Harry Wang is the founding partner of Linear Venture, a $100m VC fund that specializes in early stage technology investments related to China and greater Asia area. Harry has led investments into more than 40 technology companies, mainly focusing on data and AI-related firms. Harry has served as a CEO advisor to Baixing, Dianping & CSDN. Prior to founding Linear Venture, Harry was head of Facebook’s risk engineering team and led efforts on payments fraud control, and was one of the first engineers working on social ads and newsfeed. Harry received his MSE from Stanford and has a CS degree from Zhejiang University.
With Harry we first discuss his personal story, including moving from China to the U.S. and working at Facebook during its early years, to why he returned to China as a venture investor. Harry shares his insight into Linear Venture’s core investment areas, its sourcing and deals strategy, his views on Chinese vs. US tech leadership and strengths, and of course – if and how might Facebook enter the China market.
Transcribed by Shaolong Lin and Michael Wen; Edited by Jordan Schneider and Caroline Chen.
Link to SoundCloud (here), also available via Apple Podcasts, Google Play, etc.
[Editor's note: this interview has been edited and condensed for clarity]
From China to Facebook
Adam: Harry, thank so much for joining us today. From China to the U.S. to Facebook and back, you have had an incredible variety of experiences. Can you tell us a bit more about what you've done?
Harry: After finishing my undergraduate degree at Zhejiang University, I went to Ohio State University first for my PhD in Computer Science. At the beginning I was working in data mining. After one and half years of studying there, I spent a day at the Chicago options exchange . This experience opened my eyes to something different other than academic computer science: the application of it, rather than its theory. So, after that experience, I decided to quit my PhD.
After almost half a year, during the second half of 2005, I applied to Stanford. Around September, the economy had a turn, and it became much easier to get a job. I got job offers from Yahoo, Oracle, and I also was admitted to Stanford as well. So I started working at Yahoo while studying Management Science and Engineering at Stanford.
After a year, I began working at Facebook as one of its earliest Chinese engineers. There were about 100 people at that time in early 2007. I joined the ads team and was in the team that built the first version of social ads along with three or four backend engineers. Later I helped build the first version of the newsfeed backend. I worked on many different products, including a Facebook virtual currency product, credits, payment products, a bunch of frontend and backend products. Wherever project needs you, you have to be ready. It's the typical way of startup engineers.
After about two and a half years, I started to concentrate on things related to payments. In my last two years at Facebook, I managed a payment security team. My team analyzed every payment Facebook received and checked every transaction to calculate the probability of a transaction being fraudulent.
After four and a half years at Facebook I decided to leave the company, and the reason was quite simple: I thought the fun times were over. But I was wrong! I thought Facebook had become the king of web social networks. But not of mobile. I left at the end of 2011, and I think starting mid-2012 Mark was really keen on changing Facebook from a web-first company to a mobile-first company. It took Facebook an entire year to do that conversion, and by the end of 2013, Facebook was clearly a mobile-first company. To me, I was at the peak of ‘fun time’ before leaving Facebook, and if I stuck it out for two more years I could have had another enjoyable run (by participating in the conversion from web-first to mobile-first). But I wasn't really patient at that time and I decided to quit and do something on my own. Facebook rewarded their early employees very well, so basically we didn't have to work as much at that point. So I quit and tried to figure out what I wanted to do and for that reason, I came back to China. I always wanted to come back, but were it not for that triggering event, I wouldn't have bought the ticket.
Becoming an Investor
I came back and talked with a bunch of Internet active players like Qihoo360, the founder of Alibaba, the founder of Netease, and the founder of Dianping. Later I became the CEO advisor of Dianping for about a year and a half, and I advised the CEO of Baixing.com for the same amount of time. I had these sorts of experiences only because I came back to China.
I wasn't really sure if investment was the right thing for me, since I was an engineer. A very good engineer, but it doesn't mean I would become a good investor. That's totally different career path as far as I can tell from my experience in the past five years. After doing all that, I met my partner at that time, Michael Zhang, who had just quit his job as a VP at JD.com. I also met with another celebrity investor Xue Manzi, or Charles Xue. We three teamed up and built an investment firm using our own money. We invested in about 8 companies using that SPV, the most successful and famous one of which is Dianrong.com (点融网). That's how we gradually got into investing, and eventually in September 2014, we decided to raise a professionally managed fund, and that's how Linear was created. We started from there, and we did all kinds of angel rounds and some pre-A round investments. About 60% to 70% of those investments from Fund I are what we call data-intelligent, which is the application of big data technology and AI technologies. We wanted to focus on one thing, something that we can do much better than all our potential competitors.
Our typical investment is something like Horizon Robotics (地平线机器人). We were their very first investors. Sensors Data (神策数据) is another example, and I am sure they will become a billion-dollar company. We were their very first angel investors. We were Tong Dun’s second-round investor, and put an angel investment into Rokid, one of China's top smart home AI products so far."
How to Find Talent
Adam: What is your approach to finding talent? How do you source deals, what are some criteria beyond the fact that they play in data intelligence and AI?
Harry: You can utilize some strategies. It’s not easy to reach people at big firms. Analyzing how we did our initial investments, I wrote a lot of articles and even a book (How Facebook was Engineered) which sold well. The book was mostly bought by engineers or PMs (product managers).
So first we made an impression on the market, and by the time entrepreneurs really want to get out and fundraise, either they had already heard about us and reach out directly, or they get referred to us by friends. It's not like we’re investors so far removed from them that they have to learn about us from scratch. Working with us is like working with someone they know already, since they know how we think because they read the book. I participated in quite a few events as well, which were packed with engineers and product managers. They’re what we call "doers with dirty hands.” I'm very interested in participating in those kinds of events as well as sponsoring them. This is how we tried to maintain our focus and impact on certain types of groups. Referral is definitely very important, referral from people we know, people we respect and people we've already met.
Adam: So how do you close your deals?
Harry: It's not easy. We typically find us being competitive on two things. One is we always have a better understanding of the problem than our competitors given our background. We try to emphasize that. We talk a lot about the details and the problems they are trying to solve and give our perspectives. It has to be supported by reasoning and data, but to be honest, in most cases, the scale of the code we have written before is more than that of most of the companies we have invested in.
From that point onwards - I am not saying that it is always like that - but from that moment onwards it builds a foundation for us to try our best to understand each other. We really get to a very deep state where we truly understand what problem they are trying to solve, how they are going to solve it and how we can help them on solve that problem. We have a reputation on the market. Every company Linear supports must have great tech. I’m not saying that we are the best on judging business models – that may be another VC’s strength. We’ll spend less time on discussing business problems and business models on how to make money.
Adam: When you reflect on China over the past 5 to 10 years, a lot has happened. I think for a while there were a lot of great opportunities because with mobile, a lot of things could be built on it and China was able to emulate the West. But these days, there is a lot of focus on artificial intelligence, frontier technologies, AR/VR. It seems that now China and the U.S. are converging. Being a frontier technology investor, what are you looking at and what are some of the key trends you've seen?
Harry: Regarding key trends so far, technology innovations and application innovations are very different. From our point of view, we spend 70% of our time and energy on AI processing applications where the technology is more established. But application is relatively new, for instance trying to integrate regression models to Fintech problems by separating the bad customers from the good ones.
I think infrastructure innovations right now are largely underestimated and under-invested in in China and in the US. Now in the US you can see big companies strongly investing in those areas, including NVD for sure, Google, Intel, all those big guys. But in China, those big guys are not the most active investors. It's more startups that pick up those chances, like Horizon Robotics and Hanwuji (a Chinese Academy of Science related startup doing AI chips).
Copying from the US is lazy. On the other hand, using logical reasoning to understand the problem takes a lot of hard work.
Adam: Another thing that could happen is a lot more cross-border firms might want to invest in Chinese companies because the technology is more comparable these days. Perhaps the speed in Chinese could be a lot faster while the salaries may be a little lower.
Harry: You know if you hire somebody from the US back to China, I think the salary range is between 70% to 150% that of the US. It’s not cheap in China anymore! It actually means there could be higher salaries here given greater demand and greater potential market upside.
Developments in AI Technology
Adam: We hear a lot about how Chinese AI is getting better and better. But in terms of some areas like chips, semiconductors, GPUs and quantum computing, I heard that it's hard for the Chinese to catch up because they require a lot of investment and development. So what are the areas where China is lagging behind a bit and needs to develop further?
Harry: Chips, definitely. But the interesting thing is that China can hopefully have a shortcut. The talent is there (e.g. software and deep learning). But just like in hardware design and hardware making processes where there are expensive mistakes, those people are extremely limited and they are mostly concentrated in the US. So it’s a question of whether you can hire these people to China or whether you can buy certain smaller companies with experience.
China has capital advantage. The capital is at such a surplus on the market. Maybe we can hopefully combine those upsides to try to create a shortcut in that area. And the B2B, at least on the market side, it's picking up. I think it's still behind the US in several aspects. One is the willingness of businesses to pay for software-based services. The second is the accumulation of the understanding of key features. We always complain how hard Oracle software’s user experience is. But they have hundreds of features that you can pick and assemble to get what you want. Although it's hard to use, it is there. I think a lot of Chinese software is easy to use. But if you look at the feature set, it's so limited.
From the soft layer side, I think engineers with both product sense and business sense are in less supply in China than in the US, especially in Silicon Valley. Silicon Valley trains engineers with ideas to think big. China is more realistic, and in China a lot of the entrepreneurs are more business oriented. But it has been shifting significantly to technologists leading the ventures. So we hope to see more and more motivated, well-trained engineers and product managers try to do startups.
Adam: So for those talents, are they trained in China and developed here, or are they people who go overseas to study and work like you and then come back?
Harry: I think both types are increasing. From an absolute perspective, China has a huge population. Since China’s internet industry has boomed over the last ten years, a lot of engineers have been trained domestically. And since the market is booming, you are seeing a lot of returnees who are well-trained at good schools and good companies in the US - they're coming back as well. I would say that the percentage of locally-trained talent is increasing, while if you look at ten years ago, most were trained outside the mainland.
Facebook in China?
Adam: Well it's good to know that the quality of domestic education and talent here is increasing. It's a really good development. We also have a burning interest in your experience at Facebook and their approach and interest in China. Obviously, we all know that Mark Zuckerberg speaks a little Chinese. He's come to Beijing a couple of times and even ran in Tiananmen with no mask.
Harry: Mark is a great person. But in my opinion, regarding the likelihood of Facebook entering China, I'm really pessimistic - I don't think that's going to happen. There is so much downside and almost zero upside for the Chinese government. The Chinese government is also really realistic. Zuckerberg is one of the friendliest top CEOs to the Chinese government. And that matters. And the Chinese government will return the favor, but not on such a sensitive topic.
The biggest possibility is Instagram with filters. Instagram is easier to filter, and it's not about people connecting - it's all about photo sharing. It's easier to control potential damage there. And also WhatsApp. My opinion is that after the party's political conference, they might reopen it again (*Editor’s note: WhatsApp was unblocked after the 19th Party Congress). I think Chinese government is very pragmatic. They don't have evil intentions, rather, they just want to make things as stable as possible.
Chinese Companies Expanding Overseas
Adam: To follow up on that, WeChat is huge in China. A lot of users are Chinese too. Does WeChat want to go abroad and do more stuff overseas?
Harry: I don't think so. I don't think they will ever succeed going overseas. There are several reasons. The way WeChat built the features, the way they do account protection, the gating, is all so Chinese. WeChat also has many features which makes it too complicated for a new foreigner to start using it. If foreigners use it, they will compare it to WhatsApp, Facebook Messenger, and a bunch of others. WeChat has to be simple but have enough differentiation to make foreigners willing to try it. Without that, I don't think it will happen. The third thing is the distrust of Chinese-made products. It's there. I'm Chinese. I don't want all my data is collected. It's a never admitted fact.
Adam: Eventually though, lots of Chinese companies want to enter other markets. How do you see that happening?
Harry: I think maybe in the next 10 years Chinese companies going overseas will become a trend. But in my opinion, they should try to replicate the strategy in reverse. In a certain way, it's like reverse copying the strategy taken by Vivo and Oppo. A lot of Chinese people initially thought Vivo and Oppo were foreign brands, better than Apple even, and certainly better than any domestic manufacturer. It's about the marketing they do or their coolness factor.
For Chinese products entering foreign markets, the idea is to make consumers think this isn’t a Chinese product. China can make good quality products - digital, real physical goods. In my opinion, the user experience of WeChat is much better than that of Facebook Messenger and WhatsApp for sure. Because I’m used to it, I know that WeChat cares about its Chinese users. Companies going overseas need to apply that carefulness, attentiveness to how foreign customers use your products. Chinese companies have that capability. But you have to first localize the product. Second, the brands should look local.
The opinions expressed in this article are Harry's own and do not reflect the views of Linear Venture