized that the call was not good enough To solve the problem of the call

we must use what we just said A model with millions of parameters Then it is necessary to run a model like this inside the headset You must have an integrated storage and calculation architecture Yes At this time In fact, just look up and see Do you want to say Who has made such a chip today?

Then you can use it But if you really look at the world there is no one At that time, such a chip was actually made today We have a company that does accounting But today we define the products of that time What we want today is completely different right So actually we have a very natural choice In fact, this matter is at the company level It didn’t go through a lot of repeated discussions.

Because today it is still said that your values ​​have changed Your mission and vision have changed Then you will feel this is what you should do So when our team came up and said Then we need to run a large model When we need to have such a chip Then we will say We make this chip Right, so in fact we found it at that time We just searched the entire market

We found that the company is actually very strong in that field The strongest company We worked with them to customize and develop such a chip Then This process is actually very difficult to be honest Because you think there is no such thing in history So from What does the model look like What requirements does it place on the chip How is the chip defined?

Then how to optimize How to build a tool chain And finally there are all kinds of bugs that need to be solved and really pull it up on the product today so This project started in August 2023 Today is April and May of 2026 May of 2026 Yes Then this product

Products equipped with this chip are finally on the market What about the effect it achieves Because I’ve talked about it so much Everyone says what on earth can you do?

The effect it can achieve is actually very clear Just in any in noisy environment it can be accurate Separate your vocals from ambient noise This feels like a detailed product optimization It feels like building such a big hammer Just made such a small detailed product optimization Is this worth it In fact, in the long run It is actually It’s the first hammer in a series of hammers

But if we look at the details This is actually a very definite and a very painful pain point for users This is what we said from the call just now An effective call angle at any time Because today I believe you may also be many times Make a lot of conference calls So whatever I just talked about today under what circumstances You can make calls with peace of mind This product or this one

The change in value we are talking about Do you think that for you Is it a very, very small thing?

Airpods I use Yes In many cases you may just be able to do it from your perspective Because this is fun Because of this feature It is indeed not a Features that you feel are particularly strong In other words Does your voice follow Or is noise being brought into the conversation In fact, you don’t know In other words Did you still have it in a week today?

Sometimes I get called by the other party Disturbed by the other party's noise It’s just disturbing I think because we have daily meetings Still often encounter this situation Including saying that some people do it specifically for this reason Looking for a quiet environment to hold a meeting We actually just talked about this product Not only can it filter out noise It can also achieve an effect called soft talk That is, you are in your seat

You speak in a small voice Then the other party can hear clearly So soft calls plus extreme noise reduction Behind this In fact, it essentially represents a type of I think the first application of the problem-solving paradigm right So actually the reason why you make chips is also a perception on the user experience side It is a product insight In fact, we have always been one I think it is still a very pragmatic company

It's just that we have always The starting point for everything we do I hope it means This can produce a particularly practical customer value What about these new chips?

Will it be used in other scenarios in the future?

Yes, actually what we are talking about is call noise reduction It is just one of the scenarios where this chip can be applied Yes Essentially When today we can run a large model with millions of parameters Then in an environment with very limited power consumption At once It’s not just about things like call noise reduction it also has many other features Just this chip It is not a chip that we are all familiar with every day

Everyone seems to do it Everyone thinks it’s a matter of course It’s actually a It doesn’t exist on the market It’s the first chip in this field It’s a chip that’s really applied to products And this process is actually what I just said What you have to overcome is not just the difficulties but actually whether this is the right direction?

Can something like this be done There are many It’s an unknown risk But actually, back to the first nature You really want to think clearly today It must be like this Then you have to overcome these difficulties So come all the way In fact, the team Everyone fights monsters We fight many, many monsters Then there are also many Difficult times But in the end, we got through it

And so it was finally installed on the product Everyone can feel it In fact, the experience is very good Then looking back, I feel This process is very worthwhile So far we have seen some models with small parameters For example, Qwen’s 7B model Is it highly intelligent We think In fact, the future You definitely need to do it at many different scales There are different models on it Trillions or even trillions of models

Can be called a super brain So they solve the hardest problems in the cloud Models from billions to tens of billions There is a high probability that it will run on the terminal side So efficient Solve our on-site problems in real time And the final words So the model just mentioned has millions or even tens of millions of parameters It may be running in some perception and control organs In fact, I use the analogy of people

Then if we look at it using a deep neural network the human brain is actually a deep neural network The human cerebellum may be a different order of magnitude scale deep neural network then go down on you For example, your eyes your ears, the sensory organs and then your muscles, the control organs is it In fact, it is all caused by Controlled by deep neural networks one by one Then ask

Your eyes take all the parameters it sees Is it directly transmitted to the brain for processing?

Probably not So your eyes must have a set of optic nerves This optic nerve is responsible for processing visible light signals Then only a small amount of signals are sent to the brain On the other hand Your brain today It directly controls your muscles The movement of every cell?

No and no There is a high probability that your brain is It's actually controlling your muscles a whole movement Then back to your muscles It actually has a group of nerve cells It actually controls the specific movement of the muscles So we think We think it should probably be like this in the future That’s it You should run some on your different sensing organs and control organs related to the complexity of its problems

Or match the complexity of its problem some neural networks For example, what parameters do you think your muscle nerves should be your eye nerves and your ear nerves Approximately how many parameters should Should they be compared to the parameters of your brain's nerves?

One lower Two or more orders of magnitude So these neural networks just talked about Of course there are two schools of theory here One school is that today Today's situation actually means that does not dismantle these nerves individually but today collects the original data of all sensors Directly import it into the large model But then we will think that in the long run, this Is it really the best and most correct way?

Because you find that everyone’s frequency is different Everyone deals with different complexities So we think there is indeed a way in the future In fact, there should be So we feel that there is indeed a possibility in the future It’s these different sensory control organs Controlled by independent neural networks Then their parameter scales are much different So these small models In fact, it should be running in some places

on independent hardware Including actually when we talk about the brain everyone thinks about today the embodied brain In fact, today it is a scale of such as billions of neural parameters Then such a scale today Should I run too?

What about the chip that integrates storage and calculation?

Yes It is a chip that integrates storage and calculation You also hope that it can be used by embodied intelligent brains Really The amount of parameters we have today definitely can’t escape the brain Today I can only run a little bit about the nervous system But I believe that if we go forward on this road we will definitely reach this chip A stage that can carry a brain with billions or even tens of billions of parameters

So what is your plan for the future?

A chip that integrates storage and calculation So from our perspective we will still do a good job in perception first As for what we just talked about In fact, our whether it is the perception of sound or video So today we will make these models Make these corresponding chips Then of course We also have our own exploration of embodied business In the field of embodiment We will also explore larger models Matches today's larger models

Integrated storage and calculation chip The chip is for your own use only Or everyone can use it It's for our own use Today you put your chip first on this chip In fact Today it is completely customized from our model These operators and its model structures are all adapted to our model.

In other words it is our own little CUDA Then I will give this chip to others today.

It means that actually I want to give the model to others In that case I might become a chip company I'll do it right Companies that sell chips outside But I actually have a point of view on this That is, if you look at Apple Look at Huawei Their best chips Apple’s A series M series chips Huawei's Kirin series chips are in his own hardware equipment Are they all used by him?

Yes Because actually from this closed-loop perspective it can generate the best user value Because your own equipment Able to get the best performance out of this chip It can generate the best business value In other words As far as this chip team is concerned If you sell this chip the commercial value that can be generated is different from today You put it in your own device products

This product can be sold more expensively and the value generated is that the latter is much greater than the former Yes But the integrated storage and calculation chip includes NVIDIA Including domestic GPU companies, they will definitely join in First of all, we still have to divide it into the training side and the inference side Today I think the inference side is more suitable for storage and calculation So different models The parameter scale just mentioned

Today’s trillion-level model scale It should be said that the savings are still very difficult to achieve So we will start with one million today and then in the next stage To a parameter scale of billions or tens of billions I think this stage of computing is something I think can be seen in the next few years Yes I think the future of reasoning is indeed calculated

should be a very competitive option So how big is your model?

Our chip is actually a total of 4 trillion parameters.

Then it is actually divided into I just talked about several different types of models Then the maximum size of a single model is about two trillion.

I believe that the end-to-end side is definitely the future But how long will it take for this to arrive Will it be a main trend in 26 years?

Let’s talk about today When can you put a model that generates value running on the side then so when you Can efficiently run the model on the client side and generate good customer value At this time, the client side will explode So so First What kind of models can you run today Second What kind of models can you create today?

customer value I think in terms of expectations for the long-term vision I think is We call this trillion-level super brain It is a super brain with trillions of parameters We feel that if today the storage and calculation are really done very well I have the opportunity to be in the family At a cost of several thousand dollars To deploy this trillion-level super brain Deploy a

Then in each embodied robot Including some complex devices may need to be deployed billions of parameters We say the brain and finally in various devices There will be a large number of neurons we call which are perception neurons or control neurons.

Yeah, it's actually fun We always talk about smart home smart home It seems like our stuff is smart Then For example For example, there are more than a dozen buttons on your toilet It's smart Not smart So you find it fun we'll say that's the function Yes function or it is called adjustable section It is essentially the so-called intelligence In fact, it’s called adjustable Including today You have installed a smart home system in your home

Is it smart?

It's not smart. In fact, to a large extent it means whatever you set it to be like in advance it will always be like that.

So in fact, today’s so-called smart home Essentially, it’s called adjustable and preset furniture.

So we say In fact, all products can be divided into three stages What about the first stage?

We call it the non-adjustable stage For example, take your toilet as an example There is actually nothing that can be adjusted And what about the second stage?

It just grows a dozen buttons Suddenly became adjustable But it's not smart The third stage It should actually be able to adjust itself Then it can sense What you are doing Then adjust it Let me change the analogy For example, a stool A non-adjustable stool This is the adjustable stool we made today like an ergonomic chair There may be more than a dozen places available for adjustment today Then you find

First, you are not very good at adjusting Secondly You don’t even think about adjusting it Wait until your back hurts In fact You may think of this So this is what we call the adjustable stage And what should a truly smart stool look like?

it should be perceptible For example, you come here It retreats by itself and then lets you stand in front of it It pushes itself forward Make you interesting Including saying that it senses that you are playing games It puts the backrest back so you can lie down Feel When you are in a meeting, move your seat forward So it should be able to sense behind this know your status and adjust yourself

So this is what we think what the home should look like in the future is truly smart Then you asked that's why It seems It’s 2026 Our homes are not really smart yet The model is not powerful enough Not smart enough There is no model at all on them Yes You actually want to achieve true intelligence It requires three abilities In fact, it is perception planning control It's you who makes it smart

it must have the ability to perceive it must be able to plan itself and it must be able to control itself So today actually And there needs to be hardware underneath this Right is the perceptron Then a better controller right Then there needs to be a model on it What about software and these things You actually find that they are all in a process of gradual maturity Yes In fact, the earliest time

started to mature with self-driving cars The perceptual planning control we just talked about is typically the three-step process of autonomous driving Yes So after I matured from the inside of the car I think now is actually the time to overflow This is the ability The sensors below it The model capabilities above should inform the wide variety of products in our homes Remove the overflow Then this way After these products are remade

It becomes a truly smart product So can all consumer products be remade It’s this kind of home product It should be said that all non-adjustable or consumer products that are adjustable but require human adjustment can be turned into Really adjust your own consumer goods So this requires a device-side model This requires you to have the ability to perceive The ability to control The ability to perceive planning and control Right Then

these capabilities should be implemented one by one or an end-to-end model So you need capabilities You need models Then you also need to run these models today Efficient chips So this will make you reorganize your company’s products Products in the AI ​​era I think This will first make us rethink this company’s It’s our underlying capability What should our core technology be?

In other words, you understand That's the answer Actually, I think it was the story I just told It was when we met with the CEO of Epson Actually we said This company seems to have a history Not driven by a core technology grown out Yes But today I feel We clearly see such a core technology Just say it today able to perceive This whole set of planning and control

From chip to model Then to a set of technology stack for application When it is particularly polished and mature When it is particularly effective This is what becomes the core technology of this company Then why don’t other companies have it I think logically everyone can have it But I just mentioned it Starting from the chip to the model You find that there are still a lot of things to overcome It is the first cognitive challenge

Plus today you need to work very hard to make it this headset Is it your first product for the AI ​​era You may not know much about it Our security system In fact, it is the second largest security system in the world And if you look at the high-end market it is security systems that cost more than 500 US dollars.

We probably have more than half market share So security is actually equipped with a local large model that is, we built a model of less than 2B On top of our security base station So it processes all your videos locally In this case Your videos don’t have to leave your home Then what You don’t have to look for your videos You can ask it directly through text For example, you can ask it and say hey

Is my daughter back?

You can ask him where our cats and dogs are These questions So it is actually the application of AI It really falls on your family In daily applications Then Actually, let’s do the math In fact, our company’s There are nearly 300 AI-related engineers Two hundred sixty and seventy So actually this is not ours Right so When did we start recruiting this?

In fact, we probably started in 2017 When we started doing security In fact, we definitely need AI for security In other words You must need AI to detect people and motion detection Just 17 years ago We may have several AI engineers Today we have more than 200 AI engineers Then I remember Ying Qi (Chairman of Qianli Technology and Chairman of Step Star) told me It’s the last generation of artificial intelligence companies

Everyone tried to fight the battle of security Then in the country It was defeated miserably by the domestic security giants.

How did you develop this market overseas What we do is consumer-grade security In other words What we do is either like Hikvision Dahua or industrial grade What about consumer products?

In fact, there are also giants Amazon acquired a company called Ring Then They rely on Amazon’s cloud computing capabilities power and price they did It’s actually a three-series and the price is actually not expensive Then A security system where all data is uploaded to the cloud What about us First of all, because we are also As a Chinese company

we think we still need to have some sensitive awareness Including we think we need to differentiate ourselves from Ring So the security we do is a security that cannot be called cloud Then We place a base station in the user's home What about this base station On the one hand, it has long-distance WiFi connection capabilities As long as you install the camera far away it can still send the data back Second

This base station has storage today and can store the data locally The third one is, of course, that it does AI analysis locally Then I’ll give you a reminder or something like that So we start with Such a category was launched in 2018 Until today It should be said that At least from the perspective of market results Still successful This is a good market choice Let’s go back to the end User value That’s what I think

We saw it today Clear gap in user value and through products and technologies Today I realized this user value belongs to on this product This product itself is an existing product plus AI It is not an AI native product Do you currently have AI native products?

In fact, I can make an analogy It means going back ten years Really exactly ten years ago Then the Internet of Things was just emerging Everyone thought You see It seems like all the hardware should be redone So it seems that you have Internet of Things technology The hardware products you make are for the Internet of Things which are what we call smart home products Then you are a traditional manufacturer

You seem to have been born before this era It seems that you are also working in this category You are just not right You are relatively backward But in fact, what will happen in the next ten years In fact, you find that those who are in the same category There are very solid scenes.

A company that can evolve And the last ones are companies that have AI technology Then they must enter this category Then after the PK between the two sides Your perception Which side wins more in the end This side In the end, companies with scenes win more Yes, so I think there will be such a situation this year Observations on the past Come back and look at today’s so-called AI era AI native hardware

What do you think of this generation?

Will it be very different from the generation ten years ago How do you understand AI What do you think?

What changes will AI bring to the consumer electronics industry?

and or change what would it be like In fact, customers don’t care about AI Customers care about how they can have a better experience Let’s go back to what we just talked about is In fact, the clear path we defined is AI empowers your various devices today It seems like you need to adjust it Today it can really adapt to you It knows what you are doing It will adjust itself

Then you can use it almost without any feeling today.

It has become a living thing It is a service body It is a service body It is a service body But its existence is not as if it has its own life but in fact Of course I think there should be a distinction here If you talk about embodied robots Including robot dogs It will not have its own independent will in the future This is one thing But today you said that

is a chair that can adjust itself Do you think it is a living thing Maybe that is a bit exaggerated A bit too much A bit too much Oh So you will add AI to the existing hardware I can add models It’s not that I can I think this world will In fact, this world will make all those For example Like your microwave There are also a dozen buttons on your microwave

Then you think you will use the middle ones One or two Just one or two are commonly used I'll just use that knob for that minute number that knob except that one I don't know how to use anything else.

But what do you want Microwave ovens like today Should it be transformed into It senses what you put in it It can give you suggestions and after that it even says it can give you voice interaction so all these I think These things in the family are the things that I just mentioned that need to be adjusted by you In fact, they should all become able to adapt to your needs.

Adjust yourself So this is what I think The real value we will bring to consumers So the so-called AI native is not the first principle Is it Is the Internet of Things the first principle?

Did you find out ten years ago Everyone says the Internet of Things is coming Everything had to be done over again Everyone was very excited I thought it would bring about a big change But in the end I think In fact, consumers really don’t care about your Internet of Things They don’t care about your AI What consumers really care about today is How can my experience be better?

Just how I feel more comfortable than before More efficient and happier So today I feel This is what I think too In fact, we have always established the company's foundation That's what we say We must start from Making user value better Changing a category It will definitely need three things OK The first is the technology you want to lead What about the second one?

You need to have a real scene The ability to productize in this scenario It’s not just the ability In fact, what do you think about this product?

understanding of this scene certainly You need to have good global business capabilities So what?

Actually, for us That is We must start from creating value for customers For us We start from creating value for customers We feel that our business capabilities are actually strong In the areas that need to be changed today In fact, we still lack that technology It should be said that we go back to 2021 and 22 When we think about it seriously We feel we still lack technology Then today In this world

there is no perfect company Nothing seems to be ready today All I owe is Dongfeng Then you can make up for what's missing.

So in the past few years In fact, we just talked about Our AI team has grown bigger Build our chip team We think So to this day Of course My main purpose for coming to this show is actually We think it’s We still have a lot to do No one has done it yet So the purpose of coming is actually to tell your audience our vision for the future.

Then we can attract more people to join this company and turn this vision into reality together So we feel that what we just said What kind of people do you still lack We have the ability to commercialize We have scenario-based product capabilities Our technical capabilities are not 0 now I think our current situation may be a few tenths So we are short of good technical talents Let’s come together When we turn our technical capabilities

into a strong board then I think We can really create particularly good customers Value AI chip embodied intelligence How much investment are you willing to make?

this question It's hard to start from the point where it seems like one is the total number answer from the perspective of Or go back That is to say, what is our specific problem today?

Then we have to solve those problems And then today we solve this problem what kind of investment do we need?

Like what I just said It’s just that we most likely won’t train That trillion-level model Then So we won’t make chips of that level.

OK when we are training today what we are talking about From some neurons with millions to millions of parameters to models with tens of millions of parameters including when we are exploring embodiment Embodied field Now everyone basically has billions of models When we train these models Then the we match is the model training resources and the chip resources I think it should probably be It's on the order of several hundred million, right?

And as for AI, actually It’s hard to say how much you’ve invested in AI.

Or even if it must be counted, it can still be counted For example, We have more than 200 engineers I estimate the per capita income is in the hundreds of thousands So this is a lot It’s boring, right Right Tell us about your robot business embodied business Actually that What I just talked about was the chip I just told you about some chips

Say something sensible Then we need robots In fact, we divide robots into three forms OK Two-dimensional planar robots are today’s sweepers and lawn mowers including swimming pool robots It can be understood as a 2.5D robot This is the first stage Then what about the second stage That's today Mobile and interactive robots in three-dimensional space Today, in fact, robot dogs pets belong to this category

The third stage will definitely be the final crown jewel.

It’s the operating robot Then High probability Today everyone thinks it is in a humanoid state So so today Stages 1, 2 and 3 Then what we are actually doing You know we actually have a sweeping robot and lawn mowing robot business We are doing the first phase of the business Then we are developing the second phase of the product We are exploring Preliminary research on the third phase humanoid stuff at the forefront Right

Humanoid robot Will it be a big bet for you Today our team is still in a It’s a state of pre-research and exploration because we feel Today’s technology stack in this field has not converged at all.

if you look What is the state of the brain?

And what exactly does this person’s body look like today In other words, what does his waist joint look like Did you find these things?

In fact, everyone’s solutions are completely different So at this time We feel that may not be a product yet and it is more suitable Do pre-research What about the opposite Today you observe products like dogs It is a second-stage product In fact, the forms of each company are very close So in terms of hardware This technology stack has been relatively converged So if we go back to today's model stage everyone's approach is basically very consistent.

Today the agent runs on a model So we think such a product It should be ready for productization So we started last year We just have a team We are developing a robot dog In fact, dogs are in our lives today What role does it play?

What value does it provide?

Of course today a dog is a pet It provides you with emotional value But when you find that the first time a dog enters the family it actually plays the role of a caregiver.

It provides functional value In fact, you found today’s pet dog to a large extent It has lost its functional value back then.

And we make our own security systems Then we are in millions of seven-series homes around the world One problem we continue to face is that Your home is filled with security cameras Then including alarm equipment But if someone breaks into your home today In fact there is very little you can do So today we think that when you have a guard dog When there is an intruder today and is detected by this security system

this dog can drive away such an intruder Realizing a closed loop of real user value It is not just about finding intruders today but actually being able to drive away intruders Make your home really safe Then such a product Does it sound reasonable?

So in fact Let’s do user research I also found that our seventh-series users Today we have quite a few I am willing to spend a considerable amount of money to buy such a family guard dog In other words In fact, when we started making this product we were sure that it could be sold and it was sold in large quantities.

Because it creates particularly clear customer value It sounds like this era keeps up with the previous era Your methodology for making products There is no big difference Still the ability to define the market The ability to define products I think in any era As long as you call it a product company You must create value for customers It’s impossible to say your product The technology is extremely advanced But customers actually think it has no value

Then this is definitely not a good product company So I think over the years we started from simple products like power bank Until today it is complicated Products like guard dogs The company's belief is very consistent That is, we must do Create products of real value to customers So why does the strategy shift from shallow sea to deep sea Has there been a change of attitude Let’s still say

Today’s depth does not mean how complex the category is today Deepness represents the scale of this category In other words We say that the so-called deep refers to the sales scale of this category which is more than 50 billion US dollars.

This is actually it just like the Texas Hold'em example I just gave In fact, it’s actually your Texas Hold’em table The size of each blind bet So in other words, we say shallow sea The blind bet is several million yuan to tens of millions of people RMB market Then the so-called deep It’s a market where blind bets can range from 500 million to 1 billion.

So in such a market You find that in fact, even if you have billions of money or even tens of billions of money In fact, you may not be able to play a few games before you go down So that kind of market I think it’s actually the same as I think that is the difference between deep sea and shallow sea From this perspective What we are doing today is still a shallow sea market you understand

Even security systems A market like watchdog Its entire market size There is a high probability that it will not exceed 50 billion US dollars Yes So the shallow sea strategy still continues So from the shallow sea we define As I just said in categories where the market size is not that big we will still continue Will you go into the deep sea In fact, you can see it now

Two deep seas that will definitely grow in the future One is the humanoid robot we just talked about right And what about the second one In fact, we think it is smart glasses which is the fourth personal interactive terminal It’s a completely new form Then today human-shaped glasses We feel that there are a lot of companies ahead for smart glasses for smart glasses We think there will be three waves of big manufacturers competing ahead.

The first wave is all the mobile phone manufacturers we are talking about The second wave is all the major Internet companies Then the third wave is all the model companies so you know Because these three waves of people In the end, they all want to grab The entrance to interact with people So when any company enters this category you have to answer How can you beat the one, two, and three in front today?

These three types of major manufacturers Right, so do you enter or not So we will not get smart glasses today Which category?

Yes, but humanoid robots Like we just said It has not converged at all today.

Then we are developing it today This is what I just talked about in the form of a dog.

product time In fact, our technology stack Already extending into that field Then we hope to get to that point It’s the real technology The point where everyone has begun to converge We have the ability to challenge such a market Of course, that challenge There is a high probability that it will not be the case with today’s company.

It seems like a part of the organization does maybe from within the company Establish a more independent team to enter such a super category market It may spin off Yes It’s actually done well That’s why spin off is necessary so In fact, it is a bit like Huawei’s operator business It’s the same as the mobile phone business that grew out It actually turns into a deep sea This business is completely independent and is run by itself.

And then today it is running very successfully Is this something you must do Or are you just waiting and watching?

We feel that as a company doing Qianhai today this is actually not something I 100% have to do.

But today, when the time is right When I have the ability to do it Then we have the right leader When doing this We think we are willing to challenge things That’s it It’s your company Maybe the crown jewel But today it’s actually not sure I think today if you look at Huawei Huawei’s carrier business is essentially a large shallow sea.

together It’s just that it’s a combination of many small and medium-sized categories.

Then its mobile phone business It's actually a huge deep sea So today we feel that in a company has today’s large shallow sea business There is another deep-sea business Huawei is a good example in the past But it’s like them In fact, it requires a very good foundation Then very strong organizational and technical capabilities Then I think we are working hard in that direction today If I could summarize it in more concise language

The concept of this product that you have been working on for more than ten years Product method How would you summarize it?

From the earliest product selection to selecting people to executing it Do you have a complete method I think There is actually a complex system behind this It's hard to explain it so clearly in one sentence If I must say one thing today I think it’s better to look at the segmented user groups.

Create unique value Based on this Yes, if you must say one sentence It’s just about segmenting the crowd.

Create real and unique value for them How carefully do you look at the crowd I think all valuable businesses It all starts with crowd segmentation You go back to being like Steve Jobs Steve Jobs started from When I returned to Apple in 1996 He faces Apple and has dozens of products In fact, he also treated the crowd Divided into two groups That’s what we call

professional users and ordinary users Then he divided into two scenes One is a mobile device One is fixed on the table So after two groups of people and two scenes Four product slots have been cut out.

Then each product Each niche creates a valuable product Then the company becomes very successful So I think In fact, for all people who make products I think this is a You can understand that if we only talk about one principle Today is a definite scene for segmenting people.

Then create unique value Have you ever experienced product failure Then not too many success rate What is the winning rate?

In fact, after we made a contraction in 2022 We have never reduced categories again Have you become more cautious in choosing categories?

Yes indeed That’s how it should be said It’s not just becoming more cautious And back to the end There are actually clearer principles and methods We should start from both ends One end is what kind of people you have now These are the people They have already bought your product Then He has a basic understanding and trust in you today And what about the other end?

It's technology What underlying technology stack do you have today?

Then there are some common technologies There are also some extremely innovative technologies today So the best category for you to enter today is Your overlap among people is very high And the degree of overlap in the technology stack is also very high.

Under such development the success rate of category expansion is very high Right If your crowd is different today too Then the technology is different too Step out with your foot There is a high probability that the challenge is relatively large Are you the kind of genius product manager type person I think my product sense is good Right and then that But to be honest,

consumer electronics products are actually endless The meaning of endless is that ten years ago There are new consumer electronics products There are brand new ones today I think there is a high probability that in ten years There will also be brand new consumer electronics products Then I think I am a long-term person Oh That’s what I always wanted to say What product to make Can this product be used forever Yes And so Actually

I regard this company as my product today Today People who can make many, many young products Technical people On this platform continue to create the products they want to create Such a platform It is also a product in itself Then this product In fact, it is difficult and challenging to do In fact, I understand is far bigger than making a specific product When did you have this awareness?

Realize that what you are doing is a platform rather than a single product When we are doing strategic discussions in 2020 We realized the first thing we just talked about How many categories have we done How many categories will we do in the future?

How much will we do in the future?

Or should we say that we will be in several fields?

So I see you are drawing three areas For us today Charging and energy storage are one area Yes Audio and video is a field Yes Our third field is home automation Yes In fact, we are still working on an area It’s the incubation in the health field.

In the future, we will develop in these four fields I think the number of categories In fact, in the end, each field I think it’s normal to have 10 or 20 categories Yes But such a big vertical category Large categories 10 or 20, right 10 or 20 is a very specific product category For example, in charging and energy storage I count power banks and chargers as two different product categories Oh, how many are there now

We have about 20 now Right So today if you look at the four industry directions Then the last 10 to 20 words in each industry direction at last I think the company may have 40 to 60 categories Yes This is what we just talked about This is what we call the third type of company because we said Look at the categories below

There are a small number of super categories and a large number of small and medium categories There are three types of companies above Category 1 company that is Super category companies that are doing very little today For example, Apple Tesla What about the second category of companies That is, they do a small number of small and medium-sized categories In fact, most companies belong to the second category Category III companies Just like our vision

Today we can produce dozens of small and medium-sized categories and can make it Create good customer value In fact, in the consumer electronics category there have been companies like this in history For example, Sony back then and later Philips Then if you look at it across different industries You will find For example, Procter & Gamble It has many, many categories Many brands of companies Such as Nike Nike

is actually composed of many sub-categories Yes Then it’s like in the electronics category For example, Texas Instruments Then we all feel that actually in every industry you find There will be several successful third-category companies So we hope Anker can become one of the middle ones Become a third-category company Generally need a CEO What kind of character and personality is he?

Not a centralized person In fact, this goes to the third level We just said the bottom layer is Landscape is the terrain that you can't change middle layer Finally, three different forms of companies grew up What about the top floor?

In fact, we think they are two different management models.

One is called the King and the Knights It’s the King Domestic affairs All are decided by the king and his knights The other is called the President and the Federation Today this president is responsible for the overall construction and operation of the entire country Then Then the specific federation underneath One state one state is responsible for the operation of a specific field Right Then we think the third type of company

You found the third type of company that was finally successful in the long term In fact, it all goes to the model of the president and his federation.

Finally returned to today In fact, do you want a very specific to take care of one thing at a time or do you want to step back a little bit?

You want it more In fact, you think the reason behind this Whether it’s because you are lazy Or because of today You don’t want to do anything specific today The character behind this In fact, I think it is the superposition of several factors.

In fact, the first you may not want to continue doing such specific things Right And the second one is very important In fact, it means hey You still hope to empower some people today Empower others Empower many people to do these things Because what we just talked about you see All smart homes are worth redoing Of course you have the choice today to say I will build a brand new company

to re-make a certain category of truly smart devices But what?

IMHO Just start a business in the hardware category Become a global company The challenge is actually very big Let’s open up a company and see its value chain The so-called value chain is Today you have to create this customer value The different positions and roles we have to go through This is what we call Stringing it together is a value chain For example, our hardware

You need to do strategy first Then you need to do product development In the product, you need to do user research Then do research on the cutting-edge of technology Then develop products So actually we talk about consumer electronics Electronic products In fact, it is a company with a very long value chain So the value chain In fact, it is about creating user value today.

Different positions and roles to go through of a chain together the sum of So our company has 6,000 people We have 200 different positions In other words, our value chain If you count its length It is actually as long as 200 different positions That is, we belong to the long value chain And then even longer For example, cars Categories like rockets Then start a business in a company with a long value chain today

What does it mean Is it difficult Difficult What it means In fact, you need to be on your limited take-off runway called Runway.

With your limited time, money and opportunities You need to put these within the value chain Most of them can be polished until they are basically usable.

Doing it in small amounts may be stronger Strong So its difficulty is actually very difficult Then you said that after the value chain becomes longer Its difficulty is a linear increase should not non-linear It is non-linear because in the value chain it does not mean that every point Only communicate with the next point It is actually between a dot and a patch of dots Everyone needs to communicate and cooperate today

So you find that after the value chain grows the complexity of the organization In fact, I think the square level or even higher complexity is rising.

In fact, in this process You need to build a company with a long value chain Really polished You find it It’s actually a very, very difficult thing So you just look at the entrepreneurial success rate of hardware companies you can see The Internet era is the era of smart hardware Just ten years ago In fact, there were many, many invested companies that are really surviving today live well In fact, they can certainly be counted.

So what is facing today After we say that AI is coming So those who want to re-create these products These creators we are talking about actually also face choices Otherwise you are today It’s like I started a company I want to build this value chain from beginning to end Then we hope to provide another option Today On our company’s platform In fact, we have successfully implemented it through many categories

Then we have a mature value chain And there are people who are doing a good job in this value chain In fact, we still have this value chain precipitated intelligence So all these things Today when effectively supporting the process of entrepreneurs then we believe He can be in this category with a higher probability Survive and live well Then in the end we have to solve the distribution problem Because

if you make good products and technologies but it’s not like starting your own business If we can get a lot of returns this logic doesn’t make sense So we still say That is to say, distribute most of the fruits of creation to the creator himself So we think we hope to provide such a in addition to starting our own business Another option is to join a mature platform and then use it in this platform

to create new technologies and then new product species I can understand that you really want these entrepreneurs Creators joining your company But I’m also curious From your perspective What do you think is a good hardware product Because you are not hands-on making products now You are building this platform But from your perspective what kind of product manager is a good product manager Maybe after superimposing AI would it be different

I think a good product manager is first of all a product manager who is dedicated to delivering customer value And it's not a broad customer value It must be one that focuses on a specific group of people The unique value of specific scenarios I think a product manager who insists on doing a good job for specific groups of people The unique value of specific scenarios He is the foundation of a good product manager Yes

Of course, this point To be honest, it is the opposite of Internet products Internet platform products are not selective For example, Taobao For example, Douyin Basically all groups of people I hope it can be applied to all scenarios But in hardware we talk about going to we talk about small and medium-sized hardware It must be about selecting people and scenarios What kind of product manager Interview in front of you You will intuitively feel

He is a very good product manager The kind you will never see again First of all, we have learned many lessons from failures Then the so-called background Then the so-called academic qualifications include even the interview process a lot of processes In fact, I think is not enough to be a valid proof today So what?

Actually, back to us Why are there such things as values?

it is today To test whether a person is creative We think the most effective standard Just today Is your thinking process first-class enough In other words, today From your past career If you used to make products Then what are you different from others today?

First thoughts and what was the starting point at that time What was your derivation process?

Then secondly, strive for perfection That is to say, just thinking is useless How do you overcome huge difficulties?

Then try every means to make it And the last words This process is actually what I just talked about Do we have a long-term perspective?

Are we enough during this process?

Are we constantly aware of ourselves?

and evolve yourself As long as you find out that I keep talking about the same thing because we believe that you will see the best creators They have these behavioral characteristics So when we define our company today as a company that revolves around a creation Then the talents we attract, whether they are product managers or technical talents, even include a business talent Then let's say today

For example, you want to be in a certain country Today we can build this brand from 0 to 1 You find that you also need the first quality The way of thinking You also need to pursue the ultimate execution Finally, you also need The characteristics of growing together that we just talked about So this is not actually meant for product managers.

Or the unique characteristics of technical developers Whenever today when you want to do something from 0 to 1 You need these qualities If these people can start their own business Will he go to Anke?

Start a business in the consumer electronics category We say its value chain is very long A business with a long value chain means In fact, you have to build this value chain Its complexity is very high So in other words Your probability of success is actually very challenging Let’s look at the big data of past history like this Then after the arrival of AI today Will this change?

We think it actually involves people Because there are actually people in the value chain Then we think it’s not that easy Today is a complete change right So we think In the end, if you want to get this done Then we think our may be a better way to accomplish this matter Not to mention Today Anker is already a global A brand that everyone knows about consumers Then we have global channels

This can help commercialize a good product Bringing huge convenience We actually talked about your financing just now You launched a round of financing Then immediately after the listing started the capitalization action What do you think if you look back today?

If you hadn’t done this at that time What would Anker be like today If there was no capitalization You said that if we didn’t raise funds at that time we wouldn’t raise funds we wouldn’t go public I think actually financing and listing Objectively speaking, it gives you an external pressure but i think We are a group of people who are not driven by external pressure.

In other words Dongping and I We have been in business for more than ten years It’s not like saying how about watching others do We have to do what we do We are still us That is really firm Just step on one step firmly and then take the next step So I feel that even if you don’t raise funds, you won’t go public.

We should also reach the path we are on today For you to retain people Will there be any bad effects?

If not listed that's a good question In fact The bonus we just talked about In fact, what we just talked about is that we divide the current profit A large part of it is distributed to employees This is a way of motivation But it is indeed said that after the listing, In other words, through the realization of shares you can distribute the profits for the next 5 years, 10 years or even longer to employees

This is definitely a big attraction So In fact, when we went public China’s regulations are Unlisted companies can have up to 200 shareholders So we probably had more than 170 employees that year Have equity in the company Oh yes So after the listing The income of these employees is actually very good Will they stay after they go public Objectively speaking Actually

Many people do it after they have enough money He will reconsider his life And then but Today I think the ratio is bigger It’s not that he has money or doesn’t have money.

But he felt that such a place Is he today It can also make him continue to grow and be happy Yes We see that there must be some people After his money exceeds a certain amount he may want to choose a different lifestyle In fact, we have a very interesting example We have a classmate from finance and economics Just after the company went public He felt that he had enough money Then he Resigned

went to the Central Academy of Drama to study screenwriting Oh Then he worked as a screenwriter seriously for several years But did the ending of this story come back It came back to him again He discovered that the screenwriting industry is not that fun Yes Finally returned to this company He felt like He seemed to miss us very much Then we were doing more interesting things He finally came back

That’s when you discovered that you might have just graduated Moving forward on the same track Reached a stage This company is listed You feel like hey You give yourself a reason to rest But you find that you can’t rest all the time either You change a field After changing a track Are you happier than now?

Many people have tried it and found that it is not So what I actually want to ask is Now there are many It may be a company that makes AI hardware or a company that makes hardware It will be profitable in the beginning For a profitable company under what circumstances should one consider listing under what circumstances should one consider not listing Is there a standard answer to this From your experience I will tell you two

Different perspectives Let me talk about an employee’s perspective Let’s talk about a founder’s perspective I think from an employee perspective is almost always Just about all the time employees want this company to go public because it’s easy If it is not listed it will be difficult for me to liquidate my shares even if we give it a lot of dividends every year Just divide a lot of money But it all turned into cash after the listing.

The huge income ratio brought by that handful It's much less So from the employee's perspective I think employees almost all the time want to go public What about founders?

I think this question has to be looked at from the perspective of the founder It may be different If you look at history You will find In fact, in Taiwan, China there are many 80-year-old chairman of listed companies They were both companies at their best Maybe 40 years ago They went public when they were 40 And of course the company has survived for decades.

but you find Even this company has not grown bigger And in the long-term process For example, family members are not willing to take over.

Then the colleagues behind couldn’t effectively pick up the matter.

So Although when you were 40 years old you enjoyed the joy of ringing a bell for the listing of the stock market but when you get older when You still have to pay for this company.

Then when the company continues to run I think this matter In fact, he is sure I think for some people the long term is Or rather I think you are the only one who understands what I want to express You’re locked in here For founders, right Because your employees can leave But founder, you most likely won’t be able to leave today.

Your company went public today Then you seemed to be ringing the bell You were very happy But if you want to think about things in 10, 20, 30 years Then you'd better give it to this company One that can develop over the long term a path and a vision Can’t it be left to the second generation In fact today If your company is on a particularly stable road On a smooth road No turning around

Nor will we make today’s AI revolution Maybe today a second-generation or a professional manager can manage this company well.

But if we embark on a path of change today, Regardless of whether your category has changed today Or are you faced with something like AI today?

At a time when the whole society is experiencing a huge wave of What do you think about a second generation and a professional manager?

Can we really do this well Then if we don’t go public we can’t solve this problem.

You still have to work here What are the benefits of not being listed This is a good question Even if it is not listed today this company will still move forward But objectively speaking After not being listed today There are actually more ways for you to exit Yes Just sell it and back off say for example Today you can sell your company to a larger tertiary company like us So you find

Just when we go back to the analogy of Category 1, 2 Category 3 companies you find that a large number of Category 3 companies are growing.

On the one hand, it has created many categories internally On the other hand, it has done so through mergers Through acquisitions In fact, many new categories have been introduced So if you look at Procter & Gamble If you look at Texas Instruments That’s right It’s all like this Hey, have you ever merged with any company?

China’s current environment I don’t think it’s particularly suitable Mergers and Acquisitions Because everyone still thinks that it is a better choice to go public on your own Yes So from the moment you choose to go public you decide to be tied to this ship and you will be together for decades, right?

I think objectively speaking when you choose to go public today in China Under the requirements of today’s current market environment you don’t have just leave this Or you were tied to this ship Did you have any opportunity to make this decision at the time Because your profits were also very high at the time In fact, we thought about it clearly at that time Today Because we can use a systematic approach

Have the opportunity to do well in many categories Then we said at that time that we had a foundation We can turn ourselves into a third-category company when It’s when we systematically prepare many categories.

We think so Then we go public It seems to be a third-category company An intermediate process that is relatively certain Yes Why is DJI not listed on the market?

I think I need to ask Wang Tao himself about this.

Not going public may still have an impact on talents.

Just go back to what I just said Let’s go back to the perspective of employees Employees are almost always I hope the company is listed Yes So we have issued stock options to the first wave of employees Everyone will be very happy and get rich and free Then you have to deal with the employees behind you Then he doesn't have the problem of cashing it out one by one.

So actually we are in our new category I also drew one for the employees.

What we call equity pool Then when this new category passes the investment period today and generates profits then according to the rules of listed companies We can take the shares of these employees Will buy back again Also give it a good return So actually I think I actually took a step back in the end.

It seems that we have been talking about repayment just now But in fact, after so many years your feeling is Your best creators are those who want to create things to the extreme There is a high probability that they It’s not that returns are the first priority But you have to give enough money Yes Then we think for the best talents what they are looking for first is an opportunity Just this opportunity is a

Let him challenge a particularly difficult opportunity for something he has never done before Then in this opportunity He will grow When a company supports him today Someone can help him When someone can discuss with him He will grow Then the third step will be rewarded And I think it is equivalent to the reward There is another meaning It is what you will definitely want to say today I have made money

But when I have dinner with my classmates I also want a place where I can brag Then go back and talk to the children I also want to feel that my parents are awesome This is a meaning So we say In fact, a good company It should be able to provide its creators with what we call opportunities growth Reward and Meaning Then We feel In fact, a company

If only I could provide the creators with Just an opportunity Maybe you need to spend two years It will take three or even five years to make it mature.

Provide us with a chance to grow and be rewarded A meaningful opportunity This is already a very good company But I think the best companies should be able to sustain At least two, three, four, and five times Provide this employee with continuous opportunities for growth Opportunities for meaningful returns In this case In fact, in this three-wheel and five-wheel growth After coming down He will become a very powerful person Then he will be more comprehensive and complete

If he chooses to start a business again at this time I think his success rate will be higher Hey, you just mentioned it The consumer electronics industry is a fast-growing and fast-dying industry.

And it seems that the monopoly effect is relatively small Do you think that if you are doing consumer electronics today there will definitely be too many red oceans compared to when you started doing it?

There are so many competitors in Shenzhen DJI Insta360 Zhenghao Tuozhu etc. So what do you think is the moat of a consumer electronics company today?

What exactly is it I think I’m talking about a three-layered moat Yes The first level is actually your unique leading technology Today this matter can only be done by you At this stage only you can do it When others can’t do it This is definitely an effective moat But I think technology will definitely encounter We call it diminishing marginal benefits What does it mean Because of human perception

Human being is a simulated animal There is an upper limit to a person's perception When your technology reaches the upper limit of this person's perception You find that there is not much room for your technology to move forward At this time, others can catch up.

So your technical advantage may disappear Let me give you a few examples For example, Apple’s screen First more than ten years ago Everyone must think that The screen of iPhone is better than other mobile phones Do you still think so today Don’t you think so today Even Apple itself came out and said There is a standard for screens called Retina Retina means that its pixels are so densely packed

So that a normal person cannot tell the difference In other words People today your eyes When we can only distinguish such a fine particle size When today my technology reaches this resolution I will go further You can no longer sense it Let’s give another example What we have today is like a charger We make it small and then very powerful We put a 140-watt charger

Make it really as big as a small card You think it’s small You think it’s great When I make it twice as small you may think it still makes sense but make it even smaller Do you think it still makes sense?

It doesn't make much sense anymore So we will say When you understand technology today it will always encounter the limit of human perception When you encounter the limit you find that there is not much room for you to go higher and at this time others will gradually catch up.

So you finally understand your excess technical advantage Put it in for a long enough time It will definitely be wiped out As far as we think This is basically a first principle Okay then When you say today that your technology has been wiped out then what is your moat at this time?

What Brand It must be a brand It’s the trust you have in the hearts of users And that part when I think of your brand that emotion In fact it is very, very difficult to gain trust in a brand Consumers may need many times to build trust in your brand but you may do it wrong once or twice This trust is gone forever.

And he may never come back in this life So today you found be a good company It takes a long time to accumulate and then build trust with users Then this will constitute a huge The second layer of the moat But actually, if you look at it over the longest period of time In fact, I think both technology and brand Neither is the company’s deepest moat The company’s deepest moat is its mission

Vision and values and a group of people who truly believe in the mission and vision values ​​​​person why do you say that Because you see that there are many companies whose brands are based on trust but gradually it doesn’t seem to work.

Nokia Nokia example Even today I don't know if we should talk here It’s a brand like Sony In fact, at a certain point in time Actually you too In other words, there should be many brands like Sony like Philips Your trust in it is still there It seems that the probability of you choosing it is not that high.

Yes Finally you found out or today Does this company continue to have a group of people He is today He can maintain an extremely innovative spirit He was confident that he could just talk about the first nature Seeking perfection and growing together This group of people I think will be the longest moat for this company If such a group of people can put ten years into

Keep saying this for twenty or thirty years Your company now Do you think there are many people like this?

Let’s use our company’s own example to test The foundation of the one, two and three layers just now I think the first thing is every aspect It is not 0 and 1 It’s not that you have it or you don’t have it It’s a process that you grow from zero gradually approaching 1 Then the words here Like what we just said Starting in 2023 We established an organization to build technology To this day

Chips such as integrated storage and computing and their applications I think for a period of time you may not see anyone else having similar things So I think in the field of technology We arrived at a few o'clock But I think it’s far from ideal I think there may only be one 0.2 0.3 So there is still huge space behind We hope we can still attract good creators

Let's work together to increase the technology board from the current 0.2 0.3 to 1 Yes As for the brand This is indeed I think our past 15 years It is the result of hard work and careful care In fact, Anke’s brand is globally recognized Everyone’s trust It’s still quite high For example Our February this year

Released a set of household energy storage products Customer price is US$10,000 There is no offline store to view It’s just released You may be on our official website Just place an order Spend $10,000 on this thing You can imagine For users What a difficult decision this is Because you are so important to this brand We have energy storage worth hundreds of thousands of dollars

But energy storage worth 10,000 US dollars Today is the first time OK, this product has just been released Cloth And there are no offline stores where you can go and see it So what kind of trust do you think today can make users willing to directly pay 10,000 US dollars to buy this product online More trust in quality Still need to focus on quality to this quality Have high trust in this brand and its responsible level

So our first month There are more than 1,000 customers Everyone buys this thing online So in fact, this is a brand that is trusted So I think among these three boards maybe to this day I think our longest board is the middle board I think If you press 0 to 1 I think 0.6 0.7 There is such a length pair and then in the end this is what we are talking about

It’s a group of people today Believing in one mission and vision Everyone can be together today A group of people who first and foremost strive for perfection In fact, if you go back to we just started doing this in 2020 It’s when we transform the entire company That point I think the proportion and concentration of our people It’s still quite low Yes In the past few years In fact, to be honest

we have gone through a very painful process of change In fact, this company is indeed our Of course there are many colleagues Like you said It means we will make money after we go public Maybe We went to do other things Then we also introduced many creators In fact, this process Do you really feel It is a process from quantitative change to qualitative change It may have been a little slower in the past few years.

In the past two years, it has actually become faster and faster Because on the one hand it is also because we are speaking out about these things In fact, I have been exposed in some content Let’s talk about what kind of company Anker wants to be.

Then we can give the creator What kind of environment do they provide?

Then what kind of reward But you always talk like this It seems that the vision is a bit empty So In fact In the end it should be your product that speaks In other words, if I keep talking today Says we want extreme innovation What are we going to do In the end you find out Our product is just a micro-innovation Improved product Then you probably don’t believe what I say But actually starting from last year

Some of our products on the market For example You saw our UV printer It is actually in this category The first consumer-grade product put one An industrial equipment the size of a table Made a small device as big as a suitcase Can be placed on the table You saw the storage and computing chip we sent and the headset equipped with the storage and computing chip These are actually the real creators Different things made in this company

So I also hope to say I won’t be on the show again in two years Hahaha I hope too Our number one Our products speak for themselves Those extremely innovative products It will make everyone feel This is a company that creates Secondly I really hope to let us creators them come out and tell their stories Because actually you understand As a product manager I created this company

Today, our truly creative products are actually our technology creators in each field.

Product Creators What they create So stand on the stage today Proudly talking about my own creations Such an opportunity should not be mine It should be theirs Your company now has a bottom up culture It is still a top-down culture We are actually a culture today We call it a hierarchical authorization culture In other words each layer makes the decisions that each layer should make and then delegates authority to the greatest extent possible in other words

As CEO, what I decide is what the company will look like our strategic framework Our Competency Framework Our Organizational Framework Our Talent Framework Including how to design our returns These are my decisions Then go to our business department Of course I also said Of course I also decided which categories we should do Big macro categories But how to do it specifically in this category

We will authorize the business department to do it In a category What specific sub-categories to make and what shape they will be made into This will be decided at the business department level Then go to the product line Then the specific product will look like this Most of the details inside It is authorized to the product line to make decisions.

So in other words In fact, our company has many It is within the three-tier structure of the company that we have many independent opportunities Just because you can do this thing Most of the time when making decisions You are actually a decision-maker on your own Our business lines are the same Just mentioned that our company has many major regions There are many business lines under the major regions

They also have many independent decision-makers Right So that actually you understand This company has many opportunities to be independent We feel that for creators there is an opportunity for them to be independent In fact, it is the best way for him to grow Yes So we think It can really provide creators with a good a growth track

You just mentioned the second brand of the three-layer moat You have the highest score I’m very curious It’s because you are a company growing in Shenzhen Why, especially in the European and American markets?

Made a good brand What do you know about this?

I can share it with you In fact, what we say is the legacy of the brand is the trust with customers So today if you are doing everything They all start from the perspective of accumulating customer trust.

If you make a decision then in the end should you gain the most trust?

In other words Your goal today is to generate trust for users Then you do everything It’s time to move towards this goal Should you gain a better brand?

Is this something you figured out on day one?

I think this is a very simple reason The first principle But the first principle is actually in the face of some big decisions It will be inspected For example You may know that we had a big recall last year We recalled millions of power banks around the world This economic loss must be calculated in hundreds of millions.

Several hundred million Such a big decision Of course you can choose not to recall Yes, but the result of not recalling it is Today, many users will indeed face the problem that this power bank will overheat.

Even the risk of burning We didn’t experience it when making this decision You know It’s just repeated discussions and a lot of debates Let’s talk about what everyone thinks We saw this problem today So today we start from user value what to do Just do it So something like this When it happens at every level of the company I think the final output is Of course

I think it’s hard for me to say what each level is Because indeed for this company it’s big it’s complicated There must be a lot of local decisions in the end It may not obey the global optimal Such a principle Yes But whenever we find a problem we will definitely correct it Solving this problem It’s us Whenever we find a It was not done according to the principles We will definitely correct it

This is from the perspective of probability.

We have been in contact many times When working hard to win the trust of most users I think it finally created a trustworthy brand Yes What do you think of the current brands?

Are there any changes in the trends of the times?

I think it’s better to go back to the first point of view That’s today From the perspective of trust It will definitely not change And then if you go up What kind of emotion do you bring to users Emotion based on trust If so I think actually some brands will make you very excited Some brands will make you feel like it They are very special Yes

I think these different emotions exist naturally and cannot be changed by the times.

I saw that you also established an AI middle platform You are now on AI Do you spend a lot of time Do you personally spend a lot of time How do you learn AI?

You understand that my background is in AI so Although I didn’t have the ability to do it well back then But after this big wave comes today I feel I definitely firmly believe in its future Yes, so actually I spent a lot of time in the past six months In fact, the flowers are spent in this field Our company actually started in September last year Until now For internal use within the company

I just don’t want to talk about it Tokens used for consumer services Tokens used internally By the time the program is broadcast, it should have exceeded 10 trillion For us, it is about 150 billion every day internally used tokens And this number is still rising rapidly In this process Our company only has 6,000 people It means Everyone has 25 million tokens every day

Then there are about 500-600 million tokens every month And among these Approximately one-third are top-end models and then About 50% of them Forty percent are mid-range models Then 20% are what we call entry-level models right assigned to what to do In fact, today because people traditionally say that enterprises use AI Just use it to write code, for example.

When it comes to making products, in fact, every aspect of today’s company The functional lines we are talking about all use a lot of AI today Then what about our AI use today?

It’s not that I open a lot of accounts for everyone and then everyone uses AI on various platforms Instead, we built a middle platform Then we connected various models Then on the front end We provide chat Provide local command line Provide web page Provide various access methods Then let everyone in the company use AI on the same platform Then this is me What we call the AI ​​middle platform Yes

In fact, if we go to see it First of all, it is very popular this year Such as OpenClaw we say OpenClaw stands for We call it real-time orchestration task processing right That is, when you tell him a requirement today he will help you compile a process on the spot to help you handle this requirement Then his challenge is that today he is actually

First, he is temporary He is unstable And then the second one There is a high probability that he will Not fully understanding the data within your enterprise Then you only have processes and methods So we think enterprise-level applications It should actually be very large Pre-programmed agents That’s how I made this business successful The process and the data we already have in today’s business Integrate it into the intelligent agent

Let this intelligent agent deterministically be able to generate the value we expect So this is what we think about the use of AI in enterprises today A huge difference from today’s personal use of AI So today an enterprise-level has a variety of different functions He must be able to accumulate his abilities into a pre-programmed agent and today’s pre-programmed agent must be able to run on a platform Then it is carried and then

is distributed to different models for running In fact this still requires a complex system and a complete and comprehensive thinking Yes Back to the beginning You said that all members have unlimited tokens Was there any objection from the management at that time In fact, we thought Because many people also asked later Tell me how you measure the output of AI

That sentence is called because you believe, you see In fact, up to this point we have not strictly measured it.

All the tokens we spent What exactly did it produce today In other words We believe that the future organization It must be an organization enhanced by AI So you’d better start early today Yes, in fact, during the whole process we didn’t settle any accounts We just thought Just do it And actually Very interesting It’s

because this year our AI spending has indeed risen to a big number So let's start by talking about today Well, the cost may be as big as one We really need to take a look at it seriously today You may even need to optimize it How much This year's expenses must have been several hundred million.

Oh yes This is our internal company The cost of using tokens for employees today What is the upper limit in your mind?

I think in the end it actually matched the value.

If we can create good customer value Today we can earn a lot of money from customers Then I think In fact, there shouldn’t be any upper limit for this Everyone is talking about AI native organizations now What do you think?

Do you need to become an AI native organization This question is like what you asked just now AI native hardware So what exactly is AI native hardware?

In fact, you find that whether it is hardware or organization We feel that it must have some unchanging things Then Then I think we should think clearly about these unchanging things Then think about how AI will change it Then the organization produced in this way may be what we call the organization in the AI ​​era So back to the organization What is constant First of all, the first one is user value

You may find that in the entire field today the most likely high-frequency word User value We think in other words For a product organization We make our users happy and satisfied This is unchanged Then the value chain that creates user value just talked about What you have to go through to create this user value process one by one For example I want to make a hardware product I want to do user research first

Then I need to define this product After defining it I may want to develop some technologies There are some things I need to do, such as designing Finally I want to organize it into a good product Then I have to do a good job in quality I have to do a good job in production And I have to sell it to a good place Then this is just about producing the product

Later, I also have marketing and sales services There are many, many things to do So when AI comes will these value points disappear No As long as you find that making a product makes users feel satisfied you still have to experience these value points So then we say AI will not Change the value chain of the entire organization today Then In fact, if you take a serious look at the value chain

You notice it looks like a big organization In fact, in the value chain It is completed by small teams working together one by one.

In software development, it is called Scrum In hardware companies We now reuse the term Scrum In other words every point in the value chain is actually delivered by an organization So let’s say Today this organization will completely disappear Will it be replaced by AI?

It will definitely not disappear Then how will it be changed by AI Maybe it will shrink This is what we see today Just in the software field Give an example A traditional Scrum might have 8 to 10 roles Start with user research Then do unit testing at the end End of testing So today AI actually brought Integration of positions That is, one person can play several different roles.

In this case this Scrum will become smaller it may go from a team of a dozen people Become a team of several people But it will not disappear Then the hardware field is actually different Just like AI is coming Today I can’t let even an electronic engineer suddenly be able to draw structural diagrams Suddenly I can suddenly do embedded code Because these are indeed related to the physical world.

It still requires a lot of effort A lot of experience needs to be accumulated Yes So then you find Scrum related to hardware related to the physical world its size may It may not be reduced No change Oh yes Or it can be reduced a little So what I just said is So the first change brought by AI is Some Scrum people will be reduced And then the second one

I think the changes that AI will bring are The internal operating mechanism of Scrum will change That is, it no longer seems like humans are defining goals.

People are dismantling the process and people are completing the process Today’s AI can help you define goals Help you dismantle today’s projects and OKRs and manage it and in the end it may help you get a lot of work done so its efficiency will increase Then Then today In addition, it is reflected across different Scrums Its collaboration because in the past

everyone had to meet with each other Actually I believe that this is the traditional way of communication in the past After the AI ​​comes it will also become more efficient so pull it back a little bit What we just said is that the value of the first user will not change The second value chain will not change The third chain The delivery organizations that already exist There is a high probability that these organizations will not change

But first, some of these organizations will become smaller.

Then the second its internal and organizational inter-organizational collaboration efficiency will actually be higher.

So we feel that such an organization has completed such a transformation It is an organization in the AI ​​era.

You are transforming now How is the progress I think In fact, after more than half a year of practice we have a clear view of today’s path Yes But to be honest In fact The change of people and the change of organization In fact, it all requires a process.

I hope we get to the end of the year be able to complete such a transformation It’s already great Some people think that in the AI ​​era middle managers have become more important Some people also think that middle managers will disappear What do you think Indeed, some people think that There is no need for middle management in the AI ​​era For example, Jack Dorsey He had a very famous article some time ago Said that he felt

that the company should be controlled using a world model and that middle managers should be eliminated Then let me be honest I think today’s model can handle the context length I think there is a high probability that this is done less than In fact, the length of context Today you have to describe a problem clearly The number of tokens required For example, I have a project today Then this project will explain it clearly

How many tokens are needed Then I have a Scrum This Scrum may be running several different projects at the same time How many tokens does this Scrum require?

And today in one of my departments there is a high probability that there are several Scrums So the manager of this department today How long is the context required?

Then go up Today you add up the layers layer by layer You discover the company How long is the context required for the entire company?

Then after you finish doing this calculation, Do you understand and explain a company clearly today?

the context length it requires It must be more than 1 billion More than 10 billion Even hundreds of billions to trillions But it depends on the size of the company In this case Given the processing power of today’s model when you say it exceeds one million contexts what will happen to its capabilities?

Or lose For today you can say compression But what is left in the compression?

Leave nothing behind This is actually a very, very important value choice Today’s value choice I think the model is definitely not very well done.

So today's process must require people to do a process from a large number of complex contexts Grasp the key things Then say another point That’s what we said The relationship between the organization and its results There is a famous law here called Conway's law Conway's law says The shape of your organization

determines the results and effects of your output So I would like to ask Who defines the organizational form today CEO or grown up You will find the tissue shape after it grows out It is still organized today by people working together So today then let’s talk about an efficient organization with the right people in it He has the right atmosphere He has the right communication style These can be said today

Did the CEO alone support its growth It seems not possible It must not be It must be managed by layers of managers Go give your efforts today To really cultivate this organization today So we say You can’t assume that Today's company has all the right people and they work together efficiently In order to be able to build today The right person just said

Efficient collaboration and appropriate organization This requires managers People who put in huge efforts and hard work So in my opinion I think middle managers will not disappear because of the arrival of AI and in fact they may become more powerful What will disappear?

I hope the last ones that disappear are those that are repeated boring job There will be fewer entry-level jobs There is a suggestion for today Everyone would think that the AI ​​era has eliminated entry-level jobs so there will be no senior positions in the future.

The assumption of this statement is that senior positions must grow out of junior positions Let me give you a counterexample.

For example, construction Junior positions in construction For example, it may be building walls and plastering All in all, it’s a lot of operational work Its senior position may be architectural design Then those who planned Then if you look at ancient times In ancient times, it was really the architectural designers who came from First-line craftsmen Slowly rising that is When you become an architect today

There is a high probability that you will spend the next ten years or so I did a lot of building walls Very specific work of plastering What about modern times?

Now actually the architectural design has become a definite category You don’t need to say today that I have been building walls for several years It has been dusted for several years Only in this way can we become an architectural designer So we think it will probably be like this in the future In other words Today you don’t need For example, you have learned programming in a certain field After fixing a lot of bugs

you can be an architect at the top level today and then build solutions for customers and deliver a value That in fact We see today that this is indeed the case In our school admissions today using AI a lot Just students who have no programming ability at all I should put it this way Just don’t have any programming experience Their undergraduate major was not programming and they had no classmates who had ever written programs

Today he was in front of a clear business field Use AI to support him in delivering a complete system to support business operations Our company has many such examples So this is actually completely skipped seems to be the so-called traditional entry-level position I have reached the role of a senior position So I think actually I am not worried about AI eradicating entry-level positions.

There will be no talent for senior positions in the future Do you think the organization will shrink I think The organization will indeed become more refined more efficient Yes Will a few people earn higher salaries For example, 10 times In fact We are also talking about such a point of view Let’s just say Then First we define a word We call it N-times effective individuals

We call it N time efficiency worker and after abbreviation it becomes NEW which is N time efficiency worker It’s this new person Then In other words, it’s the person whose performance is increased by N times Then when you look at it with a magnifying glass you will find that it is actually divided into two groups You see A group of people who are using AI to build capabilities today In other words

These people are actually in the business To understand the business just mentioned Data process method precipitates into intelligent people They are the ones who make axes Then another wave We call them small new people We call this wave of people big new people To the other wave of people who use these AI agents We call them little new people Just them Maybe you don’t have to start from 0 yourself Make this agent

But they need to be able to master and use it efficiently These agents their effectiveness Also because of the existence of these intelligent agents And was promoted by N times So we say that they are big people small new people big new people are the ones who create AI agents Then the little new person may be someone who uses AI agents Yes So we think Finally

In fact, this organization may be composed of a small number of big new people and more small new people today And the distribution here Because their value has been created N times So a good organization Value should also be allocated in a corresponding way Yes Because we are a consumer electronics company In fact, let’s go back to what we talk about first

We say the value of consumer electronics is Who created it They are people who make products In fact, because of consumer electronics We just talked about quick growth and quick death It’s just that your product is really fast If you can’t make the most advanced product while iterating No matter how great a company you are Maybe it won’t work in a few years Or even gone Nokia is a very good example

So actually we say consumer electronics because you have to continue to innovate continue to be at the forefront And these must come from your employees So today we say Its value is created by employees Then we feel that a company that can continue to succeed its value distribution must also respect this law of value creation.

So then we think More value should be given to employees Just give it to these creators So we always talk about it in the company We say that creators get 7 Then the shareholders get 3 Yes, so we have seriously thought about this issue We feel that it is in the AI ​​era Then After our value creation has been increased by N times But we feel that people are the creators

Still the main body of value creation So in the AI ​​era We will still adhere to this Workers created such a distribution ratio of taking the big head Why is it 7 to 3 Not 8 to 2, 9 to 1, 6 to 4 In fact, first of all we will say The creator should get more This means that this end of creation is big then today is between 7 and 3

Let me talk about this underlying principle That is We define a word called total surplus value The total surplus value is your income After excluding all the expenses you have to pay today Then you have all that is left The ratio that can be divided among employees and shareholders This is called the total residual value So it’s fun You find that first-class companies are those with the most innovative products and the most extreme brands

They usually have a total residual value of more than 30% And what about this Then what about the 30% How to distribute it to employees?

How to distribute to employees and shareholders How much to distribute to shareholders Basically it is more reasonable We will feel because a company if today’s final net profit If there are only, say, 3 points 5 points, then this company is most likely not a that everyone usually thinks is a good company.

The shareholder value should be almost 10 points This is what we think A basic threshold for a good company So you look at it from this perspective When you have a total residual value of 30 points When shareholders want to divide 10 points it’s almost a 7 to 3 ratio So come to your company and become a creator How much cash return can he get Can you give us a quantification

In fact, we are really calculating it It’s how much money we all earn in the company Right In fact, first of all, it’s an average number And we look at the changes in the average In fact, because of what we just said we insist on creating better products That's when we started to transform from to seven series

In fact, you find that the higher the proportion of your seven series products The greater the customer value you create So in fact the total customer value you can retain in the end It’s getting higher and higher So it’s reflected in our company’s recent years Our gross profit margin can probably increase by 1.1 to 1.5 points every year Then higher income After higher gross profit Then you see that actually Distribution to our shareholders

That part has not changed much So the extra money In fact, it is all distributed to our employees In recent years In fact, the income of our employees There is a really continuous rise So we will make internal statistics Our colleagues who have been in the company since January 1, 23 His total income in 2024 compared to his total income in 23 There will be an average improvement of 20% Then comparing 25 years with 24 years

there is also an average improvement of 20% So this is what you can see When your company's products are more extreme and innovative Then your customers are worth more Then your total value is more Our employees continue to earn more And then back to the specific absolute numbers We will also count a number That’s what we call the number of people with an annual income of over one million

The number of companies actually in 24 years This number was about 500 people We were about 5,000 people Then When we had 6,000 people in 25 years the number increased to 800 Then I hope this number may exceed 1,000 this year Then The final ideal situation In fact, if you can see the best the most innovative company in China In fact, the per capita income can exceed 100

Almost everyone has a per capita income of over one million You just mentioned Now we are facing such a big wave of AI Then everyone is making changes in AI How do you think you did today How painful I just shared In fact, all of our staff use AI Then in our own center to access and use

Then bring our business experience knowledge capabilities Precipitated into the AI ​​​​intelligent body This is actually a lot of work do You understand that the basis for AI to operate today is Your data is in the system today So today Internet company Maybe naturally Most of its data is in the system But today as a hardware company in fact most of our data

is not in the digital system In this process you have to work hard and gradually sweep in most of the data Then based on these data Make the orchestrated agent we just talked about Then what is particularly important is Today’s intelligent agents in different fields It also needs to be able to connect In other words If the product’s agent is different from my R&D agent Its data objects are inextricable as if you speak german

I speak the same English This communication is actually very difficult.

So you will actually be needed behind this Just in a certain After establishing intelligent agents in each field, Then we can connect these intelligent agents across fields Behind this is actually a huge workload And what about the last words?

You find that your organization is actually going to change How to change your organization What we just talked about Then you have to break out Scrum and these Scrum to be able to fully use AI today The changes needed behind this are actually huge Yes So Of course, back to the end Your person will change That’s right, you discover that in fact, among these three In fact, the hardest thing to change is people Yes

People should be more adaptable to AI Or do you want to replace people I think it is People today have to believe that AI will not replace themselves AI will make itself stronger today and then be willing to embrace AI Then give your own things to AI Give some of your own In fact, I just talked about data In fact the first one is that a small number of people will build AI build AI capabilities

Only a small number of people will build data Build agents Then a large number of people will use this data today Application of intelligent agent For this reason returns In fact, we will summarize it in three ways The first is that we just talked about people embracing AI The second is that abilities are precipitated into intelligent agents Then the agents are connected The third thing is that your organization changes

So all three are completed Only then can we become Today is an organization in what we call the AI ​​era How will your organization change from When did it start last year We actually started in September last year Oh The most important opportunity is actually that you saw something called Speckit Oh You know Speckit is actually a set of what we call process intelligence It defines what product managers do Then how to develop

How to test and you will find on the model Things that don’t seem to be accomplished effectively Then add the description of this layer of intelligence and the limitations It runs very smoothly So from that perspective I think you start to really understand How organizations should use AI You must build your field The intelligent agent of the process comes out

So Speckit It is a set of intelligent agents for software development.

Then you think today in other fields when you can build that set of intelligence your organization will become AI-based Has anyone guided you in this process Organizational changes in AI are your own ideas We think we should be inspired Just said Inspired by Speckit But we are one That is, we believe in abstraction In other words You saw something that looked pretty

Or you saw something that amazed you You abstract the underlying principles behind this thing Then you diverge from abstraction to concreteness into different fields The way our organizations evolve is generally like this Yes Have you become more anxious in the AI ​​era Many people around me have told me that they are very anxious this year.

are you anxious I think there must be a little anxiety Because it seems to be changing very dramatically today Yes, and it’s getting faster and faster But Our habit is that you have to ask What things are constant Like what I just said Your customer value remains unchanged Your value chain remains unchanged And what happens in the end?

These beautiful values ​​of yours need to be created by creators These things are unchanging Then you will ask yourself Are you doing better or worse with these unchanged things?

We feel that those of us at the bottom who remain unchanged It’s time to get better and better Then I believe that no matter what kind of changes we face, I should be the better one Yes Of course this is actually a rational way Today I will help you overcome your emotional ways This doesn't mean you have no emotions at all inside.

But I think in the end In fact, do you understand that rationality and sensibility coexist?

AI coding is very popular this year Does this have any impact on companies like yours?

In fact, we have a big software OK team Our product manager in the software field Engineer Algorithm Engineer Then Embedded Engineer Test engineers combined I think there are 1,000 to 1,500 people Yes So what about Helping everyone really catch up with AI In fact, we think this is the company’s responsibility In other words I should not be expected to say our colleagues

I seem to be learning these things by myself Then it seems like he suddenly knows it From the perspective of empowering them The us we just talked about In fact, first of all, we are the company We will internalize a set of intelligent agents of our own software Just mentioned a set of things like Speckit It actually has something to do with our product areas and the entire data structure of our company

between our company’s organizational structure In fact, there is a lot of room for adaptation.

And what about the next thing?

You will really help our colleagues today Ability to use intelligent agents In fact, to be honest I have been working for 5 years 10 years colleagues He wants to change his inner motivation today It's one thing The organization creates an atmosphere and an environment for him Also on the one hand Yes, so what we feel today is We definitely want to say that

is empowering our colleagues To be able to catch the AI ​​bus today Instead of being left behind by the AI ​​bus Are you worried about being left behind by AI?

I'm actually not worried Because objectively In fact, I think our company is really understanding AI and embrace AI today and let it happen today our colleagues Everyone can be empowered by AI today I think we are the company at the forefront Compared to who You can't say that compared to large model companies you are at the forefront Of course, of course Today I think the relative

is in the same industry as us Compare these consumer electronics companies So actually joking When a bear comes running You don’t want to run the fastest You want to run faster than your peers Do you agree Are they concerned about AI’s actions We are actually not a special Just try hard to seem to understand what others are doing Oh In fact, you didn’t communicate with anyone in Shenzhen, right?

Originally I was an I person But I think In fact, I also found a reason for myself In fact, back to There is a book called Finite Games Infinite Games Infinite Games In Infinite Game There is a very interesting story in the prologue He said that he went back to 2004 Microsoft and Apple compete for MP3 players Because Apple’s iPod was particularly successful at the time

So Microsoft also wants to do the same thing Where is this author It happened to be at the same time I had contact with executives from both sides Then the Microsoft executive told him Look how beautiful our player is Functions are good Better than Apple What are the advantages?

Then when he went to talk to Apple executives they would say hey We think the user value Our users still have some dissatisfaction what we wish we were doing Solve the dissatisfaction of users Yes So there is no doubt in the end You can also see in the hardware industry Who has the last laugh So we think this is also a I think it is a common basic truth Because for people like us

we always want to talk about first thoughts For the companies taking the exam The first principle is to say When you are staring at the competition today You are always the last one to escape is a faster horse Not a real car today So only today You don't go staring at the competition all day long You think hard What are the most extreme needs of users today Or you think about users today

What is its value point Then look to the frontier The most advanced technology Where can it be created?

So today you take care of users’ unmet needs Then develop the ultimate technology and create products to satisfy it On this road In fact, the faster you go, the faster you go and the more stable you go, I believe that you will most likely not fail.

Yes I think it took us more than ten years to really understand this truth It's because when you were young you always felt It seems that I am compared with the people around me But at this age I feel like I have thought about these things clearly We feel like we always say hey Have you understood your first principle clearly?

Then it's over In what year did you realize it In what year did you realize it In fact, it will be 2023 when we change our mission and vision OK And what’s behind this It’s fun We also have a very true story Because our company has always been us You will think that we are just like Not so fancy Then we do Who is more fancy Who is more fancy in your consumer electronics industry

If you think so, of course you understand DJI in front Including those are very fancy You're not fancy either, are you?

Do you think I’m fancy I don’t think so Haha So what kind of person are you I think our values ​​are actually very good It’s actually a personal portrayal of myself.

that is Today I am willing to think clearly about the first principles I am willing to take one step at a time today.

Strive for the ultimate in execution Then you are pursuing the ultimate and are willing to think clearly about first principles It means you don’t have the first principles You don’t have the ultimate yet That’s why you strive for the ultimate I think this is actually the case Let’s go back to the story we just told about going upstairs.

It’s when you are on the first level You see the first principles of the first level You seek the ultimate in the first principles of the first level Then this takes you to the second level At this time You find that its first nature is different again For example, 15 years ago When I first started doing e-commerce The first thing at that time It’s completely different from what I’m talking about today.

What was your first priority at that time resigned from Google to make a power bank What do people around you think?

In fact, many students must think Oh Why make such a simple thing?

But let’s talk about the first one We feel that the first nature has not changed in the past ten years It is to create customer value But at that time Today we see the way to create customer value Included in that complexity category In fact, we just want to make an improved product.

Then just said Through customer feedback Polish this product to perfection Right and then to today I think we are in many complex categories When creating the ultimate innovative customer value Did you find what we were talking about?

This first nature It is completely different from that time Yes It goes back to the essential principles of things just mentioned We need to start making breakthroughs and innovations like chips Yes When you were a child, there was no one who aspired to be fancy.

This is a very good question Actually I think I never seem to be the coolest person Oh Then what kind of person are you That’s the first thing First I am definitely not That is It seems like he is saying the cruelest things Then, for example, wear the brightest clothes He seems to be the most fancy looking person I think we really have always been Dongping and I are In fact, we all work hard

very solid every day do better than the day before That is We are people who stand firm on one step and work hard to reach the next step It’s more like an engineer solving a problem when encountering it I think In fact, engineers definitely have this quality That is, they solve one problem and then solve more complex problems Then solve the most complex problems Are you a gambler No, if it is extreme left or extreme right

Where are you Let’s first define left and right Suppose you have 10 million and you can invest in a project today or start a business or something like that How much will you take What is the extreme left?

The extreme left is all in all in In fact, the extreme left is even further left than that It's leverage You invested 10 million of your own You will also borrow a lot of money to invest in this matter So what we just talked about If the extreme left is 5 points We say 5 points is leverage Then 4 points may be full investment Then what if you count to the right If it’s 1 point

Maybe you don’t invest even 1 penny.

2 points Invest in, say, three or four million 3 points Maybe you invested six or seven million Basically we formed a score from 1 to 5 today one option at a time This is actually how we think So when you encounter a problem We always like to find the extreme left and extreme right and then find 5 gears in the middle and then that In fact, I have always been in the middle Oh

It’s just that we won’t invest fully We will not not vote Then we will definitely vote more than half Then today Work hard to get to the next level So when we talk about going from 1 to 5 today it seems like you choose 3 points It looks like it's in the middle Then the Chinese used to have a saying called Zhongyong.

It seems that it must become mediocre in the middle Yes, but I think today If you put it in a long enough game Just long enough rounds You will definitely vote today every time And as your knowledge and abilities improve, With the creators on your team today It becomes more and more Then when your probability of success becomes higher and higher Then I think it will actually become a success in the end

The middle will lead to a success Especially it will lead to a successful third-category company It means being able to support many creators Create many categories on this platform A company that has created successful businesses in many countries Yes What kind of company do you think DJI is?

In fact, the number of categories in DJI today must have reached There are dozens of categories in one So I think it is also a third-category company Yes Is it fundamentally different from you?

I think DJI is a slave I chose a difficult mode from the first day Even a company with regional difficulty mode The category selected on this first day In fact, it is a very difficult category Then so the first few years must have been very painful.

Right, after you win a difficult category today You have accumulated good underlying technology Then you accumulate a group of people with high values When we talk about people Then expand to other categories It becomes easier What about us In fact, because we chose one at the beginning Relatively easy difficulty mode Then we have to start to hit it harder when we go up.

In fact, it’s easy in the front But it becomes difficult in the back Oh So in the past few years In other words So I have been making up lessons in the past few years Yes, so we have actually been making up lessons in the past few years Playing this difficult game Yes What about Shadow Stone 360?

shadow stone I think actually JK I think you think it’s more like DJI Or more like us DJI right From a general perspective Are you the only two types of companies in Shenzhen?

One is to choose the hard mode first The other is to choose the easy mode first Then It's basically a 2-point choice It's you or 0 or 1 Of course it is like this It may be a matter of degree You may be 0.2 0.3 Or are you 0.8 0.9 But in the end I think there is a high probability Difficult and easy You have to have a standard in the middle

If you and DJI were on the same battlefield Who do you think will win?

That's what I think We can still go back to what I just said.

Look at the elements of the three barriers It's the first Do you have advanced enough technology?

Second today in this category Have you won the trust of users?

Then the third word is whether you are first-class enough today and strive for perfection enough So I believe that in these three areas In fact, we are all growing Rapidly growing What have been your personal changes in the past five years After switching from easy mode to difficult mode Have you changed I think actually when you are in easy mode

In fact, you may do one or two things well Then you can live a good life When you get to the difficult mode Then you will find In fact, you can’t make mistakes in almost every field.

So you're talking about a company today Requires a complete understanding and cognition And you need to be able to actually practice it So I think this requirement for to be the No. 1 position is much higher I think I’ve actually been stumbling over the past few years Yes It’s called self-awareness Self-evolution It means growing up to today Beginning to realize the truth Start meditating

I think it’s right It’s just that it will be very painful for you Sometimes Then when you are in pain Meditation is a good way to solve it What did you do when you were in the most pain?

Actually today When we want to do a It can empower many categories Then there are many companies that creators can do well and today is You seem to feel When you can’t defeat those unicorns outside It’s actually very painful for you Yes But after a period of time you realize that it is not unsolvable

but the company you built today This product is missing a key link After you add in these key links, In fact, in the past two or three years you will see In fact, we It's getting better and better The solution at that time was cultural values Yes, actually what we just said is When your strategy is clear Your organization is leading in the accumulation of capabilities

Then what about your allocation of people?

When we are willing to distribute a lot of money to others then logically I think you should be able to do it Then why not Finally you found out In fact, no matter how good you are, today’s advanced method Advanced tools are all for people to use And what kind of person do you think you are today?

What kind of behavior should they use Go to daily work today When there is no clear definition No matter how good you find today's tools and methods it cannot guarantee your success in the end So when this link is added Well, I think today everyone can see something that can be seen with the naked eye We used to In fact, it seems that it’s a category that I can’t win.

Actually Turn defeat into victory This process and The process of AI transformation is actually a period of time In fact, changes in our mission, vision and values It started in 2023 The transformation of AI started in 25 years So it is actually a superimposed process Mission, vision, values ​​and your What's the relationship with your cutting-edge lab?

Just said 23 years Let’s start with two things OK The first is to dig into cutting-edge technology Deep root technology The second one is We talk about mission, vision and values Then attract the creator Yes In fact, you find a creative person It’s because he has first thoughts He often thinks when others look at him This person is so weird

He keeps suggesting directions that seem unusual Then he is very demanding Everyone will think This person seems to be very curly And long-termism is Then why do you seem to put it in front of your eyes It seems that KPI has to take care of other things So people like this are found in many companies felt by everyone In fact, he seems to be quite out of place.

His own experience in this company Maybe it's not good either But what if today you arrive in an environment Then if the environment itself advocates these things And there are more people who behave like this His work will make him feel happier So today When in this environment When there are more and more people like this He will actually change those people who are not like this in this environment It’s actually very fun

That is, people will be changed by the environment.

right So after the atmosphere of our company has changed today, It doesn't mean that you replace everyone When you bring in some good creators the atmosphere changes In fact, the rest of the people are changing You went to the company a few years ago Do you feel the changes in the environment Are there any details that can be reflected?

The details you can observe as a CEO You know I am relatively abstract Yes, then I Actually, I don’t walk around the company every day I seem to be looking at a lot of details Oh, how many hours do you go there every day?

I should be within 8 hours a day OK Within 8 hours It should be 8 hours a day for me Do you not want to go so much Or do you consciously control your time in the company I think I am actually conscious Control your time in the company Yes, and then because indeed you find that it is As a third type of company If I am in the company all day today

I seem to be pushing a lot of things directing a lot of things Then it will most likely not grow into a third category company Oh, you can understand Because the third type of company In the end, it actually depends on this system It relies on our creators to run Instead of relying on me to push you here today Just give me a push and run away.

Yes, so one such one is today This is what we call the framework. Think clearly

Then find the key people today Then today Really It’s the process of letting them make this thing Yes I think this is the path I saw today Yes What does a typical Yang Meng day look like?

In fact I will focus on the key areas that I care about and then maybe open more Neither I think there will be more discussions For example, in the last six months I actually spent more time on the AI ​​revolution Right and then Then from the level of the entire organization to the level of a specific tool actually do you think you will go Join us

to abstract from the facts and then go back to promote more Better Facts Reality Versus A Typical Day Start talking in the morning What time do you get up What do you do in a typical day?

Start talking in the morning I’ll probably get up around 8 a.m.

Then I’ll go to the company between 9 and 10 a.m.

Then I usually don’t have a particularly close schedule of meetings I probably have meetings half to two-thirds of the day Then it’s empty at other times Then after 6 o’clock Yes It's I should say it's right The environment in which I left that company And at night and usually because I am a person who likes to read

Then I like to look here and there People who read a lot of things What to watch What to watch Today’s short video Long video And indeed knowledge in various fields In fact, I know a lot of unpopular things Oh, for example A piece of cold knowledge you recently acquired In fact, I am more interested in it like politics historical military

Including some of the hobbies we mentioned like food That’s all You can still spend a lot of time to understand and you find when you read a lot of specific things When you try to abstract it a step further There are many very close laws behind it Yes What kind of similarities have you felt recently What kind of similarities have you realized Let me tell you one More fun Actually

Every revolution in productivity In fact, some very interesting things will happen in it common things First, you will find productivity revolution The Great Revolution is not a small gradual improvement Every productivity revolution creates higher value Then to whom will this value be assigned?

Who will be assigned first Military and Political You will find In fact, the value created In fact is often allocated first to the person who allocates the value Oh In other words After this watermelon is cut out In fact, it’s often the person who cuts the watermelon.

Take it first And in fact, in this process you find that whether it is the earliest bourgeois revolution or the later In fact, it's all like this In fact, there will often be a revolution later And then today Let this value be distributed more evenly Then in fact, this AI revolution In fact, it will also create a huge value And the person cutting the watermelon is the modeler I think the modeler must be part of it

And besides this In fact, there must be more These are the people who use AI in today’s enterprises And then actually So you see why Sam Altman would say In the future, we will enter a We call it a universal income society That is, the government will pay everyone In fact, what is the complete closed loop of this place?

That's today AI will bring more value to the company Then these more values ​​need to go through the government It was taken into the hands of the government through taxation.

Then send it to everyone So this completes a complete closed loop So it’s also good for CEOs I think today Actually, I don’t want to say it’s good for us because I still say I always feel that we are sincere As far as we feel We want to distribute value to those who create value So I tell everyone repeatedly in our company We said we used AI today We have improved efficiency and generated more profits

In fact, people will not say that they will be left to me by default.

Then we still say Between creators and workers today between creators and shareholders It will be in a ratio we agreed on before to allocate As far as we think In the short term, you seem to create more value It seems like you can share it with yourself But in the long run you have to make this value circulate When you can run In fact you need to allocate value to those who create value

Have you read any interesting books or watched any interesting movies recently?

In fact, I don’t read much now Because you find In fact, from a context concentration In fact, often a book’s does not have such a high information density Or It’s hard to find a book with such high information density Oh Yes What about movies In fact, movies There is also a very interesting habit that is

I can hardly remember the movies I have watched.

I can watch it again every six months It’s like new Hahaha In other words when a person today his attention Focus in the present and future times Especially in the future time In fact, he is selectively forgetting Oh, what kind of memory are you more sensitive to?

Sharper That is, photographic memory I feel that my memory of abstract structural things is continuous and firm When did this come into being?

This memory model In fact, I read many of your interviews and they all say that you are a person who likes abstraction What’s the truth?

It’s when you discover that actually You have to get good grades in the college entrance examination In fact, you’d better learn abstraction because you found The so-called knowledge points and questions in the college entrance examination It is actually highly repetitive when you can abstract it into several knowledge points several question types You are just like the process of being a cook and undressing a cow You can complete it very quickly So

I think that should be your approach to abstraction It’s my approach to abstraction The first one seems like a sweet spot Yes, and what follows is actually true Just a continuous habit that is It seems to take a step back from the specific things and then try to find the essential laws of this thing Then I think If a person or an organization only abstract people

It's definitely not a good life But if an organization actually has me Such an abstract person Then add concrete people Everyone together Just like what I just said My MBTI was completely opposite to that of our president Dongping’s MBTI was completely opposite Then hey Together like this 1 plus 1 actually gets better and better Getting better and better Yes So we are joking My company always says something

Just say hey If this company didn’t have me This company may not necessarily be what it is today It’s just getting better It seems to be more challenging More complex and more fun Then If this company didn’t have Dongping it would be dead Hahaha If Anke dies one day Why do you think it will be because Should we return to first principles?

Why do you think a successful third-category company dies?

Companies must die.

Just today when you see something like Procter & Gamble something like Texas Instruments it’s in dozens of categories Systematically can do it well And this company can continue to attract good talents When you become a very powerful person Then why do you think this company will die It seems that you don’t really want it In fact, the real third-category companies Successful Category 3 Companies When he can win what he just said in many categories

and continue to have good people In fact, this company will not die The reason why he died in the end must be because For example, his people started For example, everyone has forgotten the previous mission and vision Everyone no longer adheres to the previous values.

Then this company may be here It’s time to gradually die Today Do you think there are no signs of this?

Is today going to be good or bad I feel that we have not become a such a successful third-category company today Right So I think we haven’t reached that point today It’s the time to reach the best Let alone death If we don’t catch up with this paradigm change of AI Will Anke die?

As for what AI means to Anker I think one is user value The other is business value and employee value That’s what we just talked about Because of AI, the products around us will become truly intelligent If you can't provide users with such products Your company will definitely die Then the second reason is because of AI The efficiency of your organization The speed of growth of your people today And his ability

will definitely lag behind Then in this case Then your organization will die Yes So I think Today, whether it is the value provided to customers or the efficiency of the organization and finally the growth of people If you don’t embrace AI today you will definitely die in the end You know Ankerneng is a company that survives

How many years is the company It's actually fun There is an analogy for you to be the founder What do you think of your company?

In fact, my way is that you are raising the baby Your baby dies You must be in pain And what about the far-right approach?

this is a work There is a very interesting animal in China called pig If you raise pigs in rural areas Basically, you raise them to eat meat during the New Year So during the Chinese New Year basically you will sell or kill the pig and eat the meat So the extreme left is raising babies The extreme left is raising babies The extreme right is raising pigs And then what about the middle

There is another one called raising horses or raising pets.

So for example, your horse is actually your partner and is your tool to a certain extent.

Then Normally someone says I want to buy your horse buy your partner away You definitely won’t sell it But if this person comes forward today and offers a price you can't refuse you may also consider selling this horse.

So from raising pigs and horses to raising babies I think you need to think clearly first What do you think of your company Then Dongping and I We think We are really a bit like raising a baby Yes, I hope this company can survive for a long time In fact, why do you think so much Why do you work hard to transform In fact, you want this company

as if it were possible outlive us right You just said that you really want to recruit those creators Then why don’t he go to a big factory Don’t go to today’s big model company Are you coming to Anker?

Speaking objectively I am sure today I’m not trying to compete with big model companies for people.

Having said that In fact, I believe you must also understand In large model companies those who actually make models Although they seem to be earning very high incomes today but their sense of job crisis is very strong Many people feel that Maybe he gets a high salary for a year or two Maybe after AI evolves later he won’t have a job either I believe that many people you talk to may also have such worries

I think our characteristic is Because we are in contact with the physical world so our iteration cycle is slow so our human growth There is a long, long cycle and time I often tell everyone when recruiting at school Why If you go to a major Internet company, Between coming to a hardware company If you make a choice In the early years many people would definitely choose to go to major Internet companies It just felt like

I am 27 or 28 years old It seems that I can earn a high income Then I seem to be able to stand alone But you see This world is very fair If you come out to mess around, you will always have to pay back Hahaha You find that for major Internet companies everyone will say hey 35 years old It seems that you will encounter a continuing problem

Because the next wave of people in their twenties and seventies Grow up Then look at the underlying principle In essence, the speed at which a person matures It changes with the number of battles he fights Oh really Do you want a soldier?

When he fought in the first few battles he might have been very green right?

Then by the time he got to the fifth and tenth game he was already very experienced By the time he got to the 15th and 20th battle He is very experienced Very mature soldier So in fact, your maturity level It's with you number of times is a highly relevant Then the characteristic of the Internet is actually Every battle is very fast

A product is released today Improvements may begin in a few months Then a feature is even as short as one month So today you find a person entering a major Internet company He may fight once in three months a year and then four times in a year Then he works until the fourth and fifth year He has fought more than ten times Fight At this time he was actually already a Mature and experienced classmates

So of course his value has increased His income has also increased But as mentioned just now So today you find that the first is when your battle reaches 50 or 100 games In fact, it has not changed that much And when the people behind come up again Then you will face a It seems that there is a certain crisis at the age of 35 But the hardware is not like that Because of the hardware

You go from defining a product to making it today 9 months to 1 year is common Some complex products may actually take one and a half two years or even three years So what does this mean It means it may take a year or two for you to fight a war So Of course AI will speed up this process But it won’t speed it up to, say, three months.

So in this process In fact, you will find that you are doing it for a long period of time In fact, they can continue to grow and your value is getting higher and higher So you most likely don’t have to worry There seems to be no work today So that’s what I’m talking about Maybe that’s it This is a lower limit thinking This is a lower limit thinking Yes But let’s be honest today

In fact, you have to understand first The person who finally broke the upper limit The ratio to him must be very, very small For most people Today, in fact When you consider the upper limit You must consider your own lower limit right And I am actually in the field of large models Talents that can be absorbed In fact, it is also very limited Today you will find A successful model team Maybe it can't be too big

So to put it this way The whole world because I have always believed There are many, many creators in this world It’s not that There are only a few particularly creative people The world is determined by them We believe There are many creative people in the world today So these people Actually deserve one today opportunities they create What do you think about consumer electronics?

What kind of battle are we fighting?

Actually I think Don't think about it with a war mentality Today because of the most essential In the end, we still talk about creating value for customers.

I think electronic products today may be in the products around us In the visible future The platform that creates the most incremental value In addition to models In addition to these virtual things We say that in these actual It’s in the physical product In the visible future It can create the most incremental value Do you think hardware is an extension of large models How do you see its relationship with large models

I think models are actually the perception we talk about The brain of planning and control is OK Hardware is your human hand today Your eyes A lot of the actions you performed Body So today that You can’t just have a brain You must have a body And your body will actually follow your brain will evolve greatly

Model companies should not be able to cover this area.

As far as the extension of the model is concerned, it cannot be swallowed here.

Yes, because today actually the model is naturally in First of all, we just said that the model is hierarchical.

There is a high probability that in the future there will not only be trillion-level models but also billions and tens of billions-level models.

and millions of models These models Today it will be in distributed hardware It means deploying organic chips layer by layer to collaborate.

Yes, so I think the final world It won’t be controlled by a super brain But both OpenAI and Doubao have made mobile phones Right?

They also entered the field of hardware Regarding the leak that everyone just saw on OpenAI’s mobile phone recently Then I think in the end it’s so-called occupying the entrance In fact, I think it’s not just about occupying the entrance In fact, it is through the entrance to record and collect people's data because you want Today’s mobile phones

carry almost most of our work personal-related data Then for OpenAI today I think there is a high probability that what he wants is not that mobile phone.

What he actually wants is your your data your memories For such that its entire model Only in this way can we provide you with the best service And today Mobile phone company There is a high probability that they will not be willing to put such data Just open it up Right Shenzhen Hardware Company Today Is anyone training a particularly large model?

Actually I think Mobile phone manufacturers are making their own large models Mobile phone factory Mobile phone factory makes its own large model The second is what we call an embodied company There is a high probability that you will need to make your own model.

They also make very large ones Embodied models are all at the billion level today That’s 3 billion 7 billion It is more like a model of the cerebellum Yes Recommend a book of life to our audience Do you have one I would recommend a book called The Leadership Ladder He was written by a guy named Ram Charan Ram Charan is the head of leadership development at GE In other words

He is very interested in GE as a company How to cultivate talents Then there are actually decades of experience and his insights and thinking What about the book on the leadership ladder he wrote?

In fact divides the company into 6 different leadership ladders What should people on each ladder do?

What not to do It’s written very clearly I think this matter is important to me today It seems like hey What should I do in life What should I not do It is a huge guide and it brings a psychological explanation Our studio is called Language is the World Studio when you hear this name what are you thinking I think language is actually a In fact, language is very interesting You will find

Language is something that is very consistent with Tao The translation of Tao in English is TAO T We call it task It’s the task you have to complete today A is an action one by one It’s every word you say today Just actions one by one And what about O It’s because of the arrival of this action today Then the world has changed your observations of the world so

In fact, what I just talked about is actually something we have fun with ourselves A discussion and a realization Then every time you speak the single word you say The word itself has no meaning Or the word itself may have many meanings The reason why it produces a precise meaning and its world changes It is added behind the bunch of words spoken before It creates a new meaning so you find In fact, text and language

It naturally allows you to observe that under a task How does each token affect the entire state?

and this state What kind of token do you need below?

It's a way to start with a mission to derive What is the next action from a state and how does the next action deduce the next state?

So what I just said In fact, students who are working on robots must be particularly understandable Because there are two models today VLA model In fact, it is from the previous state To deduce what the next action should be and the world model is from an action Deducing what the world will look like next state So the reason why there are these two models is that there is no data today data like language

It allows you to observe the change from O to A and the change from A to O at the same time, so the language is a Just mentioned under the task A and O are an effective whole So in fact it is something very consistent with Tao So we think that in the future robots will be made There must also be data Then we can clearly define

the two-way cycle of A and O and finally train a model This one is a model that is very consistent with Tao is a model that is very consistent with TAO It should be a foundation for future embodied robots Yes Because on our bookshelves today Played a vinyl record of Interstellar Do you like watching this movie Oh, I really like the movie Interstellar Why

I think it is actually It is when people face huge challenges Human nature is glorious Of course there will be some negative human nature Of course there will also be very positive ones Those humanities that are very glorious today I think it is a In the face of a huge challenge Such a human nature

A very beautiful display Yes In today’s moment of AI revolution Do you still feel like the feeling described there?

I think you still have to believe in human nature No matter how big the changes are today, The good side of human nature Including the people Those who were the last creation In the end, it will bring us safely to the next era Yes