Since the beginning of this year, the "industry whirlwind" triggered by AIGC has continued to sweep across all walks of life, bringing a warm current to the long-dreary Internet track. For players who bet heavily on AI, this AI whirlwind is even more like "a long drought meets rain", and finally ushered in a day when "the willows are dark and the flowers are bright".
As a top player betting heavily on the AI track, Baidu's joy at the arrival of the AI large model is beyond words, even to the point of "talking about AIGC", the continuous improvement of operating efficiency also highlights the potential of AI from the side value.
Continue to strengthen operating efficiency
Recently, Baidu released its unaudited financial report for the second quarter ending June 30, 2023. In the second quarter, Baidu achieved revenue of 34.1 billion yuan, a year-on-year increase of 15%; non-GAAP net profit attributable to Baidu reached 8 billion yuan, a year-on-year increase of 44%; both revenue and profit increased significantly, exceeded market expectations.
Specifically, Baidu's revenue growth is driven by both traditional online marketing revenue and non-marketing revenue. The financial report shows that in the second quarter of this year, Baidu’s core online marketing revenue was 19.6 billion yuan, a year-on-year increase of 15%; non-online marketing revenue was 6.8 billion yuan, a year-on-year increase of 12%. In operation, Baidu Core generated a net cash flow of about 9.7 billion yuan. Among them, there are two businesses worthy of focus: one is the overall profit of iQiyi; the other is the profit of the smart cloud business.
Let’s look at iQiyi first. Since announcing a series of cost-cutting and efficiency-enhancing measures last year, iQiyi has achieved positive cash flow and net profit for several quarters. Public data shows that from 2022Q4 to 2023Q2, iQiyi has achieved positive net profits for three consecutive quarters, reaching 304 million yuan, 618 million yuan, and 365 million yuan respectively, basically maintaining a continuous profit situation.
In the case of positive net profit, iQIYI’s operating cash flow reached 886 million yuan, achieving positive for five consecutive quarters; free cash flow was 872 million yuan, positive for four consecutive quarters, and the total amount of interest-bearing liabilities dropped sharply , and gradually entered the stage of high-quality development. In addition to the continuous and steady growth of membership business, its online advertising service revenue has also recovered strongly. This quarter, iQIYI’s membership advertising business increased by 25% year-on-year.
Looking at the smart cloud, Baidu Smart Cloud achieved profitability again in Q2, which means that Baidu has gradually entered a stage of a virtuous circle in the cloud business. Among them, in Q2, the number of contracted customers of Baidu Smart Cloud continued to increase, including external enterprises such as Industrial Bank, China Southern Net, Hande Information, Kingdee, and iSoftStone, all of which have become customers of its smart cloud business. In fact, not only Smart Cloud, but also Baidu’s unmanned taxi Carrot Run, has also achieved continuous growth in operating data and revenue, and the commercialization of AI has continued to accelerate.
In addition, as an important direction of Baidu's AI layout, the AI large model has been emphatically emphasized by Baidu's top management, and it occupies a "heavy stroke" in the financial report.
AI large model becomes a new starting point
As an important starting point of Baidu AI, the AI large model was unanimously affirmed by Baidu's top management at this financial report meeting. From a business perspective, the AI large model is having a profound impact on Baidu Smart Cloud, search advertising, hardware and other businesses.
First of all, Baidu's search business has been reconstructed by the AI model, and the click retention rate of Baidu users has been significantly improved compared with the past. The arrival of the AI large model has turned the interaction between the search engine and the user into a two-way interactive relationship, and with the growth of the data volume of the AI large model itself, the professional question-answering ability it can provide is getting closer and closer to that of an experienced People, which reconstructs the user's interactive experience from the bottom up.
Baidu also takes advantage of this to increase user retention. Robin Li, CEO of Baidu, said at the financial report, "After using the Baidu search service with Wenxinyiyan function, users will ask more questions that were rarely asked on the platform before, and there will be multiple rounds of dialogue. This is very different from the behavior of users who use traditional search services. From the perspective of click-through rate, there is a significant increase, which can be converted into retention rate later.”
According to the general law of app retention rate increase, with the increase of app retention, the advertising value of Baidu's main APP will also be improved, which is obviously an obvious benefit for Baidu's long-term commercial value improvement. In addition to retention, Baidu’s Baidu Library, Tieba, etc. are also affected by the AI big model, and continue to upgrade to "AI".
Secondly, Baidu's smart hardware is empowered by the AI model, and its differentiation advantage has been strengthened. For example, Xiaodu Qinghe, a learning mobile phone launched by Baidu based on the advantages of its own AI large model, was released in May this year. Its main advantages are officially based on the differentiated hardware advantages of Baidu's AI smart model. Specifically, Xiaodu Qinghe has more comprehensive functions at the horizontal level, integrating the core functions of "mobile phone + dictionary pen + translator + hearing treasure + word cards"; vertically, Du Qinghe can answer questions and give lectures, Being able to practice spoken English as a companion, and correcting Chinese and English compositions sentence by sentence, basically covers the basic needs of young people's extracurricular independent learning. Regardless of the aspect, it is a differentiated category for the current learning machine market.
Finally, it is to fully empower the intelligent cloud business and drive the continuous growth of cloud services. As Li Yanhong said before, with the continuous development of large models, MaaS (model as a service) will become an important consideration for customers to choose cloud vendors, which can also be reflected in Baidu's financial report. For example, many of the cooperation between Baidu and Kingdee and other industry customers are about promoting the implementation of large-scale industry models, such as the "Kingdee Cloud. Sky GPT" jointly launched by Baidu and Kingdee. business again.
Intelligent driving two-wheel drive
In addition to the large model, as Baidu's business indicators continued to improve in the second quarter, Baidu's progress in the robotaxi business has also achieved gratifying results. At the same time, the news of Baidu and Geely's cooperation in car manufacturing has also made new progress. Obviously, Baidu's "two-wheel drive" strategy in the commercialization of intelligent driving is gradually taking shape.
According to financial report data, in the second quarter of 2023, Baidu's L4 autonomous driving business performed equally well, with the number of ride service orders exceeding 700,000, a year-on-year increase of nearly 1.5 times. According to the data disclosed by Baidu, the number of unmanned driving reservations for Carrot Kuaipao has risen from 196,000 in 2022Q1 to 714,000 in 2023Q2, and the expansion speed is obvious.
In addition, among the new cars announced by the Ministry of Industry and Information Technology, Baidu's cooperation with Geely's Jiyue 01 has officially entered the stage, announcing that Baidu's "car manufacturing" has entered a new stage. This also indicates that Baidu's commercialization of intelligent driving is further accelerating.
On the one hand, with the expansion of the operating area of Carrot Kuaipao, the increase of fleet size, and the improvement of operating efficiency, Baidu's bicycle cost per kilometer has continued to decrease. According to the current public data, the operating area of Baidu’s Carrot Kuaipao has covered four cities: Wuhan, Chongqing, Beijing and Shenzhen. In July this year, Baidu won the first batch of licenses issued by Shanghai Pudong New Area, allowing unmanned driving tests to be carried out on public roads in different areas. Currently, this progress is still expanding and is expected to expand to more urban scenarios. The total service volume and operating efficiency of Robotaxi have been significantly improved, which fully demonstrates that the balance of commercialization of autonomous driving can be expected in the future.
On the other hand, through its deep participation in automotive hardware manufacturing, it will accelerate the commercialization of Baidu's intelligent driving. As we all know, the reason why intelligent driving has been difficult to implement for a long time in the past is that the road test data of autonomous driving is too large, the cycle is too long, and it faces huge R&D investment and cash flow pressure. Solve the safety issues of car travel (such as Google's Waymo and General Motors' Cruise, which have not yet liberalized their business operations on a large scale).
In contrast, Tesla, which promotes the evolution of intelligent driving through the integration of software and hardware, has evolved from the original assisted driving mode to the "smart driving mode" and realized commercial operation. Obviously, for Baidu, while imitating Google, relying on hardware and software to promote the commercialization of intelligent driving and realize "software to make money" is more attractive to it.
Judging from the multiple intensive industrial and commercial changes of Baidu's Jidu Automobile, the game between Baidu and Geely in "car manufacturing" has come to an end for the time being. From the perspective of Baidu's overall smart driving strategy, the launch of Jiyue 01 and the expansion of Baidu's Robotaxi service show that Baidu's two-wheel drive in terms of commercialization is gradually taking shape.
Baidu AI "harvest" season is here?
Judging from the current situation, Baidu's current investment in AI is gradually entering the late stage of commercialization. So, is Baidu AI's "harvest" quarter coming? It also needs to be analyzed from multiple aspects.
First of all, from the perspective of AI large models, the combination of Baidu's large models and cloud service technology is opening up commercial space. Judging from public information such as financial reports, Baidu's current AI large-scale models are blooming in more places, forming an all-round support for Baidu's AI products, which indicates that the commercialization of Baidu's AI large-scale models is slowly opening up.
However, since Baidu has not announced traditional businesses such as Baidu search, and the data changes between the introduction of the AI model and the previous data, it is difficult for the outside world to judge the extent of the impact of its AI model and whether it can be sustained. After all, among the currently public large models, the active users and search volume of the core large models represented by ChatGPT have dropped a lot compared with early March. Obviously, its attractiveness to users is declining, and domestic AI large models may also will face this situation.
In contrast, the profitable intelligent cloud business is expected to continue the current situation. The financial report shows that Baidu's smart cloud business has achieved continuous profitability for two consecutive quarters, which means that after a long period of investment, the commercialization of Baidu Smart Cloud is gradually on the right track, and it is expected to continue to make profits in the future , or will become the earliest field of Baidu AI to achieve successful commercialization.
Secondly, in the field of intelligent driving and intelligent hardware, Baidu's AI commercialization still has a long way to go. In the field of intelligent driving, although the Robotaxi service has been launched, its current operation scale is not large enough and is still very limited. Baidu’s car has just come out, and it is hard to say how its sales will be in the “very heavy” new energy vehicle market. Whether commercialization can be rapidly expanded also needs time to verify. As far as smart hardware is concerned, the current competition in the industry is also very fierce, and whether the products supported by its AI large-scale models can be released depends on the follow-up market performance.
Overall, Baidu's AI business has passed the "most difficult" stage in its development to this day. At present, the commercialization of AI has shown a gratifying momentum, but it is still far from a full "bundle harvest".