Capturing Business Essence, Focusing on Corporate Core Competencies

Author | Yang Cheng

Baidu (NASDAQ: BIDU / 09888.HK) is grappling with a simultaneous slowdown in both its technological and business fronts.

The Q2 financial results, unveiled on August 18th, revealed that Baidu's revenue has been on a year-on-year decline for five consecutive quarters, with net profit plummeting by over 68%. Meanwhile, its advertising business, once a pivotal growth driver, continues to bleed, with the Baidu App experiencing a loss of 91 million monthly active users over the past year.

Even more disheartening is the management's acknowledgment of aiming to 'reclaim top-tier status in foundational large models,' implicitly admitting that the ERNIE Large Model, which initially held an early advantage, has lagged in terms of user adoption and ecosystem positioning.

01

Five Consecutive Quarters of Revenue Decline

In terms of total revenue, Baidu's Q2 earnings stood at 31.325 billion yuan, marking a 4% year-on-year decrease and a 2% quarter-on-quarter drop, falling short of the market's anticipated 31.59 billion yuan. Net profit tumbled to just 2.319 billion yuan, a staggering 68.33% decline from the previous year. More significantly, this trend reflects prolonged growth challenges rather than a one-time fluctuation.

Since Q2 2025, the tech behemoth's revenue has been on a downward trajectory for five consecutive quarters, primarily due to the erosion of its traditional advertising base.

In Q2, Baidu's core general business revenue reached 25.2 billion yuan, a 4% year-on-year decrease. Revenue from traditional online marketing (advertising) services, historically its primary profit generator, plummeted to just 13.1 billion yuan, a 19.1% year-on-year drop, with its contribution to general business revenue shrinking from 62% to 52% year-on-year.

The root cause of the core business's decline lies in the erosion of its user base. According to Baidu's financial report, the Baidu App's monthly active users decreased from 735 million in June 2025 to 644 million in June 2026, a 12.38% year-on-year decline, resulting in a net loss of 91 million users within a year.

Luo Rong, General Manager of Baidu's Mobile Ecosystem Group, explained during the earnings call that the emergence of new products such as AI chatbots has altered user information acquisition habits. Baidu has proactively pursued AI search transformation and temporarily slowed commercialization efforts, which has had a short-term impact on advertising revenue.

In contrast to the weakening advertising sector, the proportion of new AI businesses has risen. In Q2, Baidu's core AI new business revenue reached 12.5 billion yuan, a 25% year-on-year increase, accounting for 50% of general business revenue. Among this, revenue from intelligent cloud infrastructure surged to 7.3 billion yuan, a 50% year-on-year increase (with GPU cloud revenue skyrocketing by 283% year-on-year).

However, revenue from AI applications, including Baidu Wenku and Baidu Netdisk, amounted to just 2.5 billion yuan, a mere 3% year-on-year increase, with no quarter-on-quarter growth. AI-native marketing services generated 2.6 billion yuan, remaining flat year-on-year.

This indicates that Baidu's growth is heavily reliant on capital-intensive computing power leasing and GPU cloud services. In high-value-added, high-margin software ecosystems and AI-native applications, commercialization has yet to achieve significant breakthroughs.

Moreover, the rapid expansion of capital expenditures has further strained Baidu's cash flow. Capital expenditures for the quarter soared to 11.4 billion yuan, a nearly 200% year-on-year increase, while operating cash flow stood at just 3.4 billion yuan.

On the cost front, Baidu has adopted a highly defensive stance. In Q2, selling and administrative expenses decreased by 23% year-on-year to 4.6 billion yuan. Research and development expenses also declined by 10% year-on-year to 4.6 billion yuan. Only selling costs increased by 4% to 19.1 billion yuan due to cloud business growth.

While the robust expansion of AI cloud and GPU computing power businesses has boosted revenue share, it has also brought about high server depreciation and infrastructure costs. As Baidu's AI commercialization efforts have yet to yield substantial revenue, intensifying competition is rapidly eroding its leading position in China's AI sector.

02

ERNIE Lags Behind: An Unenviable Position in the Foundational Model Race

If the financial report merely reflects operational hurdles, the decline in the competitiveness of the ERNIE Large Model strikes at the core of Baidu's strategic vision. As the first major player in China to heavily invest in foundational large models, ERNIE now faces the risk of being marginalized in the increasingly fierce model elimination race of 2026.

Data from third-party market research firm QuestMobile in June 2026 vividly illustrates shifts in the domestic C-end AI access landscape. Doubao leads with 382 million monthly active users, followed by Qianwen and DeepSeek with 167 million and 130 million, respectively. In stark contrast, the once early-mover ERNIE Bot's standalone app has seen its monthly active users dwindle to less than 5 million.

Beyond user loss, ERNIE finds itself in a passive position within the commercial and technological ecosystem.

Today's domestic large model competition has evolved from a focus on pure model capabilities to scenario positioning. DeepSeek has rapidly eroded the developer ecosystem with its open-source strategy and cost-effectiveness advantages. Tongyi Qianwen, leveraging Alibaba Cloud's infrastructure, has constructed a closed-loop service system for the enterprise market. Kimi has established barriers in high-frequency text scenarios such as academia and law with its 2 million-word ultra-long context capability.

During the analyst Q&A session following the Q2 earnings release, Baidu management stated, 'We were among the first in China to invest in foundational large models. Along the way, we've encountered trials and errors, but our determination to make ERNIE competitive remains unshaken. Looking ahead, we will continue to invest the necessary resources to drive ERNIE's ongoing development. As part of this effort, we have further optimized our organizational structure and recently recruited top AI talent. We are confident in accelerating AI iteration and reclaiming ERNIE's top-tier status in foundational large models.'

The very mention of 'reclaiming top-tier status' confirms that ERNIE is no longer in the top tier. With the core advertising business weakening, AI application monetization stalling, and large model advantages narrowing, this dual slowdown in technology and business implementation carries far more profound implications than single-quarter financial fluctuations.

Disclaimer

Caiguan Chaotou hereby declares that all information and content provided are based on objectively compiled and analyzed publicly available data. However, we cannot guarantee the absolute accuracy, completeness, or timeliness of this information.

Any opinions, analyses, or conclusions contained herein represent the independent views of Caiguan Chaotou at a specific point in time and do not constitute any investment advice or strategy. Financial markets entail risks, and investors assume full responsibility for any decisions made based on this content. If you have any questions about our content or discover any omissions, please contact us directly through our official backend channels, and we will promptly verify and address them.