Hikvision President Hu Yangzhong: The virtual and real of the security market

The security industry cannot be developed simply by investing money. Capital may have a momentary impulse, but the exit will also be very fast.

——Hu Yangzhong

In the past two years, Hikvision has "inexplicably" become a company that has attracted much attention from artificial intelligence practitioners.

The important reason behind the "inexplicable" is that security has become the fastest-growing field in artificial intelligence. Whether it is start-ups such as SenseTime, Megvii, and Yitu, or technology giants such as BAT and Huawei, all are involved in the smart security business.

As a leader in the security industry, Hikvision has actually been involved in the field of artificial intelligence for more than ten years.

Regarding the increasingly competitive security sector, Hu Yangzhong said, "Everyone thinks the security product market is infinite, but it is not. When people find that this market does not meet their original imagination, some players will naturally withdraw."

Security, what is virtual and what is real? This is the main topic discussed by Hu Yangzhong during his interview with Leifeng.com.

Leifeng.com: What was your original intention when you founded Hikvision?

Hu Yangzhong : It's nothing fancy, it's just a way to make a living. Unlike some startups nowadays, they just make a big promise without thinking about anything else. Now we are still thinking about how to survive and keep having a meal.

Leifeng.com: What is the biggest difference between running a business now and when you founded Hikvision?

Hu Yangzhong: I used to ride a bicycle, but now I drive a truck, and the steering wheel has become heavier.

Leifeng.com: In the past 17 years, Hikvision has been able to accurately grasp many technical nodes in the development of the industry. In your opinion, which important nodes are worth mentioning?

Hu Yangzhong: There are no clear nodes. The so-called nodes are concepts deliberately created by others to simplify an industry. They are one-sided conclusions drawn by outsiders after seeing the surface and partial results of things. This is an illusion.

Many people always compare the speed of a company's entry into a certain field with other companies, but do we really know the progress of the other party? Some companies may be working hard but not telling the outside world. Some companies do nothing but shout slogans loudly. The so-called "early" and "late" are just the volume of people's voices, not necessarily true.

Leifeng.com: What do you think is real? What is an illusion?

Hu Yangzhong: What is real to me may be fake to others; what is real to others may be fake to me. We often think something is real, but it is not. The only basis for testing whether it is real is to see whether customers are willing to continue to buy your products or services.

The so-called grasp of the nodes can only be verified by the results afterwards. There are many accidents and coincidences. Some nodes are unexpected, and you won’t think about them before doing them, but they happen naturally.

Leifeng.com: Based on your experience, how do you ensure that products and technological innovations are always what the market needs?

Hu Yangzhong: No, it is impossible. Otherwise, the century-old store would not be so precious.

Leifeng.com: Why do you say that? Have you ever made any products that are not needed by the market?

Hu Yangzhong: We have made mistakes. After SARS, we thought interactive video communication (videophone) might be a better business direction. A year later, we found that there were some problems with the direction, so we adjusted the business and gave up the videophone business direction.

Leifeng.com: What words would you like to use to describe the security market over the past twenty years?

Hu Yangzhong: Fast and small.

Leifeng.com: Fast and small?

Hu Yangzhong: Yes, the security market is growing very fast, but it is a “small” market.

On the one hand, the market capacity is not that big, otherwise, companies with sales of hundreds of billions would have already appeared. The real estate market is huge, and the communications market is also huge, so some big companies can be nurtured in these industries. The mobile phone market of more than 600 billion US dollars is really huge, which is why companies like Apple and Samsung can appear. Everyone thinks that the traditional security business market is huge, so they increase investment. After ups and downs, they finally found that there is a big gap between imagination and reality, and some chose to withdraw.

The "smallness" of the security market is reflected in the fact that it is a very fragmented market with unconcentrated users, scattered applications, scattered products, and a lot of customer customization. It is a very complex system to establish a development response that meets the fragmented needs of customers and the construction of rapid delivery capabilities.

With the advent of the intelligent era, previously unmet customer demands in the security industry have become possible to be met, and the market space has been further opened up. However, this is still a fragmented market.

Leifeng.com: Huawei is now once again launching a massive attack on the security market. What is this phenomenon?

Hu Yangzhong: Every company is looking for opportunities and markets; every business department has the urge to expand itself. If it does not expand, it will most likely be cut off within the company. So it will not do much harm to increase the efforts again. With a trial and error mentality, you can take another gamble.

About 20 years ago, Tyco, Honeywell, GE, BOSCH, UTC, Simens, and Schneider all entered the security industry through mergers and acquisitions. Panasonic, Sony, and Samsung also focused on monitoring products, but now they are all very unsatisfactory. The IHS report has relatively comprehensive data. Tyco merged into Johnson Control; GE security merged into UTC; Simens withdrew from the security industry; Pelco, which was acquired by Schneider, gradually declined; Samsung surveillance was sold to Hanwha, and Hanwha transferred its equity again; Panasonic and Sony surveillance market shares have been declining year by year; Honeywell and Bosch have hardly grown.

The same is true for some AI companies that have entered the security industry. They have been engaged in security business for three or four years, but the number of implemented projects is actually relatively limited.

Leifeng.com: How do you evaluate these AI companies with strong algorithm capabilities?

Hu Yangzhong: Different industries are like different mountains. For AI companies, the first thing to do is to solve the problem of taking root. Where should the roots be planted? How deep should they be planted? It is still hard to say.

Leifeng.com: So how do you understand this industry?

Hu Yangzhong: As I said earlier, this industry is very fragmented and the demand is also very scattered. The security business is like countless sandbags hanging from the ceiling. With one or two hard fists, you can only shake a few sandbags, but you can't take over everything and form a scale. And these one or two fists may be what we now call algorithms, talents or capital. Security is not something that can be done by just investing money. Capital may have a temporary impulse, but it will also exit quickly.

Leifeng.com: Do all those technology giants with huge capital have no chance?

Hu Yangzhong: I think the chances are slim. Deep waters breed big fish, and how can the security industry breed big fish without deep waters?

This is a fragmented market. Essentially, we still have to rely on sales in countless small areas, one order at a time, and on door-to-door sales to acquire customers. The channels must be very detailed, and sales cost control is an extremely important part of the company's operations.

Taking companies in the communications industry as an example, the operating costs are much higher than security. Using the high-cost manpower in the communications industry to run security is like using a rifle to shoot flies. The input and output are very mismatched.

Leifeng.com: Has this industry undergone any changes compared to when you entered it more than a decade ago?

Hu Yangzhong: It’s still the same as before, very fragmented, with many product types and models and very scattered orders.

Leifeng.com: Has Hikvision’s core competitiveness changed in this process?

Hu Yangzhong: There has been no change.

Leifeng.com: What is that?

Hu Yangzhong: It’s still about serving the fragmented market.

Leifeng.com: Is it possible that AI will change the current landscape of the security market?

Hu Yangzhong : It’s very difficult.

Leifeng.com: Do you mean it is difficult to change Hikvision’s current position?

Hu Yangzhong: It is not just Hikvision, but other companies as well. It is not easy for companies to survive in this industry. They still have some tricks up their sleeves. In addition to AI, there are many other important technologies in this industry. AI is not a panacea. As an old security company, if we lack something, we can quickly make up for it. Moreover, traditional companies are not weak in AI at all. Hikvision itself started AI very early and has done very well in AI.

Leifeng.com: Do you agree with the term AI security?

Hu Yangzhong: The application of AI in security can indeed bring some changes, but AI is only one of many technologies in security. It does not mean that video surveillance no longer needs encoding, low illumination, and wide dynamic range. With the advent of AI, cameras still need fill light, storage, and resistance to harsh environments, and the technologies involved are very complex.

Leifeng.com: When planning AI, which factors do you think are more important?

Hu Yangzhong: We need to solve the actual needs of users in fragmented scenarios. Without solving user problems, everything is just empty talk.

Leifeng.com: What do you think about the recent popularity of “AI chips”?

Hu Yangzhong: People like hot spots. If there are no hot spots, we will create them. When it comes to AI chips, what most people see are illusions.

The development of an industry also has its own ecology and rules. You can't just throw money at something to get it done. It requires capital, time, talent accumulation, and team culture. Cooking over high heat will only make half-cooked rice, and only low heat can make good soup.

Leifeng.com: Will Hikvision make its own AI chips?

Hu Yangzhong: I won’t do it now.

Leifeng.com: Will it not be launched in the future?

Hu Yangzhong : At present, we will still choose the cooperative approach.

There are also some misunderstandings about chips. Having chips does not mean you have everything. Growing cotton does not necessarily mean you can weave good cloth, and weaving cloth does not necessarily mean you can make good clothes. We still have to respect the laws of the market.

Leifeng.com: What strategy does Hikvision pursue in its cooperation with upstream and downstream companies?

Hu Yangzhong: Look at customer value. If others make some products better than Hikvision, let them do it. There are many things that Hikvision does not make, such as chips, lenses, structural parts, and system integration.

We are not following the example of Internet companies. Regardless of whether they are good at some businesses or not, they will try to take over all of them by leveraging their huge user and traffic advantages.

Leifeng.com: Many manufacturers are now proposing software-defined cameras. What do you think? Which problems can be solved by software and which cannot?

Hu Yangzhong: The combination of cameras is determined by the scene (including sensor performance, resolution, focal length, fill light strength, etc.), which is very complex, so there are so many types. The computing power, configurable functions, or other functions that can be realized by software in integrated circuits or chips can be scheduled by software. However, when it comes to hardware such as optics and structure, the role of software is very limited. Software-defined computing, software-defined storage, software-defined networking, and software-defined cameras are just eye-catching.

Leifeng.com: Compared with top Internet companies, do you think Hikvision’s current operating model is benign or better?

Hu Yangzhong: Enterprises must abide by the bottom line of business and cannot plunder the market in a barbaric way or engage in vicious competition by burning money on subsidies. Early users seem to have gained some benefits, but in the end they have to pay more because they are kidnapped.

Leifeng.com: I heard that Internet communication giants have poached many R&D personnel from Hikvision?

Hu Yangzhong: Not a lot, but there are some. For companies with tens of thousands of R&D personnel, a certain amount of turnover is normal. Moreover, some Internet communication giants often do not really need these R&D personnel. They are willful with too much money, which harms others and does not benefit themselves. This is not only the sadness of these R&D personnel, but also the sadness of this impetuous society.

Leifeng.com: What is the biggest challenge you have encountered in the process of running a business?

Hu Yangzhong : The biggest challenge always comes from within the company.

Just like people, whether a company is healthy or not depends on how it manages itself. We should not only manage our own behavior (such as going to bed early and getting up early, not overeating), but also manage our hearts. The same is true for companies. In addition to daily operations and management, they should also pay attention to culture, whether they can stick to the corporate culture, and whether they pursue short-term or long-term goals.

Leifeng.com: Apart from the internal situation, what are you most worried about externally?

Hu Yangzhong : It is still an internal problem. Paying attention to the outside is also to see the internal problems. We can better see our own problems from the customers.

Leifeng.com: Last year, Hikvision launched AI Cloud. What is it? What problems does it solve?

Hu Yangzhong: AI Cloud aims to solve the problem of IoT implementation. IoT has been discussed for a long time, but it has not been well implemented yet. Talking only about cloud computing is not comprehensive, and talking only about edge computing is not comprehensive either. In the end, cloud-edge integration is still necessary, and cloud-edge integration can better solve the problem.

Leifeng.com: Why launch AI Cloud at this time?

Hu Yangzhong: Our own cognition has reached this stage. Many things just happen naturally.

Leifeng.com: AI Cloud has been launched for a year. How would you rate it out of 100?

Hu Yangzhong : I never give scores.

Leifeng.com: Why?

Hu Yangzhong: This is not a matter of self-rating. In the end, it depends on the users to evaluate.

Leifeng.com: How is AI Cloud doing now? What feedback do partners have on AI Cloud?

Hu Yangzhong: This is not something that can be accomplished in one or two years. The technology itself is still developing. Some pioneer users are already using it, and I believe that more and more people will agree with it in the future.

Leifeng.com: For AI startups, what do you think they need to learn from entrepreneurs of your generation?

Hu Yangzhong: I dare not. Our generation is already behind the times.

Leifeng.com: The market is very turbulent, but it seems that you are very calm inside. What do you want to say to your security colleagues?

Hu Yangzhong : Everyone has their own ideas and there is no need to listen to others. I only say to Hikvision: We must always be vigilant and not do anything speculative.