Xiaodu’s commercialization lost its way

Article | Newly Excerpted Business Review, One White

This year, Xiaodu’s path forward seems to be "overshadowed".

At the end of March, Xiaodu launched a promotion campaign for the digital collection of Genesis, promising to promote 500 people to consumers and airdrop the Palace Museum’s digital collection "Listening to the Qin" for free. Some people successfully promoted more than 500 people, but Xiaodu did not fulfill the promise afterwards. Instead, it replaced it with other low-value digital collections, and was complained by users for suspected false propaganda.

In July, a woman found multiple private videos of guests staying in a small speaker in a bed and breakfast room, which was questioned as an infringement of user privacy.

After four months, Xiaodu was fined 250,000 yuan for propaganda in the name of MIIT, which once again made headlines.

Xiaodu, which Li Yanhong calls "a revolutionary new species of AI in the Chinese market", has been highly expected, and its AI voice assistant hidden behind the smart speaker carrier is Baidu’s shield for the next decade.

However, it has been two years since the spin-off of Xiaodu, and the news of independent listing has spread again and again, but no further action has been seen. Baidu’s subsidies to Xiaodu have not stopped, and the commercialization prospects are unknown.

In recent years, with the gradual enrichment of smart home product categories, the number of substitutes for smart home entrances is increasing, the status of smart speaker entrances has been shaken, and the beautiful idea of "hardware pasted upside down, ecological money" does not know when it will be realized.

The mobile ecosystem is facing huge challenges, and autonomous driving is still burning money. How long can Baidu give Xiaodu patience, and whether Xiaodu can support Baidu’s grand AI dream expectations, everything has to be questioned.

First, the AI story awakened by a "small degree"

"Xiaodu Xiaodu, play children’s songs."

This was the "killer sentence" of a treasure mother to awaken her baby’s rhythmic cells, and it also started Xiaodu’s busy day.

Xiaodu is the name of Baidu’s artificial intelligence assistant and the wake-up word of Xiaodu speakers.

In September 2015, Xiaodu came out as a secret and made many attempts around artificial intelligence voice technology. The most widely known ones are the voice operating system Xiaodu Assistant (Duer OS), and the Xiaodu Assistant as the core. Xiaodu smart speakers and other hardware products.

In March 2018, Baidu established the Smart Life Business Group (SLG), and at the same time released the first smart speaker with a screen – Xiaodu is at home, and Robin Li personally stands on the platform.

SLG is jointly composed of the former Baidu "Du Mi Business Department", Baidu "Hardware Ecological Channel Department" and the acquired "Raven Studio Studio", and is managed by Lu Qi. After Lu Qi left, Jing Kun became the general manager of SLG.

At that time, there was no screen speaker in the Chinese market, but foreign giants had tried the water (in 2017, Amazon released a smart speaker Echo Show SLG with a screen). The combination of "screen + speaker" led by Baidu in China has brought huge increments to Baidu and virtually changed the competition landscape of the domestic smart speaker market.

The major giants followed suit. Alibaba and Tencent launched Tmall Genie CC and Tencent Jingle one after another. JD.com also released JD.com Play…

2018 was a year of Xiaodu’s anger, and it was also a year in which Jing Kun frequently exported Xiaodu’s values.

Jing Kun once said that Baidu’s original intention of making Xiaodu was to let everyone have a smart assistant, but it has no form. In the early stage of development, it needs to be attached to the form of a certain product in order to make users feel that it is an "assistant".

"On the road of smart assistants, how to meet users’ expectations for this product at this stage is very critical, which will affect whether to abandon it later or use it all the time." The form of smart speakers does not have such high requirements for the entire smart assistant, and users’ expectations are relatively limited. The initial demand may be to play music or listen to the radio, which is more in line with the expectations of the "form".

2019 Q1 financial report released, Baidu suffered its first quarterly loss in 14 years, but Xiaodu’s performance was very impressive.

This year, Xiaodu smart speaker shipments in the Chinese market first, the number of smart devices equipped with Xiaodu assistant exceeded 275 million, up 279% month-on-month, and the monthly voice interaction reached 2.37 billion, up 817% month-on-month, becoming China’s largest and most active dialogue artificial intelligence platform.

In Q2 2019, Baidu’s smart speaker shipments increased by 37 times, surpassing Google to become the world’s second largest smart speaker supplier.

Thanks to Xiaodu’s outstanding performance, Jing Kun was also promoted to vice president of the group and reported directly to Li Yanhong. At the 2019 Baidu AI Developer Conference, the "young strong faction" management led by Wang Haifeng, Shen Shuo and Jing Kun came to the forefront.

Jing Kun’s promotion also represents Baidu’s expectations for Xiaodu, or rather, the expectations of the value of voice interaction represented by Xiaodu.

Baidu has always emphasized that it is not a hardware company, and the definition of Xiaodu smart speakers is not pure hardware, but growing AI software. Hardware is just a carrier of it.

Xiaodu is Baidu’s game of chess for the intelligent voice ecosystem. According to a previous survey by Bain, voice interaction will account for about 30% of human-machine interaction in the future, and the market prospect is broad. For Baidu, which has always wanted to build itself into a "Chinese AI business card", voice technology is the soul of smart speakers, and it is also a high ground that Baidu must snatch.

Over the years, Robin Li’s investment in the voice field has been firm. From the release of the first smart speaker with screen "Xiaodu at home" to the full-duplex wake-up-free Xiaodu Assistant 5.0 update, every time Xiaodu makes an appearance, Li Yanhong stands firmly in front of the stage.

The turning point occurred after SLG was spun off.

In September 2020, Baidu announced that its smart life business group business (referred to as "Xiaodu Technology") completed the signing of an independent financing agreement, led by Baidu Capital and CPE (CITIC Industrial Capital) strategy, IDG Capital joined the investment, post-money valuation 20 billion yuan. After the financing is completed, Baidu still has absolute control over Xiaodu Technology.

The official explanation is that independent financing has a positive role in promoting the development of Xiaodu Technology and its strategic coordination with Baidu. Independent financing may become a key step for Xiaodu Technology to start the domestic listing process.

A year later, in August 2021, Xiaodu Technology completed the B round of financing, with a post-money valuation of 33 billion yuan.

Nowadays, Xiaodu Assistant has been iterated to version 7.0 and is widely used in home, hotel real estate, car, portable and other multi-scene fields. In addition to smart speakers, the landing smart hardware products have also expanded to smart fitness mirrors, smart dictionary pens, smart headphones, smart camera TVs, etc.

Xiaodu Assistant, as Xiaodu’s core AI technology capability, is already the largest conversational AI operating system in China, but the fate of Jingkun and Xiaodu Technology is increasingly uncertain.

Second, Li Yanhong’s "good heart"

Looking back on the past, we will find that Li Yanhong once had a strong preference for Xiaodu.

The same is defined as the new entrance to smart home, Xiao Ai was only briefly mentioned by Lei Jun at the end of the tenth anniversary speech, and Tmall Genie rarely appeared at Alibaba’s group-level conferences. However, Xiaodu was able to stand at the C position at Baidu’s most important AI developer conference for many years.

In Li Yanhong’s opinion, Baidu’s core business search itself is a very typical AI problem, because the essence of search is to give answers to questions, and natural language processing technology in AI technology can help search more accurately.

In the era of AI-enabled search, search results are no longer sorted by the relevance of web pages, but are digested and processed based on the machine’s full understanding of the question.

From 1.0 to 7.0, Xiaodu Assistant has transformed into an "open API" that integrates Baidu AI voice technology and image technology, and can empower third parties. Robin Li regards it as the interaction hub, super entrance and operating system of the AI era.

Jing Kun once used "Xiaodu is the search and information flow of Baidu in the future, it is a natural search itself in the family" to illustrate the importance of Xiaodu to Baidu.

Xiaodu is also an entry point for Baidu to leverage smart life. Users’ demand for smart devices continues to diverge, and once the demand diverges, the application of technology will come out.

The ideal is full, but the reality is very skinny. In recent years, the popularity of smart homes has been accelerating, and now smart homes will have built-in smart assistants to interact directly with users. The role of smart speakers as the central control of smart homes has begun to weaken, and Xiaodu’s status as the "interaction center" has also begun to shake.

At the same time as the product iteration, Xiaodu did not forget to start commercialization attempts. At the Xiaodu media communication meeting in early July 20, Jing Kun first mentioned the concept of "breaking the circle". The corresponding action is that Xiaodu connected to Kuaishou, Douyin, Bilibili, Youku and other content partners in one go in the first half of the year, and launched Xiaodu education smart screen, Xiaodu smart early education machine and other products for children.

After the double reduction, the commercial water test effect for C-end users is not good.

Xiaomi’s smart hardware ecosystem has been wildly sought after by capital, and Xiaodu also wants to follow suit, saying that it wants to create an open AIoT interactive ecosystem based on Xiaodu Assistant’s core technology, and form a set of "1 + X + Y" ecological development ideas, 1 refers to Xiaodu Assistant (Duer OS); X refers to interactive hardware, including Xiaodu’s own products and cooperative products; Y refers to ecological hardware, smart home and IoT products.

However, it turned out that the story of the AIoT interaction ecosystem was even more difficult to tell, regardless of the intense competition in AIoT. In addition to Baidu’s Xiaodu Assistant, players in the voice interaction system also included Amazon’s Alexa and Ali’s AliGenie. Huawei, a communications equipment manufacturer, was building the Hongmeng system, and Haier, a home appliance giant, had also launched its own smart home system, U-home, very early on…

Each family has an ambition to build an "ecology", and the so-called ecology is not "open". Everyone is developing their own "ecological standards". Openness is "internal", and barriers are always external.

The fragmentation of the ecological level has led to an increase in the interconnection barriers of end point devices between different brands, and the clustering of different camps has further intensified the "data islands" of each brand’s ecology. These are all insurmountable mountains in front of Xiaodu.

I still remember when "Xiaodu Home" with Xiaodu Assistant was launched, Robin Li said excitedly, "Xiaodu Home is a new product form spawned by AI, a revolutionary new species of AI in the Chinese market, and will also open up new application scenarios."

Nowadays, revolutionary application scenarios have not emerged, and the focus of Li Yanhong’s AI narrative has also been placed on digital humans and autonomous driving.

In July this year, the Baidu World Congress was held as scheduled, and the opening was no longer the familiar conversation between Robin Li and Xiaodu, but the digital human Xiaoxiao.

Baidu has confidence in AI, but it doesn’t have much patience for Xiaodu, who is already independent. One detail is that Jing Kun’s title has not changed except for the newly added CEO of Xiaodu Technology.

Lining or face

Xiaodu, once popular with Baidu and the market, has become much like a "battle with a scarecrow": it can defeat its opponent in theory, but it can only defeat its opponent in theory.

"Voice interaction is the mainstream, it’s not extra and add-on", different perspectives bring business changes, just like when touching the screen, many people think what is the purpose of adding a touch screen to the mobile phone, but Apple sees the value of the touch screen.

For Baidu, the value of voice interaction represented by Xiaodu is a very low-level value that can revolutionize many things, such as the interaction frequency between users and devices, the interaction time between users and a smart device, and even the content and information service ecosystem after the entire device.

This is a particularly low-level revolution, and it is an irreversible revolution in history, and it will only happen faster. This is why Baidu really values voice interaction, because this low-level revolution will upend so many things and is a huge opportunity.

In an interview in 2019, Jing Kun has always emphasized that the commercialization of Xiaodu is not the focus, the most important AI capabilities and user experience, hoping that the product can reach a larger scale, but "the current scale is not that large, so it needs a golden development period".

Jing Kun also mentioned more than once in public that Xiaodu’s business model does not make money on hardware, and truly attaches importance to content and service ecology, focusing on interaction, such as multi-mode interaction, remote interaction, including voice interaction self-learning system, etc., will also have a lot of technical investment in related aspects.

Three years had passed, and smart speakers had already passed the initial stage of user education. According to the official statement, Xiaodu was now the largest in the country, and the technology was constantly being updated. Commercialization should also make a big breakthrough, but from the actual results, Xiaodu commercialization seemed to be still in the "embryonic" stage.

The performance of the Xiaodu brand has been repeatedly mentioned in Baidu’s financial reports, but the specific revenue of Xiaodu has not been disclosed.

It seems that Xiaodu is more like the "face" that highlights Baidu’s AI technical strength, rather than the "lining" that supports Baidu’s performance growth curve.

No matter how beautiful the business story is, it is just a "mirage" that cannot run through the closed loop of business. The fact is that Xiaodu cannot make money by relying on hardware products, and it is more difficult to make money by building an ecosystem through AI voice interaction.

No capital can continue to pay for an idea that cannot see returns. If Xiaodu cannot go public independently and raise money in the future, it will only be a matter of time before Baidu completely abandons it.