Editor's note: In June this year, LiveVideoStack talked with Li Yutao, Vice President of Tencent Cloud, and led Tencent Cloud Audio and Video to "ALL IN overseas". After four months, LiveVideoStack once again spoke with Li Yutao, discussing his experience of the cultural differences between China and foreign countries after going overseas, opportunities in the SaaS software ecosystem, opportunities in the digitization of traditional industries, and the current status and future trends of the domestic streaming media industry.
Planning/LiveVideoStack
-01-
Review 2016-2023
LiveVideoStack : I remember we met for the first time at LiveVideoStackCon in Beijing in 2018, and we also did an interview during the period. It has been almost 6 years. Both you and the industry have changed. Can you briefly summarize the past 6 years?
Li Yutao : The past six years have been six years of blindfolded running. It can be divided into three stages: before the epidemic, during the epidemic, and after the epidemic.
Before the epidemic, (the industry) was still looking for growth opportunities based on the growth of Internet and mobile Internet users, as well as the trend of video and live broadcast. What impressed me the most is that live broadcasts experienced a very large growth in 2018, mainly driven by "chicken eating" live broadcasts. Because the "Chicken Eating" live broadcast has high requirements for viewing, the requirements for picture quality and resolution are very high. Basically, our monthly traffic has increased a lot month-on-month, and the annual growth has doubled. Everyone will think, wow, the video field will continue to maintain rapid growth in the future. Tencent Cloud Audio and Video also made real efforts to make the transcoding service intelligent this year. We added pre- and post-processing capabilities to the transcoding service, reducing user costs while ensuring users’ high-definition demand for image quality. This is (Tencent Cloud) The origin of ultra-fast high-definition transcoding service.
From 2018 to 2019, (the industry) began to compete in deep water areas. Because the live broadcast market is relatively mature, but more and more industries are using audio and video technology to solve communication and interaction problems, (in this context) it was decided to reconstruct the RTC service - TRTC was re-established.
However, in 2020, I was knocked unconscious by the "bang" (epidemic), and I was completely unprepared. I still remember that people started to rush to buy masks a few days before the Spring Festival. Some classmates who were doctors said that the epidemic would be serious. About 2-3 days later, I received a notice from the company that Tencent Conference needed support.
At first I thought it was some minor system problem and I could solve it with some help, so I went there alone, but as soon as I passed, I was left by Lori (LiveVideoStack note: Lori is Wu Zurong, vice president of Tencent Cloud and head of Tencent Conference) ), "You can't leave anymore, the number of users is growing too fast." It couldn't be stopped at all. It was growing several times or even ten times every day. Only then did I realize the seriousness of the situation. Later, I recruited many students from Video Cloud to do system guarantees, but found that the guarantees were simply not possible. Because users are growing too fast, some system architectures are very fragile, because the system design was not originally designed for such a large number of users, including capacity expansion, peak resistance, and network scheduling. Design for a large number of users was not considered. In the end, it took us about two weeks to a month to stabilize the entire system. During that time, we basically went back and forth between crashing, going online, expanding, crashing, and going online every day.
During this period, the backend architecture of Tencent Conference and TRTC has been greatly improved and a lot of capabilities have been accumulated in a short period of time. At the same time, the live broadcast business has been put on the high-priority agenda. This is mainly because the entire industry cannot conduct offline business. The physical industry, especially those related to retail and e-commerce, immediately consider how to move sales online, reach customer groups through live broadcasts, and then use logistics Implement delivery. Around April to May of the same year, the WeChat video account decided to do live broadcast. It happened that Tencent Cloud Audio and Video already had a relatively mature live broadcast solution. We provided rapid support for the video account for about two months (WeChat Video account live broadcast) is now online. If we return to the level of 2018, we won’t be able to fix it in less than half a year.
During that time, we also gained many customers that we could not support originally. These customers also needed a strong partner and supplier to solve problems such as capacity expansion and sudden traffic in the context of skyrocketing demand. Summary In 2020, the demands that have been extremely compressed for many years all exploded at the same time this year, forcing us to accumulate many capabilities through overtime work.
2021 is a relatively stable year, allowing us to take a breath and decide whether to look at overseas opportunities. Overseas, things are still in dire straits, and demand is skyrocketing. The information I came across (at that time) was how much will Zoom increase, and how much will Salesforce increase tomorrow? The day after tomorrow, Microsoft launched Teams... Many product lines of Tencent Cloud Audio and Video will begin to be internationalized in 2021. Previously, they were mostly driven by domestic customers going overseas... For example, Douyu is here The country has needs, but the traffic flow is stuck, and we need to solve it; Huya is going to launch a new application Nimo, and we have to support it...
In 2021, TRTC began to internationalize. The most difficult thing was to find customers in North America. At that time, we found an overseas customer. At that time, they wanted to do a scene similar to Clubhouse. Since Musk made Clubhouse popular, many people thought that live music podcasts were the future trend. They could use a very simple way to online Real-time communication online will replace offline meetings. After a lot of hard work, I worked overtime throughout the Spring Festival to translate documents, modify the API to adapt to the habits of American developers, did a lot of online discussions and clarifications, and finally passed the technical solution. But in the end, they were still worried about non-technical issues and were afraid of affecting the development of their own business, so we failed to become their supplier. Since then, TRTC and Chat (international version of IM products) have completely begun to internationalize, because in addition to the US market, there should be many customers in Southeast Asia, the Middle East, and North Africa who need Chinese technology to solve their online problems caused by the epidemic. The problem.
Everyone’s experience in 2022 will be quite ups and downs, and everyone is mainly focused on “reducing costs and increasing efficiency.” The focus of our products has become to help customers reduce costs or tailor products that are not so important. At the same time, we have also made some adjustments to the accumulation in 2020 and 2021 to make users' use smoother and further enhance their experience. Overseas business will continue to accumulate in 2022. Although we cannot travel overseas, the local overseas sales team will help pass on local business opportunities to the country, and everyone will actively think about and support the service docking of these customers. By 2023, our product team can’t wait to go overseas, to Singapore, Japan, South Korea, Europe and the United States, etc., to communicate face-to-face with regional sales teams and customers, to experience the needs of overseas customers first-hand, so as to continuously optimize our products positioning and direction.
It feels like 2023 has passed very quickly, and it has entered the end of October in a flash. Looking back now, 2023 is more like the first year of internationalization. Not only has there been substantial progress in product maturity, but also in market and customer contact. If we look forward to the next 2-3 years, or even 3-5 years from the current point in time, 2023 must be a good starting point.
LiveVideoStack : When did you start leading Tencent Cloud’s video cloud business? I remember that you were responsible for communications and audio and video technology at QQ before.
Li Yutao : I joined the Tencent Cloud audio and video team at the end of 2015. Not many people participated in it at the beginning, there were some voice services. At that time, Video Cloud and QQ Voice were both working on it. The video cloud business really started in 2016. After 2018, we will gradually start to develop products such as live broadcast, on-demand, TRTC, IM, and edge acceleration.
LiveVideoStack : Live broadcasting developed the fastest in 2016.
Li Yutao : Yes, the fastest development time. We haven’t reached the top yet in 2016, but we will really reach the top in 2018. The growth of live streaming has leveled off after 2018. There were several driving factors for the development of live broadcasting before 2018: 2016 was the Qianbo war, and all apps were doing live broadcasting. Several unicorns emerged in 2017, including Huya, Douyu, Bilibili, and Kuaishou. Kuaishou also solves commercialization problems through live broadcasting. In 2018, Douyin and several e-commerce platforms conducted live broadcasts. The growth of live broadcasting has slowed down in 2019, and will be stimulated by the epidemic in 2020. Then it slowed down again.
-02-
Facing huge cultural differences when going overseas
LiveVideoStack : There are definitely more opportunities overseas. Shengwang’s overseas business accounts for more than 40%. After the domestic education market was "killed", many (suppliers) are reducing costs. As we talked about last time, in terms of basic usage habits, such as documentation, API, etc., the habits of domestic and foreign developers are different. Is the problem faced by video cloud overseas just a matter of documentation and API? Or is it a systemic problem?
Li Yutao : When we first get in touch, the most direct feedback from customers is that they can’t understand the documentation or they’re not used to the API. However, after in-depth communication with customers, I found that there are deep-seated (cultural) reasons behind this. For example, in the West (culture), you have to do many things by yourself. You want to do it by yourself from beginning to end, and you don't want to disturb others. I have had a subconscious feeling since I was a child that if I ask for help from others, I may have to pay for it and may disturb others. Therefore, the habit of doing things according to documents and agreements will be formed. There are many assumptions behind this subconscious mind and habit: **If you are a commercial company, then your product should have passed various tests and verifications, and the quality is guaranteed. **Only in this way can you dare to market. Unlike the development logic of China’s Internet over the past decade or so, we are running fast in small steps, agile, trial and error, and rely on users’ negative feedback to improve. Chinese or Eastern people (subconsciously) believe that people should help each other, and they should impress customers with service. I can do it for you. You don’t have to do everything yourself, I will help you get everything done. Therefore, **many of our products were not particularly perfect when they were introduced to the market. We are waiting for one day when customers encounter relevant problems or needs, and we will solve the problems through services. **We think this (way) will be very efficient and customers will be very satisfied. But little did they know that the differences between the two cultures turned into a huge obstacle. If a Westerner comes across some of your imperfect products, they will think that the product is not usable at all, and will not cooperate with this company in the future, and will not come to you at all (feedback issue). This cultural barrier causes us to miss many business opportunities.
After we recognize the (cultural) problem, we must first make changes in consciousness, improve the quality of the product when it is launched, and add testing and SLA strategies so that customers know what stage your product is at today, what is OK and what is not. , tell customers clearly. Western culture especially hopes that you will be transparent, but Chinese people are often reserved and half-spoken, and often rely on subsequent feelings. If you see the cooperation between Chinese people, they will adapt to this situation very well, and the cooperation will get better and better. Cooperation between Westerners is in the same state from beginning to end. However, it was not easy for Westerners and Chinese to reach agreement in many aspects from the beginning of their cooperation. One side had very high expectations and could get it done in one step, while the other side believed that I could slowly improve and iterate later. Therefore, we are now also adding many students with overseas study and work experience to communicate with overseas customers to make up for the cognitive bias.
By 2023, as more and more overseas customers use our products, many products will continue to be improved and more suitable for overseas customers. We used to like to use whitelists to distinguish short-term and long-term (functional) differences. The so-called whitelist means that when some functions of my product are not perfect, the product cannot be publicly released to all customers at this time. Open it to one or two important customers first and open it in a whitelist manner. Over time, many products will have many whitelists, and many customers will use these product capabilities in non-standard ways. There are also many compromises when delivering products. If a customer can do something, the supplier can also do it. Many times the supplier does it for the customer first, in a non-standard way.
When we observe overseas products, we find that their roadmaps and product definitions are extremely clear, and they rarely do non-standard things. They believe that non-standard products should not be included in the product scope, and will let customers do it themselves through ecological partners or open interfaces . This is also worth learning. In the later product design process, extreme consideration was given to product standardization, seeking openness and collaboration with partners. Of course, I said this before 2018, and the truth is the same, but I didn’t really do it until I actually met overseas customers.
LiveVideoStack : In the last interview (in June), you also talked about going abroad. This year, you talked about it in more detail, which means you have more new experiences.
Li Yutao : Last time I went to sea with my soul, but this time I went to sea with my body (laughs).
LiveVideoStack: (Overseas areas) Mainly in Southeast Asia, Japan and South Korea?
Li Yutao : They are available all over the world. In the past three months, I have also been to Europe and the United States.
LiveVideoStack : Have you been to NAB (American Radio, Film and Television Equipment Show)?
Li Yutao : I didn’t go this year, but I will go next year.
LiveVideoStack : Went to IBC (Broadcast and Television Communications Equipment Exhibition)?
Li Yutao : I went to IBC.
-03-
Video conferencing has the potential to change China’s SaaS dilemma
LiveVideoStack : The main theme of the domestic market is cost reduction before the arrival of new opportunities and cycles.
Li Yutao : Yes. China's Internet has developed too fast in the past decade or so, and the video field has also developed very fast in the past 5-6 years, so the division of labor and resource investment in the industry are not efficient. Because many companies have been engaged in cost reduction and efficiency improvement in the past two years, the division of labor in the industry and the rationality of business will return. To take the simplest example, in 2020, almost all companies are recruiting audio and video people and forming audio and video teams. Companies that already have audio and video teams are poaching stronger people. However, in 2021, education will suddenly (double decrease), social entertainment will become legal, and everyone will realize the risks of building their own technical capabilities. Especially in 2023, it will be more obvious. I have observed that many students in the (multimedia) industry have left their original companies. The reason is that it is becoming less cost-effective to build a self-built team than to purchase cloud services, because the price of cloud services has been going down in the past two years, and the cash flow pressure brought by raising a team is increasing. At the same time, the business There is no longer any expectation of rapid growth, so industry divisions tend to be more reasonable. For example, if a few companies in the PaaS layer survive, they will continue to operate for a long time in the future, and barriers will be established. Unless there are new breakthroughs in technology, it will be difficult for new players to catch up. The situation with IaaS is similar. New players will face the break-even threshold and will find it difficult to survive. As for upper-level SaaS, China's SaaS is far from mature or has sufficient commercial expectations. Fortunately, Tencent is an explorer in the (SaaS) track. I believe that as these companies gradually commercialize and their understanding of the market gradually converges, China's SaaS charging will be a matter of time, and charging in the field of video conferencing may lead China's SaaS out of various customizations and various money losses. of haze.
Overseas SaaS is developing very well. All users take it for granted to buy a software or service. It is natural for them to spend dozens or hundreds of dollars to solve their own efficiency problems . But in China, everyone is still far from settling this account. Of course, this is also related to China's economic situation, because our labor costs are relatively low, and everyone would rather use more labor than purchase services. I believe that the day (SaaS commercialization) will definitely come, and economic development will reach this stage sooner or later.
To sum up, the domestic market has passed the epidemic, companies have reduced costs and increased efficiency, business will return to business, and the division of labor will be more reasonable. When competition is less intense, everyone has more energy to think about how to do difficult but right things and how to give better value to upstream and downstream companies . There can be no winner-take-all scenario. I feel quite confident about the domestic market. Every time you talk to me, I am quite optimistic.
LiveVideoStack : Speaking of Tencent’s investment in SaaS, what else is there besides Tencent Conference?
Li Yutao : Let me say that I am best at the fields of audio, video and communications. There are only two tracks with the greatest demand in the field of audio and video communications. One is corporate collaborative office, and the other is corporate marketing and customer service, such as call centers. We are incubating call center products and doing it in the form of SaaS public cloud. You may also notice that Genesys announced that it will shrink its product line in China and may gradually withdraw from privatized products, which will leave great opportunities. Returning to this track, it essentially solves the problem of enterprises communicating with customers through multimedia methods such as voice and video. Companies like Genesys and Avaya help companies build call centers through hardware, phones, and dedicated lines. Twilio has long been available overseas, and even AWS already has a call center product (Amazon Connect). By cloudifying all hardware and achieving elasticity, because calls have peaks and troughs, and services also have peaks and troughs, traditional hardware call centers are a waste, and cloudification solves the problem of resource waste.
With the development of AIGC this year, traditional call centers and customer service are inefficient and will be replaced by AI in the future. Not only efficiency will be improved, but management efficiency will also be improved. I believe the same is true in overseas markets. Genesys’ adjustment this time is also strongly related to the evolution of AI technology. It hopes to go all-in to the public cloud and embrace the next wave of opportunities by combining it with AI.
We started developing related products two years ago. Although the investment was not large, cloud call centers are being accepted by more and more traditional industries. This situation would have been unimaginable before the epidemic. At that time, it was generally believed that IDCs and call centers should be privatized, and public clouds were unsafe and not completely controllable. But now efficiency is given priority and cost reduction and efficiency increase are given priority. Agents can be opened with very little investment, which is very flexible and convenient. Because cloud call centers are also a very good SaaS development direction in China. Offline retail, supermarkets, convenience stores, automobile 4S stores, finance and insurance will all have such needs. We are incubating such a product called TCCC.
LiveVideoStack: (TCCC) Mainly for domestic use?
Li Yutao : Yes, it is mainly for the domestic market now, but an international version will be launched soon.
-04-
There is no end to the audio and video industry
LiveVideoStack : Is there a lot of pressure in China? Indicators, KPIs, revenue, and profit margins from the boss.
Li Yutao : It is definitely impossible without pressure. The KPI has been set there and the demand is shrinking. In fact, the management also understands that the industry is changing and will encourage us to invest in things with long-term development potential while reducing costs and increasing efficiency. (Supervisors and subordinates) After aligning the direction of development and the way of development, the pressure will be relieved. (KPI) is not a purely numerical pressure. It is okay in Tencent's system at present. After all, Tencent still has a lot of healthy, revenue-generating and profitable businesses that are growing. In the To B business, the company gave us a lot of time and tolerance, and forced us to do well in various abilities and what we should do.
LiveVideoStack : Basically everyone is like this. Frankly speaking, there is definitely pressure.
Li Yutao: (The pressure) is definitely stronger than in 2020.
LiveVideoStack: (Domestic) is a stock market. Only by snatching yours can it become mine.
Li Yutao : Yes. Everyone has a saying: Survival is the last word. Whoever makes no mistakes will live longer. When others fall, there is space for you. Many large companies grew up during the economic downturn.
LiveVideoStack: (Economic downturn) is definitely an opportunity for giants, but it is definitely difficult for small and medium-sized companies.
Li Yutao : Small and medium-sized companies are under greater pressure. On the other hand, if they operate well, it will be a good time window for winter. Because they are not prone to rapid expansion when the market is overheated, causing financial risks. Every penny spent now will be very careful and consume less. If you accumulate it correctly, you will be able to live longer than anyone else in the end, and use the accumulation during this period to gain the ability to develop next time.
LiveVideoStack : Next question, how do you see the next wave of multimedia opportunities? Whether it is economic cycles, technological breakthroughs or industry iterations.
Li Yutao : Let me talk about it on several levels. In China, the stock market that everyone refers to refers more to the Internet, but traditional industries are still in a very early stage, and there is still a lot of room. The real-time remote control solutions we use in industry, energy, and mining have already It has been truly put into use in business scenarios. Many vehicles, cranes, and excavators are in use, which has greatly improved the customer's driver safety and business efficiency, which gave me great confidence. I'm thinking that after this epidemic, many traditional industries will encounter bottlenecks when their digital transformation reaches a certain stage. Every company talks about digitization, but what value does it bring to the business? Fixed assets may have been reduced (making finances healthier), and the compliance of cloud services has been verified, but few companies have transformed their business processes and improved efficiency by adopting new technologies. We are using audio and video technology on the cloud to transform enterprise processes, take the remote control solution as an example. Originally, we were worried about safety accidents due to audio and video interruptions, jitter, and large delays. Of course, we have taken a lot of safeguards. In the end, what makes us feel more reassured is that these companies will also do a lot of deductions, guarantees, and technical verifications when they are transforming. , just a little bit of application and online.
In the future, as long as the infrastructure of a certain industry is cloudified and network conditions permit, the production process will always need to be changed to achieve unmanned operation in a more efficient way to solve the problems of safe production and digital efficiency. Audio and video are a good carrier, whether it is communication between people, communication between people and things, things between things, or even digital collection. From data reporting on weak networks to rich media, these will improve traditional industries. Production format. This is the key reason why I am optimistic about China's economy. Because the Chinese people are very diligent and the industry is so complete. It may be the most complete industrial system in the world. The value that digitalization combined with big data and AI will bring is worth looking forward to.
Next, let’s talk about the technological development in the audio and video field. Still can’t see the end (of technology). Every time you think that the compression rate reaches the bottleneck, the standard for the next five or ten years will burst out. For example, we won 21 first-place results in the MSU competition, but also in H.266, Alibaba also got very good results when participating in the MSU competition, which shows that each company has done a lot of groundwork before the technology is launched. . Tencent's H.266 is also online, and Tencent Video is already using it, and the compression efficiency is good. This gives us at least 5-10 years of imagination. As long as the performance bottleneck is broken through and the cost-effectiveness of computing power matches, the compression rate will increase by about 30%, and the video will become better-looking and the transmission will be more friendly. In addition to Codec, when we started working on high-speed high-definition in 2018, we did video pre-processing, enhancement, and bit rate control together with Codec, which resulted in further improvements in (subjective) image quality and adaptation to various scenarios. There are many benefits in various industries. Take last year's Football World Cup as an example. In a short period of time, the picture quality of football live broadcasts has been improved to a state that has not been achieved in the past. It is also used in many live events overseas, such as cricket matches in Pakistan and European events. We have them all. We support our customers in using the same method to adjust the image quality, which cannot be done by the general Codec. Next is AIGC, this year’s hot spot. With the infrastructure of large models, there are new solutions to problems that CNN and RNN could not solve in the past. The original bottleneck of content production may be broken through, and the gameplay and applications combined from this will be Become more. Finally, cloud gaming and cloud rendering are now likely to explode again. The price-performance ratio is about to break through the bottleneck, and the single-channel cost is almost equal to the unit price per customer. Once the bottleneck is broken, applications such as cloud game trials, subscription-based cloud games, promoted downloads, and real-time rendering such as digital twins can bring another wave of development. Coupled with the launch of Vison Pro next year and the maturity of the surrounding ecosystem, there will be more and more real-time space collection equipment. Many new products will surely emerge, and the corresponding content production ecosystem will gradually develop. 3D immersion is another wave.
Adding up all the above, the development of audio and video has no end in sight.
LiveVideoStack : The conclusion is relatively optimistic.
Li Yutao : I am optimistic every time (laughs).
