Baidu, Inc. (BIDU)

Baidu is the search engine that dominates China, the world’s second-largest internet market by users and revenue. Founded in 2000 by Robin Li and Eric Xu, the company built its empire on Chinese-language search — a market no Western company knew how to navigate — and has spent the past decade reshaping itself as an artificial-intelligence and cloud-services business. The company trades on the NASDAQ under the ticker BIDU and remains a bellwether for the health of China’s internet economy and the country’s ambitions in AI.

For most of its history, Baidu looked a lot like Google’s domestic business: it made money almost entirely from advertising placed alongside search results. Users searched for products, services, and information; Baidu ranked results and sold placements to advertisers who wanted their ads to appear when people searched for relevant terms. This business generated enormous cash flow for nearly two decades. But China’s internet market has evolved, regulatory pressure on ad-heavy platforms has mounted, and Baidu has repositioned itself around a question that now dominates global technology: if artificial intelligence is the next computing platform, how can an incumbent search company stay relevant?

The company’s answer has been aggressive diversification. Baidu operates Baidu Cloud, a cloud-computing and AI infrastructure service sold to enterprises. It is developing an autonomous-vehicle platform called Apollo. It owns iQIYI, a video-streaming service similar to Netflix. It has built a large AI platform called Ernie, which underlies various AI applications and chatbots. The company also retains its old core business — search advertising — but the trajectory is unmistakable: Baidu is trying to evolve from a search and advertising company into a technology infrastructure and services company.

This evolution is neither complete nor certain. The core search business still generates the majority of Baidu’s revenue and cash flow, and many of the newer businesses are not yet profitable. The company faces intense competition from Alibaba and Tencent in cloud services, from Tesla and others in autonomous vehicles, and from the entire global AI ecosystem in large language models. Yet Baidu possesses real assets for these battles: a huge installed base of users and developers on its platform, deep expertise in Chinese-language natural language processing, and access to data from search that few companies outside China can match.

The search monopoly and how it was built

Baidu’s dominance in Chinese search began with a simple advantage: it was built for the Chinese language. Early search engines optimized for English struggled with Chinese’s different grammar, character system, and morphology. Robin Li, a developer who had previously worked on search algorithms at Infoseek and RankDorf in the late 1990s, recognized this and designed a Chinese-language search engine from the ground up. Baidu launched in 2000 as a search service for other Chinese websites, then pivoted to become a consumer search engine, and by the mid-2000s had become the dominant player in mainland China.

The dominance was reinforced by the same network effects that made Google dominant globally. Better search results attracted more users; more users meant more data about what people searched for and which results they clicked on; more data meant better algorithms. Baidu’s scale in China allowed it to invest heavily in relevance, crawl the Chinese internet more thoroughly than rivals, and deploy more sophisticated ranking algorithms. By the early 2010s, Baidu held roughly 60 to 70 percent of the search market in mainland China, a position it has held for years.

Profitability followed naturally. As a monopoly or near-monopoly search engine in a country of more than a billion people and hundreds of millions of internet users, Baidu could charge premium prices for advertiser placement. A business owner in Shanghai selling machinery or a pharmacy seeking customers could buy ads on Baidu at rates that reflected the platform’s dominance. The advertising business generated enormous margins — search advertising is a license to print money in any large market where one company holds a near-monopoly.

Branching out: From search into platform and services

In the mid-2010s, Baidu began diversifying. This was partly defensive — Chinese regulators were growing skeptical of the advertising model, and there was risk that stricter ad rules would undermine the core business — and partly strategic, as technology platforms worldwide were consolidating around AI and cloud infrastructure.

Baidu Cloud, launched in 2011 but scaled much more aggressively from 2015 onward, became one of the company’s main growth engines. Cloud services are what they sound like: renting computing power, storage, and software services to businesses that do not want to build their own data centers. Alibaba dominates this market in China through Aliyun, but Baidu Cloud has carved out meaningful share by offering AI development tools and services optimized for large language models and machine learning. For a data scientist or a company training AI models on Chinese-language text, Baidu’s cloud platform offers pre-built models, APIs, and infrastructure designed for that purpose.

In 2016 and 2017, Baidu invested heavily in Apollo, an open-source autonomous-driving platform. The idea was bold: instead of Baidu building its own self-driving car, it would build the software platform that other vehicle manufacturers, startups, and systems integrators could use to develop autonomous vehicles. This would give Baidu a position in the autonomous-driving industry without requiring the massive capital and operational complexity of a car manufacturer. The platform has since attracted partnerships with other Chinese automakers, and Baidu has also launched its own robotaxi service in selected Chinese cities.

The company also acquired and operates iQIYI, a video-streaming platform that competes with Tencent Video and Youku in China. iQIYI is a subscription and advertising-supported service offering Chinese-produced dramas, films, documentaries, and variety shows. It has never been profitable at the operating level; it loses money most quarters and is more valuable to Baidu as a customer-acquisition and data-collection asset than as a standalone business.

The newest and most symbolically important venture is Ernie, Baidu’s large language model and AI platform. Released in 2023 and subsequently updated, Ernie is Baidu’s answer to ChatGPT and other large language models, but optimized for Chinese language and integrated with Baidu’s search engine, cloud platform, and other properties. The company has positioned Ernie as the foundation of its AI business, providing the underlying model that powers chatbots, document-analysis tools, and other applications sold to enterprise customers.

The competitive position in a crowded market

Baidu’s advantages are real but increasingly challenged. In search, it remains dominant in mainland China by market share, but that dominance is not absolute: Alibaba and Tencent have large user bases and messaging platforms that can route people to information without passing through Baidu’s search. Tencent’s WeChat is arguably a larger platform by daily active users and time spent, and users can find information and transact within WeChat without ever needing a dedicated search engine.

In cloud services, Baidu competes against Alibaba (which is much larger), Tencent, Huawei, and a field of smaller players. Baidu’s strength is AI and machine learning tools, but those are increasingly commoditized. In autonomous vehicles, Baidu’s Apollo platform competes against internal research by Tesla, startups like NIO and Li Auto, and other Chinese companies. It is not clear that an open-source platform model will win in that market; it may be that integrated car companies or vertically integrated AI companies end up dominating.

The one clear moat Baidu retains is its historical position in Chinese-language AI. The company has trained large language models on vast amounts of Chinese-language text and has access to search data that shows what Chinese speakers actually search for and which results they find useful. This gives Baidu a head start in building AI systems optimized for Chinese. As Chinese regulations increasingly restrict access to training data and favor domestic companies, this domestic expertise becomes more defensible.

Regulatory headwinds and the profit challenge

Baidu operates in a regulatory environment that has grown more skeptical of large technology platforms. The Chinese government has scrutinized the search advertising business for misleading ads (particularly for healthcare and financial products), and regulators have pushed the company to reduce the number of ads shown per page and to separate sponsored results from organic results more clearly. These changes have cut into the profitability of the core search business without making the service worse for users or advertisers; they simply shifted money from Baidu’s pocket to society’s benefit.

More broadly, the Chinese government has stated ambitions to control AI development within the country, licensing AI models and placing restrictions on how data can be used for training. This creates uncertainty for Baidu’s cloud and AI divisions: regulatory rules could shift suddenly, and the company must adapt quickly or lose customers. The government’s broader stance toward large technology companies has also swung between periods of support (when the company serves strategic national interests) and periods of restriction (when the company is seen as accumulating too much power).

The path to profitability in a new era

The fundamental question facing Baidu is whether its diversification into cloud services, autonomous vehicles, AI, and streaming can generate enough profit to replace the shrinking core search business. Cloud services and AI infrastructure can be highly profitable, but Baidu is not the clear leader and faces aggressive competitors. Autonomous vehicles are a long-term bet with enormous capital requirements and highly uncertain returns.

For now, Baidu still generates more than half its revenue from search advertising, and the core business remains cash-generative despite the regulatory pressures. The company has been disciplined about allocating capital, spinning off units (such as iQIYI) when they became too large to fit the parent company’s profile, and focusing investment on areas where it has genuine competitive advantage. How successful this new positioning turns out to be will depend on whether the company can establish durable leadership in AI and cloud services before new entrants or larger competitors crowd the market.

How to research Baidu

Start with the company’s annual 10-K filing (SEC CIK 0001329099), which breaks revenue by segment and provides detail on the cloud business, search business, and iQIYI separately. Pay close attention to the margin trends: the company’s ability to defend profitability depends on whether it can grow cloud and AI revenues without proportionally increasing costs.

The quarterly earnings call is where management discusses near-term pressures in the search business and the trajectory of newer divisions. Watch for commentary on Ernie adoption, Baidu Cloud customer growth, and the status of Apollo robotaxi operations.

Key metrics to track include search revenue growth (or contraction), Baidu Cloud revenue growth and gross margin, and operating margin for the company as a whole. Because the company is transitioning from one business model to another, traditional valuation metrics can be misleading; it is more useful to track segment-level unit economics and how management is allocating capital among the various bets.