TENCENT, it's a little scary to be so quiet.

In the first quarter of 2023, China's internet is surging with hidden currents.

JD.com took the lead in starting a price war, while Pinduoduo struggled to respond and faced negative news. ByteDance is increasing its investment in local life in China to compete with MEITUAN-W, while TikTok faces various risks overseas. Alibaba welcomed founder Jack Ma back to China and then announced the earth-shattering 1+6+N plan, claiming to use large models to completely redo all its business. The big model war sparked by Chatgpt has created a new opportunity for Baidu, Alibaba Cloud, Huawei, iFlytek, 360, Zhihu, Sensetime, Kunlun Wanjing, and other companies.

But during the AIGC and big model frenzy, what is Tencent, the dominant internet company in China, doing? While others are fighting tooth and nail, Tencent seems a bit scary in its quietness.

"Every time there is a huge change in the industry, Tencent is so quiet. As a person close to Tencent said, new business for internet companies no longer involves "futures-style R&D" with PPT presentations. Either stay silent or make a big move."

Now, as spring 2023 draws to a close, the cicadas of summer are about to emerge.

It is rumored that Ma Huateng is personally overseeing Tencent's artificial intelligence big model project. Only by completely overturning the technology can the industry be reshuffled. In the short video battle, Tencent has no chance of winning, at most it can only defend. If the GPT big model is a deterministic technological revolution, Ma Huateng must bet on it, and then Tencent can continue to shine for the next decade.

Tencent's aging traffic

In recent years, every time Ma Huateng holds a meeting, he talks about short videos, and the outside world is tired of hearing it.

What is the underlying logic?

As a senior traffic operator, Ma Huateng is too aware of the huge differences between traffic sources. WeChat, as the largest source of traffic within the Tencent system, used to be mainly social and graphic traffic. The emergence of short videos has had a huge impact on WeChat traffic in terms of form. It has even overturned the entire advertising industry twice.

Compared to short videos, WeChat traffic is aging in three dimensions.

First, user time.

Short videos have a lower entry barrier and more direct presentation than graphics, and with exaggerated performances, they are easy to become addictive. After all, even those who cannot read can swipe through them. It is difficult for WeChat Moments to generate such a massive amount of content.

Another dimension of massive content comes from the range of content providers. In addition to China's Douyin, foreign TikTok has almost become a global app. This means that under the power of algorithmic recommendation, users can watch short videos from all over the world with zero threshold through one screen.

But WeChat cannot do this. Although the number of WeChat friends has increased from 5,000 to 10,000 since version 8.0, even if all these positions are filled, if the content in Moments is not what users are interested in, it will not generate value. So, short videos have replaced Moments and become the new traffic source, taking away users' time.

Secondly, cognitive indoctrination.

In recent years, the proportion of brand advertising on short videos has already surpassed other channels. The advertising industry is a war for screen space, and short videos have almost eaten up all the brand advertising share of other channels. This not only brings about a decrease in the value of traffic, but also an increase in customer acquisition costs.

A service provider who does home decoration advertising on both Douyin and WeChat Moments told Lujushangye Review that from 2018 to now, the cost of acquiring a single customer from image ads on Moments has risen from 200-300 to 1300-1500.

But what troubles him the most is not the rising cost of customer acquisition, but the fact that even if he has money, he can't spend it. Because the home decoration industry is not a strong category, once categories such as cosmetics and cars carry out large-scale advertising in the same traffic scale, his scheduled advertising will be postponed, and the situation of limited traffic will become more apparent in the past year.

Especially, bloggers in short videos can use their fans to do advertising implantation, and even use their personalities to endorse and influence consumers' minds, presenting an interactive effect. This is something that the WeChat ecosystem could not achieve before. Although the problem has been alleviated after the introduction of Video Account, user habits have already been lost.

Finally, transactional attributes.

In the era where sales are greater than brand, the integration of product and effect has become a topic that most companies are discussing. Short videos can directly combine advertising with sales, which is something that other traffic channels cannot achieve.

The aforementioned service provider also compared the overall efficiency of advertising content on Douyin and Moments to Lujushangye Review. The overall efficiency of Douyin advertising can reach 40% to 50%, while the efficiency of Moments advertising is only 20% to 30%. Even for old accounts with stable models and tags, the efficiency is only 30% to 40%, which is just the lower limit of reaching Douyin.

Similarly, if you watch enough Douyin, you will be familiar with Pinduoduo's information flow advertising. They use the same short and fast script as other information flow ads, with the same magical lines, and the only goal is to make you click on the link below and go to Pinduoduo to "grab the iPhone for 1 cent".

Before this, Pinduoduo's "Dragon's Land" was still in WeChat group chats for wool enthusiasts. With TENCENT's investment, letting the ByteDance system make money is not Pinduoduo's "eating inside and out", but the efficiency of short video traffic conversion is higher.

Taking into account the above three points, short videos are a better choice than WeChat for any advertising client. As the basic business of TENCENT, the traffic business is bound to face increasing pressure. If this problem cannot be substantively resolved, it is like boiling a frog in warm water.

Ma Huateng has long seen this problem and has been trying to solve it, but there is currently no good solution. Can Video Account be separated from WeChat and face competition from Douyin independently? TENCENT has not made this radical choice yet, because even if it does, the chances of success are not high.

Business Peak, From Attack to Defense

Before the big decline of the Internet, TENCENT's investment was incredibly successful. Behind many of China's Internet giants, there is the shadow of TENCENT, including JD.com, MEITUAN-W, Pinduoduo, Beike, Didi, Kuaishou, Bilibili, Zhihu... The list is long, but the decline of the Internet industry has made it difficult for TENCENT to sit back and enjoy its success as a landlord. Therefore, last year they chose to proactively deal with JD.com and MEITUAN-W.

Recently, Parallel Nebula Investment Limited ("PNI"), an investment entity indirectly controlled by Tencent Plus Partners, reduced its stake in Kuaishou by about 51 million shares. At the latest stock price, this is at least in the range of HKD 2.5 billion.

The easy-to-accept explanation for TENCENT's investment downturn in leading Internet companies is that it is a continuation of cost reduction and efficiency improvement in the investment field. But the obvious fact is that when the development of the invested company is not as rapid as before, it is more promising to continue to "sell water" in the financial investment field, and to do business growth by oneself.

For any Internet company, not developing is an unacceptable result.

In the first half of the mobile Internet, TENCENT relied on WeChat to dominate, and then developed King of Glory, becoming a national harvester. The huge success of the investment department also made TENCENT's territory as stable as Mount Tai. The loosening of the front line was not due to the huge success of ByteDance in TENCENT's advantageous areas. The business that was originally in a growth trend also shifted from attack to defense at this time.

One manifestation of this shift is that the profit growth of TENCENT's core business has shifted from scale-driven to cost-cutting-driven.

TENCENT's 2022 financial report shows that sales and marketing expenses alone have decreased from CNY 11.616 billion last year to CNY 6.115 billion, a year-on-year decrease of almost half.

In addition, TENCENT's other main businesses have also bid farewell to high growth and even entered a decline: in the international market, game revenue excluding exchange rate and related adjustments increased by 5% to CNY 46.8 billion, while domestic market game revenue decreased by 4% to CNY 123.9 billion, and online advertising revenue decreased by 7%, with media advertising revenue even decreasing by as much as 20% year-on-year.

This is not only due to the peak of business volume, but also because the business potential of the mobile Internet is disappearing as the growth of mobile phone users saturates. TENCENT's strategy of using new mobile users to bring incremental business revenue is completely outdated.

Under the logic of stock competition, if TENCENT still relies on investment to lay out its business and does not seek new growth curves, its result will be to be constantly attacked and occupied by companies like ByteDance in the core business field. Except for passive defense against benchmark businesses, there are not many opportunities for active attack. Therefore, the video account that has been given the title of "the hope of the village" by Ma Huateng and continues to increase its presence is also a typical defensive business. In the latest WeChat public class, lecturer Zhang Xiaochao also mentioned that WeChat video accounts are more like components of the WeChat ecosystem, relying on underlying atomic capabilities to allow users to circulate in various segmented scenarios of WeChat, with a convergent and folded form.

An Internet practitioner who has been tracking the short video industry for a long time commented on this: "The WeChat team's understanding of video accounts is still at the level of a certain factory in 2019."

In other words, the upper limit of the future growth of video accounts is largely not in Tencent's own hands. Although it is Tencent's "hope of the village", it still cannot become Tencent's entry ticket for the next era.

The crisis behind the large model

Even though there is no mature commercialization plan yet, no one doubts that large AI models are the entry ticket for the next era of Internet companies.

In the "pre-AI era" represented by short videos, Tencent can still rely on the number of users accumulated in the past ten years to compete with it. But when various companies have successively launched their own general large models and rushed into the "iPhone moment", if Tencent is "missing" in the large model, it will be the seven-inch moment.

Tencent is not idle. In fact, in February of this year, Tencent established the "Mixed Yuan Assistant" project team for Chatgpt.

According to the disclosed list, many heavyweight employees including Tencent's chief scientist Zhang Zhengyou participated, and in the progress in early April, they won the first place in the five most authoritative cross-modal video retrieval datasets of MSR-VTT, MSVD, LSMDC, DiDeMo, and ActivityNet.

Compared with large model manufacturers such as Baidu and Alibaba Cloud that have already released products, Mixed Yuan Assistant has lost many opportunities in the competition in the track due to its lag in release time. These opportunities include but are not limited to C-end users' testing of large models and the user habits established thereafter.

More importantly, in the field of collaborative office, DingTalk and other manufacturers have already demonstrated a certain degree of empowerment effect of their own large models. For example, at the DingTalk Peak Conference in the spring of 2023, DingTalk President Ye Jun has been able to realize functions such as dialogue or taking pictures to generate applications, one-click creation of answer robots that can automatically learn documents, automatic creation of document copywriting, and intelligent extraction of video conference and group chat message summaries by accessing the slash "/" of Tongyi Qianwen.

Although these functions can only be opened to enterprise version DingTalk and personal VIP customers for internal testing so far, most of the content generated by these functions has undergone compliance review. In the process of promoting enterprise intelligence, DingTalk has already taken the lead over WeChat Work.

It can be seen that the corresponding scenarios for content generation in the IM track where WeChat and Mixed Yuan Assistant are located are obviously more complex than e-commerce and collaborative office. Even if Mixed Yuan Assistant may have advantages in performance, it will only delay its release time. The time difference generated may already be enough for other vendors to carry out simple commercial landing practices. When will the large model empower TENCENT's current business, especially WeChat, which is most likely to be empowered? This is everyone's expectation.

After all, on the other side of the ocean, large models have begun to empower Microsoft products such as Bing and Skype. Google has felt an unprecedented sense of crisis and quickly adjusted its strategy. Meta, which previously bet on the metaverse, also found itself on the wrong path.

After all, for today's Internet companies, large models are very similar to the mobile Internet ten years ago. If they can't get tickets, they are likely to be beaten to death on the beach.

According to multiple reliable sources, Ma Huateng is currently personally supervising the large model project internally. Previously, Tencent President Liu Chiping publicly stated on a conference call that Tencent has the strength to create competitive products in the field of AI large models, because Tencent has resources for usage scenarios, advantages in data, blessings in cloud computing infrastructure, and the ability to create artificial intelligence applications for long-term practice in providing assistance to existing businesses.

He believes: "Our first product will be a product after multiple iterations, and the whole process will be long-term. Chatbots are just one of the many products we will release in the future. We regard it as a long-term development opportunity rather than an urgent threat that needs to be resolved. We regard it as a work focus, but with a long-term attitude."

Obviously, Tencent does not want to easily make a move, but once it does, it will definitely want to win.

TENCENT still needs to fight the "old white rabbit"

Almost all companies have the disease of large enterprises. The terrible thing about Tencent is that many VP-level executives have become old white rabbits.

There are also internal corruption issues. As we all know, Tencent relies on social and gaming to accumulate an unparalleled user base in China. As long as social users are introduced into games, regardless of the high or low payment ratio, corresponding game recharge revenue will be generated. Based on this, third-party game channel publishers will inevitably contact Tencent-related personnel, and the rent-seeking in the middle can be imagined.

A supplier close to Tencent told Lujiu Business Review that in terms of Tencent's internal mechanism, once there is a change in the top management, the middle-level personnel who are the main force of business promotion will inevitably undergo a major reshuffle, which leads to extremely unstable internal morale of the group. In order to prevent rent-seeking, the group must regularly replace the corresponding responsible persons, but this forces the relevant personnel to have a sense of crisis, and not focusing on business and timely realization becomes the only choice, which becomes a dead loop.

A person close to Tencent said that the distance between Tencent's core high-level executives and grassroots employees is too far, which has caused many information to be unable to be transmitted upwards. For example, if Pony has a business trip, the surrounding area will be crowded with various VPs and senior VPs.

This has been summarized by another person close to Tencent as the "gravity" of large companies-whether it is external introduction or promotion from the grassroots, the probability of the emergence of "strong people" is lower than usual. Zhang Xiaolong was only an exception back then. Many VPs seem to be responsible for business, but executives at this level tend to mix things up more and are less responsible for the final results. The big corporate disease of TENCENT is reflected in the fact that projects personally handled by members of the TENCENT General Office are often good. However, at the VP level, many things are left unresolved, and their combat effectiveness can be said to be lacking. (Executives in the TENCENT General Office are responsible for critical decision-making issues such as controlling the company's strategic transformation and architectural adjustments. In addition, they must grasp product direction and handle coordination when cooperating across business groups. To some extent, the TENCENT General Office plays two critical roles: investor and coordinator.)

Lujia Business Review once wrote that in TENCENT's "Xiao'e Pinpin" project, there have been cases of upward management based on false orders, and the same person in charge has also led the TENCENT digital collection project "Huanhe", which was also shut down six months ago.

Earlier, TENCENT also continued to bet on the VR track, and for this reason, it promoted the acquisition of Black Shark mobile phones. In the early stages, there were hundreds of millions of yuan in investment, but also due to internal corruption, TENCENT terminated its cooperation with Black Shark, and the Black Shark mobile phone brand disappeared from the mobile phone industry as a result.

It can be seen that TENCENT has also explored new business curves several times. But either the business person in charge is not capable, or corruption occurs internally, always casting uncertainty over the business prospects. Although the members of the Hunyuan Assistant project are of good quality, TENCENT still needs to work hard under internal management to ensure that these capable assistants "do not love money and are not afraid of death."

"A person from the TENCENT Research Institute once told Lujia Business Review that for a considerable period of development time, TENCENT has tried countless times to take off the coat of "social and gaming," but has never been successful due to the aforementioned reasons."

Some people say that Zhang Xiaolong is just an exception. After all, in the era of mobile Internet, WeChat brought by Zhang Xiaolong gave TENCENT a ticket. Today, the large GPT artificial intelligence model has given TENCENT the opportunity to use its accumulated technology with certainty. After "stabilizing the domestic situation," the era has given TENCENT the opportunity to redefine itself.

This time, can TENCENT seize the opportunity smoothly?