No one can become the YouTube of China.

No one can become the YouTube of China.
- youtube
  • This article is authorized to be reproduced from the official account: red Tide AKASHIO (ID:AKASHIO)
    Author: commenting on the corpse recently, Watermelon video, bilibili and iqiyi are debating "who is China's YouTube"
    YouTube can be said to be the holy grail of China long Video Company, including Tudou, Youku, six rooms, 56
    com, Sohu Video, Tencent Video, iqiyi and now bilibili and Watermelon Video
    All the video websites that have existed in China, except Letv, have wanted to build themselves into the "YouTube of China" at some stage
    But they have either failed or are on their way to failure
    What is YouTube? In the eyes of users, YouTube is an open video community, a great philanthropist who can skip ads, a top product with first-class functional design, and a long video recommendation stream as accurate as Douyin
    YouTube is the promised land for all video users
    If the experience of all long video sites is not good, it must be because it is "unlike YouTube"
    But YouTube is not a charity
    What is the nature of YouTube as a business? Is it accurate advertising? Is it an ultra-high percentage of creators' sharing plan? Is it a great homemade channel? Neither
    YouTube's business logic is simple: a video site that pays almost no Internet fee
    The server and bandwidth cost of the video industry if you have a basic understanding of the server cost of the video industry, you can skip to the second section
    Why did Chinese video websites fail to do UGC or PUGC before? Because the server and bandwidth costs are unbearable
    Why was Douyin only allowed 15 seconds of short videos when it first launched in 2016? Because the server and bandwidth costs are unbearable
    Why did Wechat's moments rise from allowing 10-second short videos to 15-second short videos at the end of 2019? Because the server and bandwidth costs are unbearable
    In an article [1] of the Daily Economic News in 2012, Chinese video websites at that time were divided into four categories: the first category is Youku and Tudou, which uses UGC (user shared content) to lengthen the legitimate video content; the second is the UGC model represented by 56net and Ku6, which focuses on user sharing content and interaction; the third is authentic HD videos represented by iqiyi, Tencent Video and Sohu
    The fourth category is the video information category represented by the exciting network and the first video
    In retrospect, you will find that the third category of video sites based on legitimate content has become the current mainstream
    In the first category, after the merger of Youku Tudou, which both UGC and legitimate dramas have to do, it has actually become the third category
    The second category of video sites, which are mainly UGC, PGC and community, have died out

    Oh, no, it was a dead round, and now bilibili is the second round
    Specifically, let's take a look at Tudou's data at that time: according to Tudou's 2011 annual report, Tudou's server bandwidth cost that year was 180
    2 million yuan, accounting for 42
    1% of the revenue cost
    In other words, nearly half of Tudou's costs are paid to operators
    According to a report released by iResearch that year, Tudou had 193 million monthly active users that year
    [2] here, we divide the two and invent a new concept to compare the inflation of paid server costs: "return on infrastructure investment (too long, hereinafter referred to as IROI)
    " Tudou's IROI=1
    07 person / yuan, that is, 1 yuan per year on server bandwidth, can serve 1
    07 more monthly active users
    Some people will say that the overall cost of server bandwidth has been falling over the past decade, and the operating costs of today's video sites cannot be so high
    This is actually caught in the trap of Moore's Law-computers at the same price today perform more than ten times better than those of 10 years ago, but can you still install Windows XP on today's computers and continue to use the software and games of that era? In the past 10 years, users' requirements for picture quality have changed from 480p to 4K 60 frames with a high bit rate
    And the significant upgrade of shooting equipment will make the current pure UGC platform more cost-effective
    If you don't believe me, take a look at bilibili's server bandwidth cost
    Bilibili's bandwidth server fee for the whole of 2019 was 919 million, accounting for 16
    5% of the cost
    Bilibili had 130 million monthly active users at the end of 2019, so it can be calculated that bilibili's IROI=1
    3/ 919 million = 0
    14person / yuan bilibili invests 1 yuan more per year and can only serve 0
    14 more people
    In other words, in terms of service capacity, instead of falling, the cost of server bandwidth has increased sevenfold in the past 10 years
    (about five times excluding inflation) with such high service costs, only by frantically squeezing commercial value can we become a normal business rather than philanthropy
    Tudou failed to survive such a high cost on its own and was finally eaten by Youku
    So what can bilibili rely on to smooth out this seven-fold cost increase and make a profit? Of course, it depends on "changing the taste"
    If you can't cut expenditure, you can only open source
    Bilibili, as a highly sticky 172 million monthly active Internet product, is still "pure" so far, and there are still many places to add advertising
    Whether you continue to channel games, pay for knowledge or be an e-commerce in the Z era, you may even have aIt should be no problem to make a profit
    One spoonful of commercialization is not enough, seven spoonfuls, up to profit
    But when that day comes, will bilibili still be the "promised land" in the hearts of users? In the same way, the same is true of watermelon videos
    Why can YouTube be YouTube?? Let's first set up a scenario: Baidu has obtained a magic, so that Baidu network disk can run without a server and bandwidth, what will it do? Baidu's online disk will certainly remove all capacity, download and upload restrictions
    The price may be reduced to a symbolic 1 yuan or completely free, and then make money by indirect means such as scanning user data and placing accurate advertisements, when Baidu's online disk must be the best user experience on the planet
    When a business becomes unprofitable, it usually becomes less profitable
    Because every penny received is earned more, it is more important to squeeze competitors with a good experience than to earn an extra dollar
    And YouTube is the product that gets the magic
    To explain this, let's start with the domestic network environment
    If you surf the Internet early enough, you should remember that it is very slow for domestic Internet Unicom users and telecom users to access each other, transfer files, type voice or video
    Until now, some domestic online games are also divided into telecom service, Unicom service and mobile service
    The reason is that even though China's operators are state-owned enterprises, companies still account for each other independently
    They each build their own infrastructure (which used to be), which is equivalent to forming "several physical networks"
    As a Unicom user, the access fee you pay to Unicom is used to access the resources hanging on the Unicom network
    Whether you are chatting with a telecom user or downloading content from a website hanging in the telecom room, half of you are using telecom infrastructure
    So in theory, you have to pay for telecom
    However, this thing can not be done too well, or it will really become two nets
    Therefore, an agreement will be signed between Unicom and Telecom, which is rarely mentioned in China, so there is no name
    In foreign countries, it is called Peering Agreement, peer-to-peer agreement
    To put it simply, it is for the convenience of users of Unicom and telecom to access each other, although Unicom users do not pay to Unicom, and telecom users do not pay Unicom, but they can get a certain free data exchange quota for each other
    For example, it is agreed that every day all Unicom users can send 1TB content to the telecom network, and every day all telecom users can send 1TB content to Unicom network
    Within this amount, neither Unicom nor telecom will pay the other, and the two users will not have to pay for the broadband
    But the name "peer-to-peer protocol" sounds bad, because there are not so many equivalents for things on the Internet
    Let's take it a step further and now pull the video website in
    2005Youku was founded in #, it is headquartered in Beijing, the territory of Unicom
    Let's assume that its initial computer room is only connected to the bandwidth of China Unicom
    What will happen then? Youku is developing rapidly, and more and more users are watching videos
    Telecom will find that Unicom's daily quota for accessing Unicom's online resources is used up in an hour, and Unicom has to pay a settlement fee for the remaining 23 hours of running traffic
    Not only that, and all telecom users access Youku very slowly, the card is not good
    The only solution to this problem is to let Youku also connect directly to the telecom network, so that telecom users do not have to Peering to watch videos on Unicom
    However, this solution has developed two different directions of compromise at home and abroad: 1
    In China, the operator is absolutely strong, only it collects money
    If you want to develop users in South China, you must come to our telecom backbone network to pull a line
    In this mode, website operators need to pay bandwidth fees to operators to ensure user access speed
    two。 In foreign countries, Google is absolutely strong, you a carrier access to Google is so slow, waiting for users to change broadband brands
    Therefore, operators should use Peering to actively dock with Google's computer room
    In this direction, website operators can even charge operators for money to improve user access
    Google started its series of operations here: first, in 2005, Google was revealed to have bought a lot of dark fiber (Dark Fiber) [3]
    Dark optical fiber means that operators tend to lay more optical fiber when laying optical fiber for cities
    For example, they expect the building to use 1TB broadband in the future, but will lay 10TB underground
    Because the cost of optical fiber laying mainly comes from digging the ground, digging the wall and burying the wire, just like the wiring of your home decoration, it is only reasonable to open the wall once
    Therefore, for a long time, although there was only 1TB network access in a building, the 9TB bandwidth optical fiber was still idle in the wall
    Google rents dark fiber everywhere at a low price
    Where you rent a more complete piece, you will build an edge computer room there and spread out the edge computing and CDN of some of your own servers
    In this case, for example, you are a subscriber of a broadband operator in City A, and the broadband operator and Google Peering
    Your request for access to YouTube will be routed to the edge computer room of Google in City A, and then transmitted on the intranet of Google, which greatly saves the number of network routes and improves the speed of access
    And Google will not go to City A for this
    The operator pays for the bandwidth-though it should have
    In 2009, Google was already the largest holder of Dark Fiber in the world
    Some media began to question whether Google wanted to hijack the Internet on the grounds of public rights
    So Google simply launched its own broadband operator Google Fiber the following year
    The project has largely stalled in 2016
    However, some foreign media have always suspected that the purpose of Google Fiber is never to provide services to users, but to frighten operators [4]
    Now it seems that the purpose of scaring operators is two: 1
    At that time, operators in Europe and the United States, especially in the United States, were affected by the collapse of Bubble on the Internet in 2000, and were reluctant to invest too much in infrastructure, which hindered the development of many new services of Google
    The vernacular is that operators are reluctant to speed up and reduce fees for fear of bankruptcy
    Google Fiber was the first operator in the United States to produce "low-cost 1G home fiber", the equivalent of putting a catfish in a box of sardines
    two。 As an operator, it is better to do Peering, with the operator after all, "if your users want to access Google, quickly, you have to pay me, not me
    " This logic is unacceptable in the global ISP community
    But if it is not Google, but an operator Google Fiber, it is in line with the general principle of equality of Peering Agreement
    Both of these goals should be achieved that, Google Fiber does not cover many areas and does not have many users from the beginning to the end
    , Google Fiber came to a standstill in 2017 after the three major operators in the United States launched similar gigabit broadband
    After Google Fiber stagnated, Google Peering did not stop, but was incorporated into Google Cloud, and developed better and better
    You can see the specific development data on the peering
    com portal
    So, is YouTube part of Google Peering? Apparently
    According to a presentation on Google Peering by Ritz Campbell, product manager of YouTube, at the 2014 APNIC Conferences (Asia Pacific Internet Information Center Annual meeting), PPT shows
    There are three advantages for operators to join Google PeeringOne of the items listed separately is to improve the quality of YouTube access
    This practice has caused resistance in France, where broadband users almost all over the country found that access to YouTube became very slow in early 2013
    The reason is that all operators in France do not Peering, with Google and they think that if Google wants French users to access YouTube faster, Google will have to rent French broadband from them (Chinese operator model)
    [5] of course, France surrendered at last
    To further "intimidate" operators, Google has been rating ISP services since 2014, a tool that is actually "measuring how your network accesses YouTube
    " If your test results are not good, YouTube will give you detailed science popularization, which is mainly the operator's pot, and suggest you switch to a broadband that supports Google Peering
    YouTube, a product owned by Google (Alphabet), has rarely disclosed key data since it was acquired in 2006
    Until 2020, Alphabet reported the revenue of YouTube for the first time in its annual report, but did not split the cost of YouTube
    In Alphabet's financial report, the cost of server bandwidth and infrastructure maintenance is an astronomical figure of about 290 billion RMB, which is about 10 times the cost related to Tencent, but this includes the cost of all Google services around the world and Google Cloud sold by Google, so it does not have much reference value
    So how do you prove that YouTube is only a small part of this? Back in 2009, Arbor Networks, a data monitoring company, released a report saying that "the cost of YouTube bandwidth is almost zero
    " According to the fact that in that year, YouTube, as almost the only video website in the world, accounted for only 6% of the total network traffic with 100 billion broadcasts
    [6] data monitoring companies generally cooperate with ISP to set up bypass traffic monitoring on the backbone network for market research, similar to TV "ratings monitoring"
    Therefore, Peering traffic cannot be detected without entering the public network route
    The huge gap between Google's user active data and third-party monitoring data means that Peering has played a huge role
    The largest video website of your ball,The number of monthly active users exceeded 1
    5 billion in 2017, with 900 million video hours per month, accounting for only 17
    31% of Internet traffic
    In the following year, YouTube accounted for a much lower proportion of public network traffic than Netflix, even though it had dozens of times the number of broadcasts and users
    In 2014, Netflix accounted for 32
    39% of total uplink and downlink traffic in North America, while YouTube accounted for only 13
    [7] later, with the rise of the mobile Internet, without the blessing of Peering, YouTube accounted for more than 30% of the total mobile traffic in North America [7]
    This is the bandwidth cost that websites of this magnitude should have
    Then, Google worked out the mobile virtual operator Google Fi, tried to replicate the strategy of pipelining the typical operators, but because the base station construction involved in the mobile network is more complex than the fixed network construction, it has not made enough achievements so far
    There is no doubt that without the free bandwidth that Google Peering provides for YouTube (or bargaining power for operators), "becoming China's YouTube" is a pipe dream
    Because, to the operator bargaining power, what changes not only is the cost, but also the competition mode of the video website
    After reducing rigid operating costs to almost free, YouTube became the "bandwidth-free Baidu network disk"
    The essence of YouTube's PUGC business is not to incubate potential stars, but to keep unlimited typing monkeys-as long as you want to share videos on the Internet, whether you are professional creators or high-paste videos shot by young people in a small town, YouTube will provide you with high-definition, high-bit-rate, high-access video hosting services
    There will always be some monkeys who accidentally become Shakespeare, bringing high commercial value to YouTube and video creators
    In China, only short videos can use this model, and long video websites can not even beat the cost of server bandwidth when they do the UGC mode of raising unlimited monkeys
    After defeating competitors with low-cost and high-quality services, YouTube is purely post-paid for content creators
    It is almost impossible for Youtube to skip the data-sharing creator sharing plan and directly pay high signing fees for Youtuber, as domestic video sites do
    (this kind of thing only happens on a few channels such as YouTube Red Camera
    In addition to Google Peering's bandwidth savings, YouTube can also take full advantage of Google Cloud's in storage and computingIdle resources to offset the inventory cost of cloud computing business
    In addition, YouTube's accurate advertising recommendations also reuse Google Ads's R & D and data collection costs
    This part of the headlines and Tencent may have, but may not be able to reuse so well, bilibili and iqiyi certainly do not have
    Of course, it's not impossible to have to do it, but the more you do it like YouTube, the faster you die
    Reference: [1] "can Youku eat Tudou to make video sweetness", https://m
    html[2] "China's online video market", "three heroes compete for hegemony", Https://lmtw
    com/mzw/content/detail/id/66547/keyword_id/-1[3] Google wants' dark fiber', https://www
    com/news/google-wants-dark-fiber/[4] Why Google Fiber Is High-Speed Internet's Most Successful Failure, https://hbr
    org/2018/09/why-google-fiber-is-high-speed-internets-most-successful-failure[5] YouTube sucks on French ISP Free, and French regulators want to know why, Https://gigaom
    com/2013/01/02/youtube-sucks-on-french-isp-free-french-regulators-want-to-know-why/[6] "reports that the cost of YouTube bandwidth is almost zero"
    Sandvine Global Internet Phenomena Report reports / feedback of http://tech
    shtml[7] over the years
Related Content: