Interview with Wang Xiao: What kind of enterprise is worth investing in in the era of pan-AI

Posted on 2021年7月30日

【Editor’s Note】

Artificial intelligence empowers various industries like electricity, profoundly changing human society. China is at the forefront of the third wave of global artificial intelligence development. “Algorithm Weekly” will focus on artificial intelligence “Shanghai Heights” and China’s new infrastructure, and continue to pay attention to the forefront of global AI.

“With the further development of artificial intelligence, AI will become invisible, and we will not even perceive the existence of AI. For example, when autonomous driving develops to a certain level After the stage, everyone will become accustomed to unmanned vehicles; in the meta-universe, we are in the state of digital survival.” At an investment and financing forum during the World Artificial Intelligence Conference, Wang Xiao, founder of Jiuhe Ventures, analyzed the trend of the AI industry. Said. “If you look at artificial intelligence in the two dimensions of the bottom layer and the application layer, the bottom layer is the data and technology platform, and the top layer is the industrial platform. The bottom layer is the changes brought about by data and digital intelligence, and the upper layer is the various scenarios. Opportunities brought by digitization.” From the bottom to the application layer, he dismantled Jiuhe’s investment “card position” layout one by one.

In an exclusive interview with the news (www.thepaper.cn) before the conference, Wang Xiao explained in detail the changes in the AI entrepreneurial ecology, the technology investment route of Jiuhe Ventures, and his optimism The focus of the track. In the venture capital circle, Wang Xiao’s label is “technical” and “engineer-type” investors. The technical background of one of the “Baidu Seven Musketeers” has also implanted unique technology genes into Jiuhe Venture Capital.

There is no shortage of drumming games in the VC field. But looking at the investment landscape of Jiuhe Venture Capital, Wang Xiao’s technical complex is quite heavy. His style is rational and rigorous, and his answers are very logical and rhythmic. Talking about the insights of early investment institutions on technology, he concluded that investing in technology must first understand the technology itself. “The core logic of Drumming and Spreading Flower is to rely on capital to promote the development of the project. As long as the financing is sufficient, the probability of winning is high. Long-termism The logic lies in believing in the essential laws of things, rather than using the power of capital to catalyze its growth in the short term. Jiuhe believes in technology and long-termism, one is because we know more about technology, and the other is that we want it to be in line with the development of things. It’s own rules instead of pulling out seedlings.” Wang Xiao, founder of Jiuhe Venture Capital

Wang Xiao, founder of Jiuhe Venture Capital

Entrepreneurship ecology in the era of pan-AI

As technical barriers and application thresholds continue to decrease, AI is becoming increasingly instrumental and normalized. In the era of pan-AI, AI entrepreneurship has also stepped into the deep water zone, constantly “sinking”.

According to Wang Xiao’s observation, in the field of AI entrepreneurship, the previous mainstream founders were pure AI algorithm technicians. But now, many core algorithms and training frameworks are gradually being open sourced, the difficulty of algorithms is gradually decreasing, and AI is becoming more and more infrastructure-oriented and tool-oriented.

The combination model of the AI entrepreneurial team has changed significantly. “To some extent, the current team founders may not be proficient in AI technology, but they know more about products or specific industries. The co-founders or CTOs of the team are AI experts. This is the result of pan-AI. It has become a technical language that everyone can understand, similar to English and computer programming.”

Algorithms are increasingly difficult to become barriers between AI companies and specific industries. The attributes of the combination and application industries are becoming more and more important. The generalization of AI application scenarios is also happening at the same time. “Originally, it was more in the “high” fields of finance and autonomous driving. Now AI has penetrated into the supply chain and industrial platforms, and has entered a variety of enterprise services. Scenarios In generalization, entrepreneurs’ awareness of this is also increasing, and the proportion of pure AI technology entrepreneurs is decreasing.”

He further analyzed that according to technical Features and problems to be solved, AI entrepreneurship has two different paths and ways of playing.

The first path is that in many landing scenarios, AI technical indicators are no longer the core difference, and the more core part of the solution is the industry’s know-how SaaS (software as a service), software, and data collection.

For example, in the field of enterprise services, after passing the technical threshold, whether the algorithm can continue to be optimized depends on the application of the scene and the ability to obtain data. Only when the scenario is clearly defined and rich high-quality data obtained from customer feedback can the optimization and rapid iteration of the algorithm be realized, and practical problems can be solved. The key to building a moat lies in industry-specific solutions, sales capabilities, and brand in the industry.

The second path is that in areas where AI technology is more difficult, technical reliability and safety requirements are high, and technology is still the strongest moat. Breakthrough AI technology can even create new application scenarios.

For example, in technology-intensive fields such as autonomous driving, startup companies need to have very strong technical capabilities. After self-driving technology has accumulated high enough potential energy, many new application scenarios have been defined, including parks, logistics, and automated driving of taxis.

Wang Xiao believes that overall, relatively few startup companies can take a “technological breakthrough” development path in the near future. The bottleneck restricting the development of AI companies often lies in the lack of clear landing scenarios, which makes it difficult to create unique commercial value for customers. In addition, when the AI craze in the past few years, many companies with AI concepts emerged, including pseudo-AI companies, which in essence did not have real AI technology. “After the ebb tide, these’pseudo AI companies’ exposed their shortcomings, unable to use technology to solve customer pain points, and will inevitably fail in the process of commercialization.”

In the era of pan-AI, the evaluation criteria of AI entrepreneurial teams by investment institutions have also undergone a relatively large shift.

In the past, investment institutions paid more attention to technical difficulty, the background of technical leaders, etc., and used technology as the core to make judgments. But now, it is not only necessary to evaluate the ability to use technology, but the proportion of “industry understanding” in the evaluation system has increased significantly. It is more important to see whether AI technology can create real value in the industry and whether the market scale is large enough.

Wang Xiao said that the Nine-Party Conference will focus on investigating whether AI startups can form a closed loop between technology, data, and scenarios. The four dimensions to measure the ability of AI companies to land include: the right vertical industry; the right time to enter; the ability to solve problems with technology; and the ability to build a sales team in a timely manner.

Six major tracks

Jiuhe Ventures currently manages four RMB funds and one USD fund, with an asset management scale of approximately RMB 3 billion , Has invested in nearly 300 early-stage start-up companies. The investment rounds cover angel rounds, Pre-A rounds and A rounds. IRR and DPI have always remained at the leading level in the industry. Its investment cases include Momenta, an autonomous driving technology developer, Smart Sales SaaS service provider Xunji Technology, image-assisted diagnosis AI medical platform Yingtong Airdoc, artificial intelligence customer service company Xiaoduo Technology, deep learning overall solution provider first-class technology, AI Drug R&D company Qingyun Ruijing, active noise reduction and mute solution provider Ansheng Technology, drone logistics technology company Xunyi Technology, and second-hand luxury goods e-commerce platform Red Bulin and other projects.

In 2011, Wang Xiao left Baidu and founded Jiuhe Venture Capital. During Baidu, he was responsible for technology and To B business, including products, technology, sales, marketing, operations, etc. He is also the head of Baidu’s internal product division, including Baidu Maps, Baidu Mobile Search, Baidu Mobile Input Method, Baidu Enterprise Search Software, etc. Ten years of Baidu experience has laid the foundation for his investment in technological innovation and business innovation after the establishment of Jiuhe, and also left a deep imprint on Jiuhe’s overall investment style: stepping out of the only technology theory, respecting the evolutionary route of technology itself, and paying more attention to technology Integration with industry and commercial landing.

Wang Xiao said that Jiuhe will reorganize and reshuffle the track every six months. This sense of rhythm is to “take a beat” faster than its peers. After sorting out in the first half of the year, the focus will be on the following six areas:

Frontier technology fields, such as chip or launch vehicle “hard technology”; in the context of carbon neutrality In the fields of new energy and new materials, Jiuhe has a layout in the fields of fuel cells, smart grid chip applications, electric trucks, etc.; in the field of biotechnology, especially in the field of bioinformatics, the combination of computing technology and biotechnology will bring breakthroughs Opportunities for industrial platforms, Shell is a typical case, and such platforms are likely to be born in various industries; fast-growing companies in the consumer sector; corporate services and SaaS companies.

Frontier technology often comes with high uncertainty. From this perspective, the most cutting-edge technology may not be the most suitable for investment. How to judge the right time to intervene in a certain cutting-edge technology field? Wang Xiao’s answer is that it depends on the maturity of the underlying core technology.

“The possession of these underlying core technologies actually takes time. If you can’t make a breakthrough, you can’t make a breakthrough. But once a breakthrough occurs, many scenarios based on the underlying technology The application-level technology advancement will be relatively fast. “Take bio-information technology as an example. When basic technologies such as gene editing technology and synthetic biology technology mature, capital entry has the prerequisites.” In the past year, we believe that bio-information technology Basically to a relatively good level, so I started to lay out in this respect.”

There are also underlying technologies that are still in short supply and still need patient technical direction. Wang Xiao believes that from an investment perspective, breakthroughs in core technologies will take time in fields such as natural language processing (NLP), digital twins, and new energy chips.

Rhythm: I don’t care much about whether the market is hot or cold

“Jiuhe may have a feature, that is, when we invest in the project, we don’t care about the hot or cold market. I won’t say that the market is hot so we invest, but the market is cold. So don’t invest less.” Wang Xiao has a clear personal style of technology investment philosophy. “I think investment is a very independent behavior. It must be forward-looking and independent. Everyone thinks it’s good. It’s very expensive. Others think it’s average, you think it’s good, and the price is high.”

Jiuhe Ventures’ unusual path is reflected in the early days I began to pay attention to the To B field and deepened the B-side layout step by step. B-end means more vertical, higher requirements for industry understanding and industry experience, and slower speed. To C companies rely on traffic to take off, but in the face of the industrial Internet, the iron law of traffic has all failed.

In the era of consumer Internet when traffic dividends broke out at that time, waiting for To B to grow is an “alternative” behavior. It wasn’t until the last four to five years that the domestic SaaS industry accelerated its heating up, and capital surges poured into the To B market.

Wang Xiao believes that in To B, his judgment logic has always been not whether the market is hot or cold, but whether it has value for the company. Enterprise service can improve production efficiency and enterprise competitiveness, and it has practical value. “At that time (ten years ago) the market’s perception of To B was that no one wanted to buy software services. The To B market was too slow and there were no large To B companies in China. This perception was a posteriori, meaning that there was no , There won’t be any in the future. This logic doesn’t work in my opinion. There was no before, is it possible now? This is what I want to judge.”

Slow There are also benefits of slowness. In his opinion, although To B and Industrial Internet are slower, the advantage is that there is a contract amount and a stable income. “It is nothing more than slower, but it will not die. As long as it survives three or five years.”

Although the willingness of local companies to pay for software services is improving significantly, compared with the US market where corporate service unicorns continue to emerge, Wang Xiao believes that there is a 10-year cycle between the two , There are still plenty of opportunities. “China’s software service industry still has 5-10 years of growth opportunities in the future.”

has a broad customer base and data, and can mine value from data; entrepreneurs are The industry has a deep understanding; the front end uses software as the service carrier, and the back end is algorithms and data-this is a rough portrait of the high-quality enterprise service company he described.

For investment institutions, it is much more difficult than choosing a track to choose a driver, that is, to choose a person.

Wang Xiao said that a simple understanding of the business of the track is to add, subtract, multiply and divide. . But human judgments are two orders of magnitude more complicated than business judgments.” In the field of artificial intelligence, Jiuhe’s measurement standards for entrepreneurs are firstly the understanding of artificial intelligence technology itself, and secondly the use of scenarios. Comprehension, “After these two conditions are framed, the selectable targets have been significantly reduced, and the overall success rate has also been significantly improved.”

domeet webmaster