Episode 3: Kai-Fu Lee’s Journey From Google China to Sinovation Ventures

Kai-Fu Lee’s Journey From Google China to Sinovation Ventures

GGV Capital’s Hans Tung and Zara Zhang interview Kai-Fu Lee, the founder and CEO of Sinovation Ventures, an early stage VC fund in China, and a renowned computer scientist known for his work in artificial intelligence. Kai-Fu was the founding president of Google China, and played a key role in establishing Microsoft Research Asia as well. Kai-Fu recounts his journey from Taiwan to Tennessee, from a computer science student working on speech recognition to one of the world’s experts on artificial intelligence, from helping US giants like Google and Microsoft expand into China to helping Chinese entrepreneurs as an investor, mentor, and thought leader.


HANS TUNG: Hi there. Welcome to the 996 podcast brought to you by GGV Capital, and co-produced by the Sinica podcast. On this show we interview movers and shakers of China’s tech industry, as well as tech leaders who have a U.S. China cross-border perspective. My name is Hans Tung. I’m the managing partner at GGV Capital and I’ve been working on startups and investing in them, both in the U.S. and China for the past 20 years.

ZARA ZHANG: My name is Zara Zhang. I’m an investment analyst at GGV Capital and a former journalist. Why is this show called 996? 996 is the work schedule that many Chinese founders have organically adopted. That is 9 a.m. to 9 p.m., six days a week.

HANS TUNG: To us 996 captures the intensity, drive and speed of Chinese internet companies, many of which are moving faster than even their American counterparts.

ZARA ZHANG: On the show today we have Dr. Kai-Fu Lee, who is the founding president of Google China and a renowned computer scientist known for his work in artificial intelligence.

HANS TUNG: Kai-Fu is the founder and CEO of Sinovation Ventures an early-stage VC fund in China that he started in 2009. He was born in Taiwan and emigrated to the U.S. at young age. He has worked as an executive in Apple, SGI, Microsoft and Google. He played a critical part in establishing Microsoft Research China which will become Microsoft Research for all of Asia. That was a game changing event in the field of tech for China. Then he became the founding president of Google China, and then a few years later in 2009 he left Google to start his own VC fund.

ZARA ZHANG: Kai-Fu obtained a bachelor’s degree from Columbia University where he gave a commencement speech this summer, and a Ph.D. in computer science from Carnegie Mellon. Besides being a prominent investor and advocate for artificial intelligence Kai-Fu is also well-known in China as “Kai-Fu laoshi”, teacher Kai-Fu, a career coach and influencer with over 50 million followers on Weibo, as well as his own WeChat official account, and he has published five books, and frequently gives speeches at universities across China. Welcome to the show Kai-Fu.

KAI-FU LEE: Thank you. Thank you for inviting me.

HANS TUNG: First question, we’ll start with the early stage part of your life. Can you tell us when did you leave Taiwan? What prompted that? What kind of experiences you had when you first went to the U.S.? Was there any adjustment period for you back then?

KAI-FU LEE: It was I think in the early 70s when my brother, who was a Ph.D. in the U.S., came to visit my parents and our family after many years of being absent. And he saw that there was a large gap that the U.S. had in its education system compared to all of Asia, and he said, “Why don’t you come live with us?” So I did. It was late 1972.

HANS TUNG: Older brother?

KAI-FU LEE: A lot older, yes, more than 20 years older.

HANS TUNG: 20 years? Wow, okay.

KAI-FU LEE: So he was in Oak Ridge, at Oak Ridge National Labs in Tennessee, which I went to and lived in. It was really quite an experience.

HANS TUNG: You lived in the South in the 70s.

KAI-FU LEE: In the South, and I was the only Chinese student in an entire high school which is probably something that no one ever thinks of nowadays.

ZARA ZHANG: Did you speak good English back then?

KAI-FU LEE: I didn’t speak any English when I went, but I think with the typical southern hospitality I was really welcomed with open arms. And in fact the Oak Ridger, the newspaper of the small town of Oak Ridge, recently had a three-part series interview with me, so I really treasure the days, and I think that people were very warm and wonderful.

HANS TUNG: Interesting.

KAI-FU LEE: And the teachers were very patient and let me get up to speed with the language.

HANS TUNG: Did you encounter any difficulty at all?

KAI-FU LEE: Well, of course, there are always some people unfamiliar with the Asian culture.

HANS TUNG: Or orient.

KAI-FU LEE: Yes. So there were maybe a few, but very interestingly there is a lot more warmth, and a lot of people reacting when occasionally someone picks on me, and I didn’t really know they were picking on me. There would always be some other warm student who would come to my defense, so I never had any real serious issues or problems.

HANS TUNG: So how many years of school did you do in Tennessee?

KAI-FU LEE: Six years.

HANS TUNG: Six years?

KAI-FU LEE: Yeah, middle school and high school.

HANS TUNG: Wow. But your Chinese is amazing, so how did you continue to have such depth and grasp both languages?

KAI-FU LEE: When I first went to the U.S. my mother said, “I have two wishes, two things you must do before I let you go. First you have to marry a Chinese wife,” which I did. Not commenting on whether that’s necessary for others or not, but I did as she wished. The second was she said, “You need to write me a letter every week.”


ZARA ZHANG: In Chinese?

KAI-FU LEE: In Chinese. And then she corrected every mistake on the letter. At the time phone calls were expensive, there was no e-mail, so it was all based on this one letter from me to her. She wrote the letter back, correcting my previous letter, and then that helped me keep up my Chinese.

HANS TUNG: That was my problem. My mother moved there with me, so I never had to write anything to her.

ZARA ZHANG: But did you have anyone to speak Chinese with in Tennessee?

KAI-FU LEE: Well, my family, so my brother, sister-in-law, and their child.

HANS TUNG: Got it. And then you went to New York.

KAI-FU LEE: I did.

HANS TUNG: Very different city. Columbia.

KAI-FU LEE: Columbia, right.

HANS TUNG: What prompt you to choose Columbia?

KAI-FU LEE: Well, Columbia is a very famous school.

HANS TUNG: Of course.

KAI-FU LEE: And I applied to a bunch of famous schools. I mean being of Chinese background you’d understand, you go to the highest ranked one that accepted you, so I just tell the truth.

HANS TUNG: So you weren’t in New York after all.

KAI-FU LEE: New York was a positive factor, but I think I was still very much Chinese mentality. You go to the ranks and say, “Okay, Columbia is ranked number seven. Since the other six didn’t accept me I will go to that one.”

ZARA ZHANG: Did you know you were going to study computer science when you went there?

KAI-FU LEE: I did not. I enjoyed computer science a lot in high school. I actually got exposed to it in high school.

HANS TUNG: Played with the mainframes?

KAI-FU LEE: I did, yes. IBM, equivalent of 360. So when I went to Columbia I was going to study math, because in Tennessee I was the state winner in mathematics. I thought I was really good, but it turned out Tennessee was a small state, so when I was at Columbia I did not do that well. I was placed in the most advanced math class of seven people, and I was probably number seven in the class. And the teacher gave me an A minus nevertheless. He said, “You get an A minus just for getting into the class and the effort.” So I realized I was no math genius, I was pretty good at it. but then I also discovered computer science which was a lot of fun, and I thought I was a lot better at it.

ZARA ZHANG: Do you remember the first time you came across a computer?

KAI-FU LEE: First time it was actually at the University of Chicago. It was during high school, sophomore year. I went there to study number theory, as a special math class. And then they gave me an account. I barely knew how to program, I mostly played games on it. That was a lot of fun.

HANS TUNG: But your brother didn’t show you how to program a computer?

KAI-FU LEE: No, he didn’t know how to program a computer.

HANS TUNG: What did he do?

KAI-FU LEE: Well, he was he was a biochemist.

HANS TUNG: Oh, got it. Very interesting. And then after Columbia what came next?

KAI-FU LEE: I went to Carnegie Mellon.

HANS TUNG: Straight to school.

KAI-FU LEE: Straight to school, to the Ph.D. program. I was fortunate to have had a professor, John Kender, who was a CMU grad, teaching at Columbia. He liked me a lot and recommended me, so I got into CMU, and that was the best school, and I was very excited.

ZARA ZHANG: So when was the first time you heard the term artificial intelligence?

KAI-FU LEE: It was in my sophomore year at Columbia University.

HANS TUNG: That early?

KAI-FU LEE: It was that early. It was in a course called Natural language processing. I was taught by a professor who was a student of Roger Schank who was one of the early cognitive science-based natural language experts. So we built systems using Lisp, and actually it was a dialogue system. It’s what we would call a chatbot today. So we actually built a chatbot for my 1981 class project.

HANS TUNG: When I was in undergrad we also heard the term artificial intelligence. That was in the late 80s, but people tend to think that’s a field has been around long time, and it takes a while for a breakthrough, a lot of theory, and who knows when will it come to reality. In your mind what changed? What were the few critical factors that changed that made AI so popular and so prevalent today?

KAI-FU LEE: I think the biggest factor is really the maturation of algorithms. Even though there were multiple ups and downs, and there were times when neural networks were hard. I used Hidden Markov Models for my Ph.D. thesis which were the same class as neural nets, but they kept improving. And even though there were periods of downtime, deep learning, reinforcement learning, those algorithmic improvements and the people who continually worked at it, that’s the most important factor. Of course, the other almost as important factor is the availability of huge amount of data.

HANS TUNG: The internet companies and software companies create so much data.

KAI-FU LEE: Exactly. A lot of people think it’s all algorithms.

HANS TUNG: You need to train them.

KAI-FU LEE: Yes, you need to train them with large amounts of data and you need to tune your algorithm based on the amount of data you have. So you get more data, you improve your algorithm, you get more data, you improve it further. And then that iterative cycle is what drove today. And as you said, who has the most data? Well, it’s the internet companies. So in the U.S. it’s Google, Facebook, Amazon. In China it’s Baidu, Alibaba, Tencent.

ZARA ZHANG: So when you finished your Ph.D. were you planning a career in academia?

KAI-FU LEE: I was. I stayed at Carnegie Mellon. They made me…

HANS TUNG: The youngest assistant professor.

KAI-FU LEE: Yes, I was a research scientist and then assistant professor, skipped the postdoc. They really made an exception because my Ph.D. at the time was considered an important breakthrough in speech recognition. Of course, in hindsight it really wasn’t anything close to good enough, but I stayed and taught for two years.

HANS TUNG: And what lead you from there to work in the industry? How tough was it to leave academia to do that?

KAI-FU LEE: Well, academia really wasn’t quite for me, so I felt all the work you need to do to get grants, get student, was a lot of busy work without the corresponding ‘aha’ moment when you get to do exciting research, do something exciting. And actually more importantly, the University wasn’t a great place to make things happen. You write papers. It’s still kind of ivory tower, still academic. I wanted to change the world. And, of course, it was Apple that called, and it was a similar style of recruiting, they said, “Look, do you want to spend your life writing useless paper or do you want to…”.

HANS TUNG: Join us and change the world.

KAI-FU LEE: Essentially it’s the same things that Steve Jobs said to John Sculley about selling sugar water. For me it was useless papers.

ZARA ZHANG: How big was Apple when you joined?

KAI-FU LEE: I think it was about 3,000. It’s been a long time, but I think it was 3,000 people strong.

ZARA ZHANG: What do you think was your proudest achievement when you were at Apple?

KAI-FU LEE: Well, my proudest achievement was getting on Wall Street Journal and Good Morning America when we demonstrated our speech recognition, which still again did not work. It was the closest to a working system because we had a great team of engineers that made my Ph.D. algorithms real-time. It was really, really fast, faster than anybody else, but it wasn’t good enough. But nevertheless Wall Street Journal covered it, the stock went up a few points.

HANS TUNG: When was that?

KAI-FU LEE: It was in 1992, 1993.

HANS TUNG: That’s when I first remember hearing about you from reading the newspaper covering what you did.

KAI-FU LEE: Yeah, I’m sorry that didn’t work as well. If you were looking forward to buy a Mac you could talk to. That product did hit the market. It was not a failure but it was not a success.

HANS TUNG: How long were you at Apple and what took you from Apple to join SGI?

KAI-FU LEE: I was at Apple for six years and then towards the last two or three years it was just continual rounds of layoffs. I think from a career development point I learned a lot about adversity and how to consult people, managing teams in good times and bad, and also I got a bunch of promotions because my bosses kept getting either fired or left. So I went from being a manager to a VP in just I think about two and a half years. So it gave me a chance for accelerated growth. One could also argue it was too fast and I had to go back and relearn in other companies. But ultimately the iterative rounds of layoffs were just too much and the SGI reached out and I thought…

HANS TUNG: That was a fast riser.

KAI-FU LEE: That was an exciting company, and little did I know I was jumping out of the boiling pot into the fire.

HANS TUNG: How long were you at SGI and what was your proudest achievement there?

KAI-FU LEE: Just two years. We actually built the first set of web servers, so I went to a totally different business. It was with a profit and loss. We built a 3D technology for the web, and then we built web servers. The web servers were a big success and made a few hundred million dollars, and then the 3D for the web was too early. It was called VRML, virtual reality in 1996, way too early.

HANS TUNG: And from there you went to Microsoft.

KAI-FU LEE: Yes. At the time I was actually ready to give up. I felt like every company I went to ended up getting killed by Wintel, Wintel was Windows and Intel for the younger listeners.

HANS TUNG: They were the 800-pound gorilla. They were just crushing everybody else.

KAI-FU LEE: They were crushing everybody else. So Apple was facing the market share issue and Apple couldn’t deal with it. Wintel products were almost as good and much cheaper. And SGI the same thing.

HANS TUNG: They were the first mass-market solution that just kept on eating up everybody who did the high margin stuff that was too niche.

KAI-FU LEE: Right. At first SGI workstations were the only things you can use for movie animation, but later Wintel became good enough. That caused me to say, “Okay, I’m just going to have to work for either Microsoft or Intel”, because that’s the Wintel duopoly. So I was interviewed at Microsoft and interviewed at Intel, got two job offers, and chose to go to Microsoft.

HANS TUNG: What made you to choose Microsoft over Intel? That was a very important decision.

KAI-FU LEE: Yes, it was.

HANS TUNG: In hindsight it changes your career.

KAI-FU LEE: Both companies offered me to head their research lab in China. And I think it was really my roots in software, and that I really didn’t know much about semiconductors. I felt I wanted to work in a company where my expertise is at the core of the company, as opposed to something peripheral. So Microsoft was a software company. I felt I was trapped in hardware company, at Apple and SGI, where software was evaluated purely as, “Can it help us sell our hardware?” And then Microsoft was maybe the first chance I could do software for real.

HANS TUNG: How come both of those companies were thinking about setting up an R&D center in China? What were some of the factors that were going on that propelled the management team to think about that?

KAI-FU LEE: I am guessing that one of them decided to do it first and the other heard. I don’t have any facts. I suspect Microsoft chose first, and then I think Intel said, “Hey, if Microsoft’s doing it we should do it too.”

HANS TUNG: Had Microsoft already have an R&D center in India at a time?


HANS TUNG: So nothing outside of the U.S.?

KAI-FU LEE: It was actually research. Core, pure, fundamental research, and both companies were doing that. And there certainly were no strong indication China would be an academic power at the time.

HANS TUNG: Exactly, that’s why it’s so surprising.

KAI-FU LEE: Microsoft only had Redmond and Cambridge, that’s Cambridge UK.

HANS TUNG: That’s right.

KAI-FU LEE: And China, Beijing would be the third.

ZARA ZHANG: Did you expect that you will be working in China when you joined the company?


HANS TUNG: You were interviewed for that position?

KAI-FU LEE: Yes, it was specifically for that position. I had met a lot of great Chinese employees, fellow students at Carnegie Mellon and Apple. I saw that the people in China were incredibly hardworking, very smart, very motivated, maybe didn’t have as good education as in the U.S., but I felt Microsoft could do something if the brand was strong, and we could hire the smartest people.

HANS TUNG: Being Taiwanese. Did you have any hesitation, or did anybody in your family have any hesitation for you to work in Beijing?

KAI-FU LEE: No, it was actually one of my father’s wishes on his deathbed that I would spend time in China, because my parents had emigrated from mainland China. Not emigrated, I didn’t use the right term, they moved during the civil war, with the Chiang Kai-shek. So my father had always wanted to return to roots. He hoped to move back himself, which didn’t materialize in his lifetime, but he thought I should try it.

ZARA ZHANG: How was your adjustment process when you first started working in the mainland, because you haven’t really lived in mainland before that?

KAI-FU LEE: Yeah, but it was a very open environment. I think all the universities academics were very curious. They were wondering the same questions you were, “Why would you pick China? It’s not like we are powerful academically. What do you hope to do? What is your goal? Is this a PR move?” And then we showed that, “No, we really want to bring the smartest students.” So we brought back from the U.S. some of or top researchers.

HANS TUNG: You did an amazing job recruiting a lot of people which ends up being very famous in its own right later.

KAI-FU LEE: Yeah. We were very fortunate there was just a wave of Chinese PhDs.

HANS TUNG: First wave.

KAI-FU LEE: Yes, for mainland China. I would say probably Hong Kong University of Science and Technology was the first greater China wave, but largely not mainland. But definitely MSRA was the first attracting these typically PhDs in their early mid 30s who were rising stars in their areas.

HANS TUNG: And almost all of them were born in China and went to the U.S. for grad school.

KAI-FU LEE: Yeah. The senior people were maybe 80 percent in mainland China. The junior people were 100 percent from Tsinghua, Beijing University, and so on.

ZARA ZHANG: And how did you first meet Google? According to a recent TV show they said that you met Eric Schmidt on a golf course.

KAI-FU LEE: Yes, I did. I actually went back to Microsoft headquarters and then when I heard that Google was starting its own Google China presence I thought I had the right combination, because Google wanted to do R&D, they wanted someone who understood China, someone who could communicate with the Silicon Valley culture, so I actually volunteered myself and sent an email to Eric Schmidt, and then we were arranged to meet on the golf course. Mostly because there were too many Microsoft people interviewing and they felt that I would prefer not to be seen by a lot of Microsoft people coming in and out.

ZARA ZHANG: So you were a great recruiter both at Microsoft and at Google. What what tips do you have for people who are trying to attract talent to join their companies?

KAI-FU LEE: Well, I think you have to find people with the right cultural fits, and you have to have a very high bar. You have to show the people’s roadmap in your company. And I think you will find that people who have the right culture and the background will generally find it to be mutual fits, because people will want to work with other people who are similarly minded, equally smart, that they can learn from, in a culture that’s nurturing, and also consistent with the values that they believe in. So I think part of it is you’ve got to pick out who to target and then have an incredibly important original nucleus of people who are magnets themselves, and then attract more people like them.

HANS TUNG: What were the top three points in your sales pitch to recruit people, these young PhDs in their early 30s, in the mid 1990s to move back to China from U.S.?

KAI-FU LEE: To Google you mean?

HANS TUNG: To Microsoft or Google.

KAI-FU LEE: To Microsoft, the top reasons were, “This is the only place that can do fundamental research with the freedom, and you will be helped by a team of super-smart young people who are incredibly hard working. You remember how it was for you?” And then Microsoft as a brand. For Google it was that this was a chance to really build up a business that could be phenomenal, that combined business engineering, everything. We have the empowerment of the headquarters. Second is the Google brand itself. At the time Google was clear and by far the most attractive company to work for.

HANS TUNG: It was in 2005, the year after it went public.

KAI-FU LEE: Exactly. And then the third would be all the wonderful, actually not so important, but incredibly important for recruiting, the wonderful benefits. We had an Executive Chef, we had lobster for lunch. We had all kinds of benefits. In the U.S. there was taking your dog to work, doing laundry for you, and all the other good stuff. So all those benefits actually did work in attracting people, even though I don’t think they were core as far as the long-term retention.

ZARA ZHANG: How would you define the culture of Google China back then in its early days?

KAI-FU LEE: We wanted to really have an independent mindset because I felt the only way Google China could succeed was to believe this is your own company, and go after it. And in order to do that we wanted to have the Google values, but have the freedom to really roll out new products, features, and iterate and show numbers in terms of market growth and also advertising revenue growth. So it was almost like a startup within the company.

ZARA ZHANG: And when do you think you had the first inkling that Google might need to leave China?

KAI-FU LEE: I didn’t. I think the Google business went up for the first three years that I was there. The market share went up significantly. Revenues were almost a billion dollars. The search share went up. Search share was about tripled during my tenure. But the last year I think there were some difficulties. I think it’s partly just due to Google feeling that, “Okay, it’s time to build one Google, and not have everything being different”, because all the products, features, a lot of code was forked, and people were questioning, “Hey, look Google now has expanded successfully throughout the world, and maybe China should just be like any other country.” So I didn’t have as much freedom as before and also I think it was the one year when Google’s financial numbers were not the greatest, so there was less budget for us to attract traffic, and usage, and marketing. So those were some of the challenges we faced. But also I think the bigger reason I left Google was not because of the challenges, but because I saw the opportunity. I saw that mobile was going to be a huge deal. Partly because being in Google I saw the power of Android. And then I saw that the entrepreneurial energy, the entrepreneurs and the VCs were growing in China. But there was a gap in the early seed stage funding, so I thought filling that gap would fit my technical background, and also be a lot of fun to help young people start their companies. So that was really the major reason.

HANS TUNG: The model you first started was the incubation model.

KAI-FU LEE: That’s right.

HANS TUNG: Now as you’ve evolved and added the VC component became bigger and bigger. As you evolved the model over the last seven or eight years, what were some of the biggest lessons and takeaways that you had evolving this model?

KAI-FU LEE: Our model today is basically a VC model. We invest in series A and B. We still do angel deals when we can. We don’t incubate, not anymore, but we would invest in early stage, pre A deals, when one comes along. I think we pride ourselves on being very technically knowledgeable, picking the right trends. And we still have a lot of that early spirit that we love our entrepreneurs. We have a very strong service team to help in terms of marketing, PR, finance, legal, and so on, so we have a big team that continues to help the entrepreneurs even though it’s series A and B. But the service mentality, the pro-entrepreneur mentality is the same. And then we have now an AI institute which is now implementing AI, but also helping our VC with deal referral, judgment, and so on. So we’ve evolved to a model that’s maybe a little bit like Greylock and Benchmark in the technical type, a little bit like Andreessen Horowitz in the strong services operations, and maybe a little bit like Google Ventures in having our own engineers. So we’re trying to make this model work.

The fund is doing well.

But in retrospect you’re asking me what things I learned that were not quite as I originally expected. I think it’s probably obvious to you, Hans, and many people in VC now, and myself too, but at the time I thought a nurturing incubation environment could help a lot of people fill up in areas in which they are strong, and increase their chance of success. However, I think in retrospect the strongest, finally successful entrepreneurs, are so strong, they’re so confident, I think they feel an entity like an incubator, or someone who watches over them like a teacher or even a boss, is something that’s highly undesirable. The great entrepreneurs want to be independent. They want to be funded. They want to be given input. But they need to be their own boss, and they’re big and strong, and powerful and confident. They’ll make mistakes, but their growth comes from learning from the mistakes, not from a teacher who guides them along the way. So that’s how we made our role shift into one that I think is more complimentary to people who are of this type of strong entrepreneur.

HANS TUNG: As you started backing different entrepreneurs, do you also see them being older than the ones you helped to incubate earlier?

KAI-FU LEE: Yes. The age changed a lot because in the early stages when we were doing mobile, that population was very young.

HANS TUNG: Mid 20s.

KAI-FU LEE: Yes. Because at that time PC people were in the mid 30s and they couldn’t change their mindset so easily in general. And so we backed young people, light, very lean startups, very low small amounts of money. And that very small amount of money was enough because it was three engineers. But now as we go into heavier styles where you have to think about new retail, new health, and think about new education, you actually have big customer acquisition costs, you may even have offline activities, and also the Chinese competition evolved differently from the U.S. I think U.S. continues to feel light startup is good, light business model is good looking.

HANS TUNG: Looking for the next Mark Zuckerberg.

KAI-FU LEE: Exactly. But in China the competition is so fierce that if you have a light model someone might copy you or ride the next wave. So you have to really build very big deep moats, and the moats often have to do with ugly offline and expensive capital expenditure.

HANS TUNG: Which requires someone who has that experience from previous startups or big companies.

KAI-FU LEE: Exactly. So the amount of money we put in has to be a lot more, even though it’s series A and B, and people tend to be older and more experienced with running larger organizations.

ZARA ZHANG: So I’m wondering what advice do you have for U.S. companies who want to enter China, especially given your experience with Google. Tim Cook that recently that you can’t change things from the sidelines, you have to participate, when he was defending Apple’s move to comply with Beijing’s regulations.

HANS TUNG: I have more to that question. The one side that have done well tend to be more traditional economy companies, whether it is a Starbucks, or a GM, or even an Apple, but the Internet guys have a much tougher time.

KAI-FU LEE: I think all foreign companies will have a tougher time than before. On the one hand China is a big lucrative market, everyone should try to enter and participate, but the good days when Procter and Gamble could come and take it all is gone. Starbucks probably was one of the last entrants that have the kind of mindshare they have. I think for the next ones it will be tougher because Chinese brand builders are getting better. So even traditional American companies shouldn’t take it for granted. Internet companies have it maybe a little bit harder. Apple is one example of success but you could argue they are not that internet company. So for an American technology company I would have two advices. One is if possible get a Chinese partner.

HANS TUNG: One of the big companies.

KAI-FU LEE: Well, some have also tried to get VCs to find the local entrepreneur, that’s another model. There are multiple different models. Joint venture, or maybe if you’re in certain areas maybe a state-owned company or a bank. It all depends. If you are an internet company maybe BAT was good. So I think you have to evaluate, but having someone who knows local that’s core, and that’s important. The other advice I have is if you have a very strong technology differentiation, consider selling the technology, consider license the technology, because that’s another way that China is now much better in respecting IP. I think technology licensing could be a very good and easy business. You don’t have to deal with the complexities of the local market and running in it.

ZARA ZHANG: What do you think is the most successful U.S. tech company in China?

KAI-FU LEE: Apple would certainly be one of them. They’re doing okay so far although I think they’re facing some challenges as well. Their share appears not to be growing, but I think they care a lot about the market, and they’re trying hard. I think actually some traditional companies have built strong enough routes, and they’re still doing okay. IBM being one example. They have a very strong brand, very strong local partnership ecosystem. Those are probably two styles to consider. But there are also hardware companies like Intel and Nvidia who I think are also doing okay, because they’re at that infrastructure level, so it’s a little bit easier. It’s a bit like licensing that I mentioned, so those are some examples.

HANS TUNG: Do you think more companies should look at what Yahoo had done with Alibaba, or Uber, where it took them a while to realize to do a deal with Didi. Do you think that more of them should be doing more of these investment deals?

KAI-FU LEE: Yes, I think so. I think the Uber deal was very shrewd to be done at the time that it was. I think the Alibaba and Yahoo was maybe a little bit lucky, because I think at the time it was hard to envision the way things would play out. But nevertheless I think Jerry saw Jack as an amazing guy.

HANS TUNG: When they known each other for a while it makes a huge difference.

KAI-FU LEE: That’s right. So I think it’s maybe well-informed from the people angle, but the business side is different. We did not see it play out this way but it was great for Yahoo.

ZARA ZHANG: You recently wrote a piece for The Economist about a phenomenon known as OMO, or online merged with offline. So that boundary between online and offline is being blurred. Do you think China is ahead of the U.S. in OMO, and which area of application are you most excited about?

KAI-FU LEE: Right. So online merges with offline is when you can capture everything in the physical world, digitize it and upload it, and combine with online. So when you’re shopping the fact that you picked up some products, or bought some products, that’s captured and combined with your online patterns. That could be done with just the online payment. It could also be done with a video camera or other sensors and mechanisms. The entire OMO is huge in China because we see it hitting a lot of areas including retail, including public transportation, including education. And also if you look at the amount of digital activity that’s happening offline China is way passing the U.S.

Online-wise China’s mobile users is maybe three or four times that of U.S., but in terms of offline mobile payments it’s much bigger. Offline mobile payments are off the charts, over 50X. So in China I think the biggest enabler, as much as we love AI and OMO, we think a huge enabler that’s already happened is the mobile payment. In China, over 600 million people can do peer-to-peer micropayment almost frictionless, almost without commission exchanges to each other. And that will empower many types of applications. So our investment in Mobike, some of the live streaming and food delivery applications, that’s all taken off because mobile payment will be the enabler that will make everything happen a lot faster. So I think the offline payment will unlock a lot of growth and accelerate it even faster than before.

So if you ask me to predict the future of AI it’s possible China or the U.S. will be ahead. But if you ask about the future of OMO, about the mobile payment driven online offline merging, creating value, there’s no doubt China will be the leader in this area.

HANS TUNG: One of my last questions then, what you did at Microsoft made a huge difference once you had to open letters, “gongkaixin” to a lot of Chinese tech students. Also you recruited a lot of people that end up becoming the future presidents of Baidu and CTO of Tencent and so forth. What made you capable of recruiting all these top people in the first place? And what do you think someone like Microsoft or any other Western company could have done to retain that talent?

KAI-FU LEE: Well, I think there is a place and time for everything. There was a time when Microsoft’s brand was the most amazing. There was a when Google was the most amazing. And there are times when BAT were a great brand, and now I think it’s the age of entrepreneurs. So I think it’s hard to fight against the trend. I think actually the entrepreneurial trend is the most powerful including the government’s support for the mass entrepreneur and innovation campaign, as well as I think that the parents and the young people themselves are thinking differently. So I think it’s hard to twist somebody’s mind and force them to still think some company is hot when it’s not. I think right now if you want to follow the trend people want to be entrepreneurs, they want to learn how to be entrepreneurs. And I think we’re in the fortunate position as a VC to be able to connect to those people and find the way to fulfill their dream.

HANS TUNG: My last question is, what you and I have seen in China with the internet space and the VC space in the last 15 years is nothing short of transformational for our entire society. Do you think that’s a China-specific phenomenon or do you think some of these models or companies themselves, could export these models to other developing markets around the world, if not even U.S. and Europe?

KAI-FU LEE: Well, I think the China model is very hard to replicate. I actually don’t think it can be replicated at anything close to China’s scale. Partly it’s because I think China have all the things happening at the same time, very hardworking industrious population, plenty of capital coming in, the willingness to take risks, willingness to learn from the U.S. at the right time, and the right revolutions that came along with internet, mobile, now AI. So it is almost a perfect storm, a perfect script designed for China to go from a poor country to manufacturing, now into tech. That cannot necessarily be replicated in other countries because the initial step of using cheap labor to gain a strong ground hold in the world may not be doable because AI is taking a lot of those jobs. So it’s not clear that having more unskilled laborers is a benefit to large population countries anymore. That’s the number one reason. The other thing is those perfect storm elements are just hard to realign for another country. So I think what’s most likely to happen is China and the U.S. will each become a very strong powerful corners of the world, with their software, technologies and know-how, and going into other countries.

HANS TUNG: The Chinese and American companies themselves will go to other countries.

KAI-FU LEE: Right. So that will happen in many different ways. Facebook is used to expanding their own presence with subsidiaries. In Chinese companies maybe BAT.

HANS TUNG: Or even Xiaomi.

KAI-FU LEE: That’s right, in China some of those companies are expanding themselves like Xiaomi. Some companies are doing investments like BAT. Some companies are doing startups, but still not leaving China like the total copycat to different countries, and the live streaming copycats, and the shared bicycle copycats. Those aren’t necessarily owned by Xiaomi or BAT.

HANS TUNG: But it started by Chinese entrepreneurs who understand what’s happening around the world.

KAI-FU LEE: That’s right. So I think in the end China and the U.S. will really play very big roles. The other countries are obviously the ones with bigger markets and more advanced infrastructure and they will still enjoy good growth, but it’s hard to really replicate China.

ZARA ZHANG: Who is an entrepreneur that you admire the most and why?

KAI-FU LEE: The entrepreneur I think would have to be Steve Jobs. I think he has the boldness to dream the impossible, and also learning to become a better leader. Now he’s obviously had a lot of issues and faults, and had to leave and come back, but I think he is the inimitable one.

HANS TUNG: What was the best and worst decision you ever made?

KAI-FU LEE: Best decision was probably I think I have to say marry my wife. But professionally moving to China. I think the worst decision is probably just generally believing speech recognition could work in the 90s.

HANS TUNG: That was too early.

KAI-FU LEE: So continuously trying that without learning from the previous lessons. But eventually I learned.

ZARA ZHANG: What’s a habit that you are glad you have?

KAI-FU LEE: Reading.

ZARA ZHANG: Thank you so much for your time. It was a pleasure to have you.

KAI-FU LEE: Thank you.

HANS TUNG: Thank you.

KAI-FU LEE: Thanks.

HANS TUNG: Enjoyed it.

HANS TUNG: Okay. Bye bye.

HANS TUNG: Thanks for listening to this episode of 996. By the way, we also produce a weekly e-mail newsletter in English, also called 996, which has a roundup of the week’s most important happenings in tech in China. Subscribers have told us it is informative and fun to read. The newsletter also features original content and analysis from Zara and me. Subscribe at 996.ggvc.com

ZARA ZHANG: GGV Capital is a multi-stage venture capital firm based in Silicon Valley, Shanghai and Beijing. We have been partnering with leading technology entrepreneurs for the past 17 years, from seed to pre-IPO. With $3.8 billion in capital under management across eight funds, GGV invests in globally minded entrepreneurs in consumer, internet, e-commerce, frontier tech and enterprise. GGV has invested in over 200 companies including Airbnb, Alibaba, Ctrip, Didi Chuxing, Hello-Bike, Houzz, Keep, Slack, Square, Toutiao, Wish, Xiaohongshu, YY and others. With 29 IPOs and 17 unicorns to date. Find out more at GGVC.com.

HANS TUNG: If you have any feedback on this podcast or would like to recommend a guest please e-mail us at 996@ggvc.com. This podcast is co-produced by our friend and business partner, Kaiser Kuo, the host of the wonderful Sinica podcast. It covers China’s economic, political and cultural issues.