Two data points stood out particularly in Baidu's recently released Q2 earnings report. First, in July, AI-generated content accounted for 64% of mobile search result pages, with over 60% of search result pages on Baidu App displaying rich media content in the top position. Second, Baidu's AI new business revenue broke through 10 billion yuan for the first time, with intelligent cloud business revenue growing 27% year-over-year, and Apollo Go completing 14 million ride services in a single quarter (as of August this year), maintaining its global leadership position.

Behind these numbers lies Baidu's relatively aggressive strategic path in the AI era.

Amid the global AI giants' full-speed competition and the narrative that "AI is killing traditional internet search," the search businesses of Alphabet and Baidu represent highly valuable comparative cases for observation. What we can see now is that the former, backed by the world's largest advertising system and market share, is introducing AI Overviews in a moderate manner, while the latter, driven by both advertising revenue pressure and corporate strategic transformation, has chosen the path of first redesigning, then exploring commercialization.

Just last month, Baidu Search underwent its largest reform in ten years, upgrading the "search box" to a "smart box," replacing hyperlinks with multimodal, intelligent agent content. In other words, Baidu chose a more direct and thorough reform path.

To a large extent, the stakes of this gamble lie in whether Baidu's performance can achieve a new round of growth when search business is fully driven by AI.

**Steady Growth of AI New Business, Foundation for Baidu's Reform**

In Baidu's performance, what deserves the most market attention remains search. Data shows that Baidu is continuously accelerating search transformation, with intelligent, structured, and multimodal-priority AI answers gradually replacing static hyperlinks. In July, the proportion of AI-generated content in mobile search result pages rapidly increased from 35% in April to 64%; on Baidu App, over 60% of search result pages displayed rich media content in the top position.

This figure is not only far higher than Alphabet's disclosed 13.14% AI trigger rate in March this year, but also means that Chinese internet users are already deeply experiencing the new ways AI search brings in their daily use.

In July this year, Baidu launched its largest-scale search business redesign in nearly ten years, not just "adding an AI answer to result pages," but systematic transformation. Baidu Search has replaced traditional hyperlinks with multimodal content and AI intelligent agents. Meanwhile, in terms of interaction logic, users within Baidu's product matrix can not only obtain AI-generated answers but also directly invoke plugins and services to complete one-stop operations. Additionally, across the product's various formats including images, text, short videos, and real-time search, AI has become the default entry point.

Beyond the "aggressive" search business, another major highlight of Baidu's Q2 financial report is the scaling of AI new business revenue.

In terms of AI computing power, the financial report shows that on the evening of August 20, during Baidu's Q2 earnings call, it was disclosed that intelligent cloud business revenue grew 27% year-over-year, with Non-GAAP operating profit achieving year-over-year growth. "The unique end-to-end four-layer AI architecture has become our core competitive advantage," said Baidu founder Robin Li, stating that full-stack AI capabilities are driving healthy growth in the intelligent cloud business.

Recently, IDC released the "China AI Public Cloud Service Market Share, 2024" report, with Baidu Intelligent Cloud ranking first for six consecutive years. In the latest IDC report released on August 20, Baidu Intelligent Cloud achieved first place in the large model platform market.

In Q2, Baidu successfully activated a 30,000-card Kunlun chip cluster, a chip truly designed for large models with excellent performance, capable of simultaneously supporting full training of multiple hundred-billion-parameter large models and supporting 1,000 enterprises in fine-tuning ten-billion-parameter large models simultaneously.

On the evening of August 20, during the Q2 earnings call, Robin Li revealed that Apollo Go's global ride services exceeded 2.2 million times in Q2, growing 148% year-over-year. During the quarter, Apollo Go successively reached cooperation agreements with Uber and Lyft, two major global ride-sharing platforms, to deploy autonomous vehicles in Asian, Middle Eastern, and European markets.

In Q2, businesses represented by Baidu's new generation digital humans released by DuBo, GenFlow 2.0 - the world's first full-end universal intelligent agent released by Baidu Library and Baidu NetDisk, and AI tools like Comate AI IDE also showed considerable growth. Revenue generated by digital humans increased 55% quarter-over-quarter, approximately 500 million yuan according to calculations based on information disclosed in the earnings call.

In recent quarters, the usage rate of digital human technology has gradually increased, being applied by more and more merchants in industries such as e-commerce, legal services, education, automotive, and healthcare.

The above achievements indicate that Baidu's AI strategy has formed growth pillars not only in search business but across multiple business lines. Particularly noteworthy is Apollo Go's rapid expansion in global markets, which is an important showcase of Baidu's "full-stack AI capabilities," with its breakthrough in international progress deserving special attention. Additionally, Apollo Go holds a leading position in the global right-hand drive Robotaxi market.

In Hong Kong, Apollo Go achieved three expansions in half a year, with test routes now extended to Tung Chung and Southern District, covering more complex commercial and residential areas. Leveraging its experience in Hong Kong's right-hand drive, left-side traffic environment, Apollo Go is expected to accelerate its globalization pace.

Citi believes that the partnerships with Uber and Lyft validate Baidu's global recognition in the autonomous taxi service field, having positive significance for Apollo Go's further development of new markets with commercialization potential, highlighting Baidu's core capabilities in the autonomous driving value chain.

**Aggressive Search Reform: A Certain Winning Formula?**

Over the past decade, search engines have almost been considered a mature, competitive track.

With new engines for the war machine, they dare to dismantle old carriages.

Globally, the AI transformation of search engines has become an irreversible trend. Alphabet has proven that traditional search will maintain its leading advantage in the AI era. In March this year, before officially launching AI Mode to the public, 13.14% of Alphabet's queries triggered AI Overviews; Alphabet's latest financial report released in July shows that AI Overviews currently has over 2 billion monthly active users across more than 200 countries and regions worldwide.

The reason for describing Baidu's search business reform as "aggressive" is that Baidu has hardly left any "retreat" space for its search business, directly pushing AI-driven experiences and results to users as core experiences in the most direct way.

Alphabet's AI Overviews are essentially still based on hyperlinks, merely adding AI-generated overview content above traditional search results. The difference in AI search coverage rates between Alphabet and Baidu further confirms the aggressive stance of the latter - replacing hyperlinks with multimodal, intelligent agent content - while AI Overviews are actually still hyperlinks, Baidu has taken a step further.

To a large extent, when promoting AI search revolution, Baidu doesn't need to be constrained by the pressure of protecting advertising revenue like Alphabet. Currently, it appears that Alphabet's deployment of AI Overviews has been constrained by its business model.

Why is Baidu making such a big bet on search? Does this aggressive reform truly have a certain winning formula behind it?

Baidu's answer is built on three pillars: growth and confidence brought by AI new business; Alphabet has verified that search will still maintain its leading advantage in the AI era, while Baidu chose a more aggressive evolution path; and Baidu directly skipped the "AI Overviews" stage, reconstructing search experience with multimodal and intelligent agents.

These three factors together support a conclusion: aggressive search reform is not putting all eggs in one basket, but Baidu seeking a new, certain growth curve.

The Q2 report reflects why Baidu made such choices. Baidu already has AI new business revenue support sources - the continued growth of cloud computing and autonomous driving enables it to consistently deliver solid performance in financial reports. The operating profit margin of Baidu Intelligent Cloud and the scale effects of Apollo Go's rapid expansion in global markets together constitute the foundation for Baidu's search transformation.

Alphabet's experience has proven that AI products often don't emphasize revenue in early stages but focus on user penetration and usage time, with capital markets having high tolerance for this. Baidu hasn't rushed to forcibly bind AI-generated results with advertising but chose to first polish the experience, then gradually explore new commercialization models.

Regarding the construction of the overall search ecosystem (including third-party platforms), Baidu Search Open Platform has integrated over 18,000 MCP services and more than 220 AI applications, covering multiple scenarios including lifestyle services and financial transactions.

Baidu AI search's aggressive redesign is more like a long-term strategic gamble. Traditional search advertising has slowed in growth, with a very mature business model. Baidu's new AI business is growing rapidly, giving search reform more space - it doesn't need to rush to directly monetize AI search but can first strongly position for the future, making product formats completely align with new generation user needs.

Baidu's choice is actually a high-risk strategy that may yield high returns in the future. If users accept and form habits, Baidu will reconstruct new search growth points in the AI era. But if user experience is poor, or if the market is subsequently led by new product formats, Baidu will also face greater pressure.

**Full-Stack AI: The Ultimate Battlefield for Tech Giants**

In fact, the AI strategies of Alphabet and Baidu reflect competition between two AI routes of global tech companies.

Alphabet takes gradual improvement, advancing AI while protecting existing business models, while Baidu chooses aggressive reconstruction for the possibility of generational leaps. The former has controllable risks but may miss the opportunity window of paradigm shifts, while the latter involves high risk-high return and faces unknown market acceptance tests.

Beyond disagreements, the consensus among tech giants is that owning only chatbots cannot support the future path to AGI. Real-world problems including user retention, customer acquisition costs, traffic value, and business models will constrain and then force AI giants to turn toward building full-stack AI ecosystems.

From underlying computing power, foundation models, development platforms to upper-layer industry applications, Baidu covers the complete closed loop of AI industrialization. This strategy is reflected not only in technical layout but also verified its comprehensive advantages through industrial implementation.

Baidu has built a four-layer technical architecture from underlying chips (Kunlun), frameworks (PaddlePaddle), models (ERNIE), to applications (Apollo Go). The core value of this layout lies in balancing short-term benefits with long-term exploration. The stable cash flow contributed by Baidu Intelligent Cloud supports its investment in cutting-edge fields such as digital humans, GenFlow, and MCP.

Baidu's Q2 financial performance reflects industry and market recognition of its full-stack AI capabilities, while supporting overall financial performance exceeding expectations, which is precisely the cost-effectiveness and competitiveness brought by full-stack AI.

This, together with aggressive search reform, constitutes Baidu's winning formula in the AI era. If AI and search must have a battle, then Baidu chose to make the first move. If AI's certainty has been verified, then the earlier and more thoroughly search is reconstructed, the more likely it is to establish a new pattern in advance.

When search boxes finally disappear, the outcome of this competition lies in who first defines what users will actually need tomorrow.

Baidu's AI strategy is not fragmented. Whether AI search or intelligent agent ecosystem, they will ultimately flow back to the basic needs of the commercial world, integrating into the super entrance of AI, tools, and services.

The advantage of full-stack AI layout is having both current certain commercial revenue and giving Baidu the confidence to explore a more innovative future.