Li Yanhong is sprinting on the road of large-scale models.

Author | Zhou Zhiyu

Editor | Zhang Xiaoling

At the end of last year's internal meeting, Baidu's founder, Robin Li, expressed his hope to create a "1" by 2023, meaning at least one high-growth, innovative business.

For the past decade, Baidu has been telling the story of artificial intelligence, but when it comes to revenue, investors have been repeatedly hopeful and disappointed.

However, with the unexpected success of large-scale models in 2023, Baidu took the lead in launching Wenxin Yiyu, becoming the pioneer of large-scale models among domestic technology companies, and it has the potential to become the "1" that Robin Li has been expecting.

Baidu seems to be entering its "singularity moment."

Looking at the financial report, Baidu performed well in the second quarter, with core revenue reaching 26.4 billion yuan, a year-on-year increase of 14%; core operating net profit (non-GAAP) reached 6.51 billion yuan, a year-on-year increase of 27%, both exceeding market expectations.

This is mainly due to the recovery of Baidu's advertising business. As a core business, online marketing services generated approximately 19.6 billion yuan in revenue in the second quarter, a year-on-year increase of 15%. As an industry that follows the economic cycle, Baidu benefited from the recovery of various consumer sectors in China, including tourism and local services.

On the other hand, revenue from cloud and other AI businesses reached approximately 6.8 billion yuan, with a growth rate of only 12%, lower than the overall revenue growth rate. Shen Dou, President of Baidu's Intelligent Cloud Business Group, explained that this is mainly due to the drag from the "to G" project, while industries such as manufacturing and internet services continue to grow steadily.

However, the cloud business maintained its profitability in the second quarter. Robin Li revealed that Baidu's Intelligent Cloud achieved revenue of 4.5 billion yuan in the second quarter of 2023, a year-on-year increase of 5%.

The second-quarter operating performance is not the focus. Investors and the market are still concerned about how Baidu's generative AI and large-scale models will change the company and enable it to monetize its decade-long investment of over 100 billion yuan in the field of AI.

In the past, investors believed that Baidu would be the first to achieve large-scale commercialization in the field of intelligent driving and intelligent cabins. Now it seems that the progress of large-scale models may be even faster.

After the financial report was released, Robin Li talked extensively about the innovation brought by generative AI and large-scale models to various industries, as well as how Baidu will benefit from them during a conference call.

The improved financial indicators, coupled with Robin Li's grand blueprint for restructuring Baidu, have once again energized investors.

On the evening of August 22nd, Baidu's stock rose by 2.75% in the US market, and the next day it rose by 4.39% in the Hong Kong market, leading the Hang Seng Technology Index.

Institutions including Morgan Stanley and Daiwa Securities have focused their attention on Baidu's future performance growth in large-scale models. Daiwa Securities explicitly stated that the commercialization approval of Wenxin Yiyu will be the main catalyst for Baidu's future stock price.

Robin Li is well aware of this. During the financial report meeting, he emphasized multiple times that the development of AI technology has brought about a transformation in Baidu's business, and Baidu is using the Wenxin large-scale model and Wenxin Yiyu to build AI-native applications and completely restructure its product line.

Baidu's CFO, Luo Rong, believes that with the support of the Wenxin large-scale model, advertising will become more customized, and the return on investment for advertising clients will also increase. Generative AI and large-scale models will empower various industries and improve work efficiency. At the beginning of the birth of large-scale models, people from all walks of life hoped that it could further improve productivity and hailed it as the "Fourth Industrial Revolution." Now, Baidu has gradually explored some paths.

The difficulty lies in the fact that large-scale models are still in the early stages of development, and investors' valuations and pricing of players in the field are based on imagination and anticipation. The surge in stock prices of leading players, including Microsoft and NVIDIA, is also based on extremely high expectations for the future.

However, everyone is still exploring business models and profit models.

Although Baidu launched Wenxin Yiyu early on, Li Yanhong has repeatedly emphasized the commercialization path of Wenxin Yiyu, but there is no clear example of commercialization.

The market's concern is whether Baidu is "all in" on large-scale models due to the trend, or whether it has accumulated technology. After all, in the past decade, Baidu's focus in the field of artificial intelligence has always been on autonomous driving.

And now, Baidu hardly mentions its autonomous driving business anymore. Multiple autonomous driving departments and teams within Baidu have been downsized and lost to various new energy vehicle companies or autonomous driving companies, leaving Baidu with only the name of an industry autonomous driving academy.

Jidu Auto, in which Baidu and Geely have equal shares, has also changed to be controlled by Geely and renamed as Ji Yue. Baidu has stepped back and become a provider of intelligent driving solutions behind the scenes.

Li Yanhong regards generative AI and large-scale models as Baidu's new growth engines, hoping to promote Baidu's long-term growth through them.

Baidu's management also acknowledges that the business model of large-scale models is still in a very early stage. Shen Dou, President of Baidu Intelligent Cloud Business Group, said that it will take some time to see significant revenue contributions, but generative AI and large-scale models have opened up a new growth space for Baidu's cloud business.

As the "fourth cloud" behind Alibaba, Huawei, and Tencent in terms of market share, Baidu is also being pursued by players such as ByteDance. These players are betting on large-scale models through MaaS (Model as a Service) in the early stages of developing the MaaS ecosystem. Naturally, the capabilities of the cloud become the key to determining market share during this period. Whether Baidu can stimulate the growth of its cloud business through large-scale models and reflect it in actual business revenue remains to be seen.

Whether it is in the field of AI or autonomous driving, Baidu always takes the lead in technology, but when it comes to implementation and commercialization, it always seems unsatisfactory. Now, in the era of large-scale models, Li Yanhong and his Baidu need to accelerate their pace.

With ten years of technological reserves, can Baidu maintain its first-mover advantage and make some highlights in the cloud and application side, leading the way in commercialization? Can Li Yanhong also turn the tide and regain the glory of the AI era?