AI unicorn’s dependence on station B

Article source: AI Liaoyuan

Image source: Generated by AIImage source: Generated by AI

The AI company that spends the most money on advertising on Site B is about to change ownership.

According to observations from third-party data platforms familiar with Station B, in the second half of 2024, Kimi is no longer the AI company that can spend the most money on advertising on Station B, and related investment in byte-based AI products is increasing. Several million-level fan UP owners and MCN leaders at Station B also told Hu Xian that starting from the second half of 2024, byte series has gradually become the largest source of their AI business orders.

At the beginning of December 2024, a knowledge area head UP owner told Hu Xian that the main source of his monthly business orders was the different AI products owned by Byte, and there were 3 Byte series business orders waiting to be created immediately. I am too busy. At present, Byte has a variety of AI products, such as Dream, Bean Bag, Button, Cat Box, Smart Headphone ola… From large models to AI social networking, AI creative tools, AI education, etc., the entire product system has been covered.

In addition to bytes, the launch of AI companies such as “AI Six Tigers” has not slowed down. Marketing agents of several AI companies told Huxian that starting from the second half of 2024, AI companies, including AI Liuxiaohu, will become more competitive in terms of the launch schedule and amount of Station B.

“Everyone found that simply growing users may not completely allow them to win this AI product competition. Therefore, many customers began to spend more money on brand marketing in Q3 to build brand mentality; and in Q4, it may be due to pressure from investors or performance evaluations at the end of the year, and the overall marketing strategy will be biased towards user growth.” One of the marketing agents concluded.

As various AI products attract users on site B, more and more UP owners have begun to make money from AI advertising in the past year.

Hu sniffed learned that in early 2024, there are three main UP owners who can receive advertisements from AI companies: software applications, campuses, and workplaces. The functions that AI products need to be promoted also mainly focus on improving the efficiency of office and learning scenarios. Among them, UP owners in the software application category often write code themselves, or use AI to implement or start businesses, making it easier to obtain a budget for brand mental marketing.

In the second half of the year, the situation changed. With the maturity of large model technology, more types of AI products are entering the market, and more and more scenarios are being penetrated by AI. For example, products such as AI search, emotional companionship, AI-generated videos, and AI-generated music are gradually taking shape.

As a result, UP owners in more categories have also begun to receive AI advertisers, such as UP owners in games, ghosts and animals, and video clips.

At the same time, the advertising revenue of Station B has also risen. By the second quarter of 2024, the number of AI industry customers at Station B has increased by more than 200% compared with the previous year. In the third quarter, Station B’s advertising business revenue was 2.09 billion yuan, a year-on-year increase of 28%, and it also achieved a historic profit.

The situation where everyone is happy is accompanied by some unavoidable confusion:

·A launch agent that has served many AI companies, while earning the service fees of the AI company, is puzzled and feels that the AI company is wasting a lot of money-why does the AI company make a lot of launches before its products have been verified and feasible? If we find PMF (Product-Market Fit)?

·A UP owner who did not receive advertisements from AI companies in 2023, but the revenue from AI commercial orders received in 2024 accounted for 30% of his overall revenue is happy and worried: Will it last in 2025?

01、AI unicorn, looking for a “living” path at Station B

Since March 18, 2024, Douyin issued a large number of advertisements to regulate and restrict the streaming of AIGC products, the situation of non-byte products being released in Douyin has changed. Companies such as Dark Side of the Moon, MiniMax, and Smart Spectrum have shifted their focus to platforms such as Fast Hand, Xiaohongshu, Station B, and WeChat.

Xiang Huxian, a manager responsible for streaming and user growth at AI Six Xiaohu Company, said: “Currently, our company is unable to deliver on all traffic ports in Bytesystem. Even the tap content of some creators who publish content on multiple platforms can be published on Station B, but it will be downloaded when Douyin is online.” This person showed Hu Xian the company’s main “investment” situation in the second half of 2024: Station B accounted for almost more than half of the weight, and the weight of fast hands began to increase in the fourth quarter of 2024. As of the end of 2024, Station B and fast hands contributed more than 85% of the investment channels of this six tiger company.

Another product leader of AI Liuxiaohu Company said that his company’s streaming in Douyin would basically stop after May 2024. “When bean buns gradually rose, our investment in Dianxi began to plummet. On the one hand, there are restrictions, and on the other hand, a large number of creators are saturated with orders for bean buns.”

Many agents, MCN institutions and bloggers covering the above different platforms said to Huxian that Station B has become a preferred channel for AI companies to release content. We summarized several main reasons from exchanges:

·Whether it is an app or a webpage, compared with short content platforms such as Xiaohongshu, Douyin or Quick Hand, Station B can provide more different advertising spaces, and customers can choose more widely and flexibly. For example, there can be an advertising space to guide downloading under relevant videos. For example, video advertisements placed by customers can appear in the information stream, such as the location of the homepage focus map, etc.;

·”The content of Station B is mainly medium and long videos, which is suitable for comprehensive analysis of AI products. Short content is suitable for planting grass, while medium and long content is suitable for in-depth understanding and learning of products. Therefore, for AI products, Station B is very suitable for educating users.” The relevant person in charge of the “Forward Four” MCN organization, which is good at serving technology Internet customers, emphasized this. If a video lasts for more than 5 minutes on Douyin, Fast Hand, Video Number, or Little Red Book, then in order to comply with the user habits of short content platforms, the video needs to be deleted, which will lose some information.

·”Creating stories” is also a customer demand that has been mentioned many times. According to the analysis of the agent, some UP masters in Station B have stronger technical backgrounds, are more capable of integrating, and are good at creating events. For example, UP owner @ Lin Yi LYi has an advertising video. Based on a domestic model, he developed a game in three months and created six AI-driven NPCs. In the game, these six AIs lived in the same apartment, and then observed their evolution in interests, personalities, and behaviors. The video currently has 1.72 million views on Station B, with nearly 4000 comments.@ The commercial person in charge of Lin Yi’s LYi account told Hu Xian that similar cooperation is no longer a traditional business-order relationship. Customers can jointly develop some long-term operation projects with UP owners, package them as core cases, and even follow up on their projects. Promote and promote them. As a result, Station B is equivalent to providing advertisers with the option of “creating stories”.

According to observations by many UP hosts over the past year in their comment area,”In the first half of this year, users ‘focus on AI video content was mainly on what abstract things AI was doing and what style it was doing, which tended to be banter; In the second half of the year, when AI companies’ products had been greatly upgraded and improved, users began to face AI and began to regard it more seriously as a productivity tool. Whether in terms of mind or content atmosphere, there have been major changes. People are starting to think it is something useful, and at the same time, it also means that the AI industry is getting better.”

However, AI model products are still in the stage of educating users. A technology UP owner familiar with various AI products said frankly,”At present, the interaction difficulty of conversational AI products is still relatively high, not smooth, and is still in the middle, and the user cost is too high.” This explains to some extent why the user retention effect of AI products is not good.

Product upgrades do not mean that the needs of more users can be solved, nor does it mean that people can use them at high frequencies or even pay for them. There is still a long way to go before maturity.

However, pressure from competition and investors did not leave enough time for AI unicorns to mature.

According to Orange IT data, from January 1 to December 5, 2024, there were 439 financing cases in the domestic AI field, with a total financing amount exceeding 56.4 billion yuan, about 80% of last year. Nowadays, there are fewer and fewer venture funds on the market, and the financing environment is deteriorating.

An investor who invested in one of the six AI tigers revealed his helplessness to Huxiao,”The valuations of large models are generally very high now. VCs are basically unable to invest and need deeper money. Next, who has the ability to cast the next round? It may be state-owned assets or money from overseas sovereign funds other than the United States.” He further explained,”How can you persuade people to invest in you? It is no longer feasible to tell the story of iterative upgrades of large models. More specific product capabilities are needed.”

“Big model companies now need to prove themselves. Without a certain number of users and interactions, they may not be able to prove their products, technologies, and model capabilities, and subsequent financing will be more difficult.” The above-mentioned person said.

This is also consistent with the feeling of launch agents starting from the second half of 2024: AI unicorns are increasingly investing.

Coincidentally, in the same month, several large models successively completed new rounds of financing in the second half of 2024. Smart Vision announced that it had completed 3 billion yuan in financing and stepped up to announce that it had completed hundreds of millions of dollars in financing. According to the above-mentioned investors, there are one or two large model companies in the circle that “have completed financing in the second half of the year but have not been officially announced.” If those who cannot get financing and cannot get admission tickets quickly, they will have no way to continue. If they develop upwards, they can only shut down or transform.

This may be the helplessness expressed by the above-mentioned investors.

02、How can the fanatical investment continue?

AppGrowning platform data shows that since March this year, Kimi has spent tens of millions of yuan in advertising almost every month. An industry insider said,”In October and November, Kimi’s advertising amount on the entire platform exceeded 100 million, and bean buns were even more.” Some relevant sources said that the advertising amount of a certain AI head product other than Kimi in the entire platform during the same period was approximately “between 220 million yuan and 270 million yuan.”

In addition, according to the “AI New List” data: from 8.1 to 10.29, the total number of external advertisements placed by Kimi in the past three months exceeded 2500, of which more than 2100 pieces of content (including graphics and videos) have been posted in fast hands, accounting for 83.6% of the total.

In the second half of 2024, some AI unicorns will begin to switch to fast hands. Some analysts believe that the logic behind this behavior is: placing advertisements on Short Video platforms, guiding downloads through information flow, and paying through CPA or CPM. The effect is relatively stable and controllable, as long as money is spent continuously. Short videos have a shorter content production cycle, and users can be guided to complete product downloads within a shorter update cycle.

Hidden behind this behavior are the financing challenges and cost pressures that AI unicorns are facing.

Previously, media reported that Kimi’s cost of obtaining customers on platforms such as Station B was as high as 30 yuan per person.However, according to Hu Xian’s verification from various parties, this data is false.Relevant sources revealed that the industry’s comprehensive bidding situation is not that high“In September, the average bid range at Station B was 10-15 yuan, and the current AI customer effectiveness optimization target of Station B was mostly an activation single bid, with the average bid around 12 yuan.” A person familiar with the matter revealed.

“Some AI unicorns burn money hard in terms of investment. I don’t understand this phenomenon even when I launch.” An agent responsible for investing in several AI unicorns said,”If it weren’t for the rapid increase in data in the short term to attract financing, it would be difficult to understand why everyone would invest in money without any rules.” He said that some AI companies will even advertise to bloggers with poor data performance, irrelevant categories, or extremely low cost performance.

According to his estimation, the customer acquisition cost of AI products can range from 12 yuan, 15 yuan, and 25 yuan, no matter which one is relatively expensive. “The specific price depends on the user group acquired. This is related to factors such as Android, iOS and region, age, etc. The higher the quality, the more expensive the user. High-quality users usually refer to those who use the product frequently and can exert their product capabilities.”

He believes that it is already unreasonable for a group of startups that have suffered serious losses, cannot find a suitable business model in the short term, and have not yet been verified and feasible at this stage.

UP owners, agents, and MCN often also have a large number of customers from other technology industries. After the contact process, they feel that AI company customers still lack launch experience.

·On the one hand, the product is still slowly maturing, and its selling points are constantly changing. Moreover, there are not many cases that can be referred to when launching this emerging track;

·On the other hand, compared with traditional digital manufacturers such as OPPO and vivo, which have clearer strategies and more focused user groups when launching, AI companies often do not know who their products are for and who are more willing to use them.

In terms of marketing, China’s big model companies have no overseas cases to imitate. In contrast, the foreign AI unicorn Anthropic has not yet placed advertisements, mainly relying on every release of relevant technology to increase its popularity, rather than traditional advertising channels; the same is true for OpenAI, its first marketing director, who will be in charge of 2024. Just arrived in mid-December.

According to the judgment of many marketing professionals, the aggressive launch of domestic large model unicorns may continue for at least half a year.

The root cause of today’s dilemma is that domestic AI products are still not enough to become stable and reliable efficiency tools. For an entrepreneurial track that has only existed for a year or two, we should remain tolerant and give time to innovate.

For a rapidly emerging industry, an AI industry headhunter has witnessed the emotional ups and downs of the big model track in the past two years. He said that at first, everyone joined the big model with high spirits, but after two years, there were still no strong ones. When profitable products came out, everyone was still a little disappointed. “I will come into contact with many candidates from MIT and Tsinghua University. The thing that everyone pays most attention to is often how to stay at the table. They are more concerned about which big model company will raise more money next and which one should join.”

“It may still be a concern about the huge uncertainty of the times.” He added that in the exchange of eyes, he could feel everyone’s desire, vitality, and confusion. “This emotion should be transmitted from top to bottom. Perhaps the bosses of big model companies are also very confused.”

In 2025, big model companies are still full of uncertainty.

发表回复

您的邮箱地址不会被公开。 必填项已用 * 标注