Competing with Google, can the AI model really support the IPO of Going Out and Asking?

Source: Bullet Finance Wang Yajing

November 30, 2022 is a day of subversive significance for the AI industry.

On this day, “ChatGPT” was born out of nowhere, which produced a huge circle-breaking effect. Google, Microsoft, Apple, Baidu and other technology giants have announced follow-up plans. No one does not want to be able to occupy a place in this blue ocean.

Under this carnival of algorithms, Mobvoi Inc., an AI company that had received Google’s strategic investment, was the first to attack the Hong Kong Stock Exchange. On May 30, Mobask submitted a prospectus to the Hong Kong Stock Exchange. According to the prospectus, this is an AI company with generative AI and voice interaction technology as its core business.

This 597-page prospectus runs through AIGC (“AI generated content”, meaning artificial intelligence generated content) and general large-scale model concepts that Momenask highly recommends.

There is no doubt that this is the hottest trend in the AI industry this year, but it cannot support the people who go out to ask questions. According to the prospectus, until 2022, AIGC will only contribute 39.857 million yuan to the company, accounting for 8% of the total revenue.

As for the general-purpose large model “Sequence Monkey” that was just upgraded this year, Li Zhifei, the founder of Gomenwen, once scored at the 2023AIGC strategy conference held in April this year, "If GPT-4 is 100 points, our model is only 50 points. "

Go out and ask, using popular concepts to tell stories, can it really impress the capital market?

1. Benchmark Google’s ambition

Mobask was founded in 2012 by Li Zhifei, a former Google scientist.

Although it has been deeply involved in the AI field for many years, the prominent general-purpose large-scale model that has been emphasized in the prospectus until 2020 was not launched until 2020. It was named “UCLAI”, and this year it was upgraded to “Serial Monkey”.

Generally speaking, a general-purpose large model refers to an artificial intelligence model that can process data of various types and fields and provide multiple intelligent services. For example, the popular ChatGPT is a product based on the general large model technology.

According to the introduction in the prospectus, using the general-purpose large-scale model “Sequence Monkey” as the underlying technology, Mobvoy has currently launched three major solutions: AIGC, enterprise services, and AIoT.

(Picture / Go out and ask prospectus)

Specifically, AIGC is mainly aimed at creators, and its products include “Magic Sound Workshop” and overseas version “DupDub” that provide AI dubbing, “Magic Writing” that can automatically generate articles, “Wonderful Yuan” that assists video generation and virtual live broadcasting. ".

Enterprise services mainly provide AI voice interaction and intelligent customer service for enterprises. For example, it provides in-vehicle voice interaction control, and implements functions such as answering calls, navigating, checking the weather, and booking hotels through default or custom hot words.

The audience of AIoT is consumers, and the products are mainly smart watches, smart treadmills and other software and hardware combination devices, which are mainly used in wearables, cars and smart homes.

Li Zhifei is an ambitious entrepreneur. In an exclusive interview with Entrepreneurship in 2021, he said: “I hope to build a company like Google, with great technology and a successful business model, so that technology and business can continue. cycle iteratively.”

Driven by this vigorous ambition, Mobvoi has won the favor of Google and many well-known investment institutions.

According to the prospectus, from February 2013 to September 2019, Mobvoy conducted seven rounds of financing. The strategic investment team included Sequoia Capital, ZhenFund, SIG Heiner Asia, Google (Google), Volkswagen Group, etc. Well-known enterprises and capital institutions, the cumulative financing amount is about 233 million US dollars.

(Picture / Go out and ask prospectus)

The post-investment valuation of the company has skyrocketed from US$5.1 million in February 2013 to US$757 million in September 2019, and the valuation has doubled by nearly 150 times in just six years.

2. R&D and marketing costs are equal

Surrounded by capital and at the same time standing at the forefront of the industry, it seems a bit “high self-esteem” to go out and ask.

In the prospectus, Momenwen claims that it is “the eighth place in AIGC’s revenue market in the world and the number one in Asia” and “one of the few AI companies in Asia with the ability to build general-purpose large-scale models.”

Whether it can be worthy of these famous names is still hard to judge, but what can be judged is that Going out to ask is one of the companies most eager to go public at present.

In the prospectus, China Insights Consulting, commissioned by Mobvoy, made statistics on the top 10 AI technology companies in the world that have commercialized AIGC (hereinafter referred to as: Global TOP10). In the global TOP10 revenue ranking, Mobvoy ranks 8th.

• ** (Picture / Go out to ask the prospectus) *

Compared with other companies, Going out and asking questions is obviously lagging behind. Calculated by 2022 revenue, AIGC’s service income is the highest in an AI R&D and deployment company established in 2015 in the United States, with an amount of US$65 million to US$95 million, while Mobvoi is US$5.7 million.

From the perspective of AIGC’s revenue alone, there is at least an 11-fold gap between Mobvoi and the world’s No. 1 company.

Even so, Going out to ask is still seeking to be the first listed company. “Interface News · Bullet Finance” sorted out and found that, except for going out to ask for the submission of prospectuses, none of the other nine AI companies in the above list were listed.

In addition, “Interface News·Bullet Finance” found that, as an enterprise that urgently needs to flex its muscles in the field of AI, it did not devote more resources to research and development when it went out to ask questions.

According to the prospectus, from 2020 to 2022, the research and development expenditure of Mobwenwen will be 97.1 million yuan, 91.51 million yuan, and 119 million yuan, accounting for 21.18%, 23.0% and 23.7% of the revenue respectively.

That’s almost on par with marketing expenses. During the above-mentioned period of time, the sales and marketing expenses of the enterprise were 74.528 million yuan, 106 million yuan, and 97.12 million yuan, respectively, accounting for 28.2%, 26.6%, and 19.4% of the revenue in the same period. Among them, an average of more than 60 million yuan per year is used for promotion and advertising.

(Picture / Go out and ask prospectus)

Such a huge amount of marketing and promotion has eroded the company’s net profit to a certain extent.

The prospectus shows that from 2020 to 2022, the company’s revenue will be 265 million yuan, 398 million yuan, and 500 million yuan, respectively. During the same period, the net profit of the enterprise was 56.146 million yuan, -276 million yuan, and -670 million yuan respectively.

(Picture / Go out and ask prospectus)

However, Momenwen also disclosed the net profit under non-IFRS, excluding interference factors such as changes in the book value of preferred shares and common shares, which is considered to reflect the company’s main business operations to a certain extent.

According to the prospectus, the adjusted net profit of Mobwenwen in 2020-2022 is -157 million yuan, -73.439 million yuan and 109 million yuan respectively. From this point of view, the company achieved the adjusted net profit in 2022 for the first time. Become regular.

(Picture / Go out and ask prospectus)

Judging from the adjusted net profit, the company’s main business will be profitable in 2022, making it one of the few AI companies in the industry that can achieve profitability.

However, it should be noted that the “adjusted net profit” in the Hong Kong stock prospectus is generally disclosed voluntarily by the company. The items adjusted by each company are different, and investors need to further comprehensively analyze the company’s overall business development. judge.

3. Difficult transformation

In fact, in the early days of entrepreneurship, Mimenwenwen did not focus on software development, but focused on the development of voice interaction, such as the development of hardware products such as smart watches, wireless smart headphones, smart rearview mirrors, and smart speakers. Among them, the smart watch TicWatch series is the leading product of the company.

(Picture / Go out to ask the official website)

However, compared with hardware products such as smart watches, AI software can tell more stories, and it is easier to win the favor of the capital market.

According to the “2023 Global Artificial Intelligence Market Report”, it is estimated that by 2028, the global artificial intelligence market will reach 2.5 trillion US dollars. Machine learning, natural language processing, and computer vision are the three major application areas of the artificial intelligence market.

In 2022, Mobvoi is consciously transitioning to AI software. According to the prospectus, from 2020 to 2022, AI software solutions will contribute 17%, 15%, and 60.6% of operating income respectively.

(Picture / Go out and ask prospectus)

The AI software solutions business segment is mainly composed of two parts: “AIGC solutions and AI enterprise solutions”. 44.476 million yuan will rise to 263 million yuan in 2022, and the proportion of revenue will also rise from 16.8% to 52.6%.

However, “Interface News·Bullet Finance” noticed that the current income from AIGC solutions is still very weak.

Although Mobwenwen stated in the prospectus that “as early as 2020, the company has developed a general-purpose large-scale model and vertical optimization technology modules, and then developed leading AIGC technology and applications specifically for global content creators and enterprises”, but in 2020 In 2021, 2021 and 2022, the AIGC solution revenue of Going out to ask is 496,000 yuan, 6.822 million yuan and 39.857 million yuan respectively, accounting for 0.2%, 1.7% and 8.0% of the total revenue respectively. It can be seen that this A business scale is still very small.

In addition, there is another outstanding problem that may cause many investors to worry about going out to ask questions-the excessive concentration of large customers.

According to the prospectus, from 2020 to 2022, the revenue from the top five customers accounted for 22.4%, 37.0% and 62.8% of the total revenue respectively. Among them, the revenue from the largest customer accounted for 8.8%, 24.1% and 42.6% of the total revenue respectively.

Just in 2022, Mobvoi provided AI enterprise solutions and intellectual property arrangement services for the largest customer from the automotive industry, thereby earning 213 million yuan in revenue, accounting for 42.6% of the total revenue.

(Picture / Go out and ask prospectus)

In fact, when a company’s performance has a strong relationship with its regular customers, it poses a test for the company to maintain stable and sustainable development, especially when external competition is becoming increasingly fierce.

According to the incomplete statistics of “Interface News · Bullet Finance”, as of now, Baidu’s “Wen Xin Yi Yan”, Ali’s “Tong Yi Qian Wen”, SenseTime’s “Consultation SenseChat”, 360’s “360 Smart Brain”, etc. General-purpose large models have been released or officially announced for internal testing.

On April 9, Minsheng Securities released a research report and statistics. At least 30 large-scale models have appeared in China. The producers include Internet giants, AI concept listed companies, leading server companies, scientific research institutes and primary market startups. The parameter scale of the large model is close to, even surpassing the scale of ChatGPT (hundreds of billions).

It is undeniable that in the field of transforming AIGC and general large-scale model technology, Going out to ask has a first-mover advantage. But in the face of the strong entry of giants, can a company with a revenue of only 500 million yuan compete with it?

After all, the difficulty of playing against the giants should be deeply touched when you go out and ask. When Li Zhifei recalled the peak moment of the company from 2018 to 2019, he once told the media that when he went out to ask, he directly confronted giants such as Xiaomi, Baidu, Huawei, etc., and was completely helpless in the face of the giant’s subsidy behavior.

“That was the first time I realized that the competition with giants is very scary.” Li Zhifei said.

After that, Going Out and Asking cut off six or seven product lines, leaving only two, closed all 20 offline stores, and reduced the team size from 1,000 to 500.

Today, at least in terms of the number of employees, the vitality of going out and asking questions has not yet recovered. According to the prospectus, as of the latest practicable date, the team consists of more than 400 employees.

If it successfully lands on the Hong Kong Stock Exchange this time, it will become the “first AIGC stock” in Hong Kong stocks. However, it is undoubtedly that it has not yet formed a large enough scale in terms of AIGC solution revenue, and whether it can really support this Titles stand the test of time.