Founder of Baichuan Intelligence, Wang Xiaochuan, Discusses the Price Wars of Large Models: Why I'm Not Joining the Giants' Game

After observing the launch of various large model applications over the past six months, Baichuan Intelligent has debuted its first AI application, "Bai Xiao Ying." Founder Wang Xiaochuan asserts that the timing of this launch is not late; he believes it may even be slightly premature given the current capabilities of AI models. He aims to avoid crafting a product that users abandon due to ineffectiveness. Although "Bai Xiao Ying," which launched on May 22, does not yet fully realize Wang's vision for a super application, he remains hopeful it will enhance usability compared to earlier versions.

Wang, who has largely remained out of the spotlight, maintains a candid approach. In evaluating other applications, he criticized a competitor's product from a leading internet company for lacking usability. He defines a super application as one that achieves a minimum of 30 million daily active users (DAU) and discussed the ongoing price war within the industry, clarifying that Baichuan Intelligent will not engage in price reductions.

Reflecting on this price war, Wang, a seasoned industry leader and founder of Sogou, views the trend among major firms as predictable. He notes that companies are overly optimistic about this era and willing to operate at a loss to enter the market. Following ByteDance’s pricing cuts, other major players like Baidu, Alibaba, Tencent, and iFlytek have also reduced prices. Wang argues that this price disruption differs from past competitive events, emphasizing that it focuses on productivity supply in the B2B sector rather than altering production relationships.

Wang identifies intense competition among cloud service providers in this landscape, highlighting the roles of ByteDance’s Volcano Engine, Alibaba Cloud, Tencent Cloud, and Baidu Cloud. He emphasizes that AI has opened new market opportunities for these companies, transforming China’s cloud services landscape. He is optimistic that as large models become more affordable or even free, the B2B application market will flourish, creating new cycles of opportunity and disruption, even at the risk of forming new bubbles.

On the topic of price competition, Wang firmly stated that it is a game for larger entities—one in which Baichuan will not participate. His future strategy involves a "dual-drive" approach, focusing on simultaneously developing foundational large models and AI applications. He advises startups not to limit themselves to B2B ventures, as the consumer market is significantly larger. With experience in developing products for 600 million users, Wang expects consumer applications to thrive and aims to craft a super application from the outset, targeting DAU figures between 30 million to 300 million.

Wang envisions these applications as not just tools, but as companions for users. "Bai Xiao Ying" differentiates itself from other AI applications through excellent search capabilities and user interaction. For instance, if a user asks, "What should I do if my car hood won’t open?" the application first collects information about the car’s make and model along with the user's location before providing tailored advice and contacts for nearby repair shops.

Wang aims for an interactive environment where AI can accurately discern user intent, assisting them in refining their queries without needing precise prompts. He recounted a relatable instance involving a corporate executive frustrated with the inability of models to perform complex multiplication, underscoring the idea that large models should be considered partners rather than simple computational tools.

He also thoughtfully designed the "Bai Xiao Ying" logo, choosing warm colors. While many AI applications prefer cool tones to reflect technology, he believes modern technology should exude warmth.

Though he acknowledges that current AI applications, including "Bai Xiao Ying," are yet to fully embrace the companionship aspect, he remains confident in Baichuan's ability to develop a super application using this generation's technology.

Regarding the business model, Wang criticized existing monetization strategies that involve charging both consumer users and B-end clients, which he finds unappealing. Instead, he believes the path to successful monetization lies in establishing a solid user base first.

Since launching "Bai Xiao Ying," Baichuan has not yet introduced any paid features, as Wang feels it is premature to discuss monetization. He prioritizes identifying scenarios that encourage users to pay, suggesting a successful model emerges only when a scenario appeals to millions of paying users.

In contrast to competitors that aggressively market to attract customers—some reportedly spending heavily per user—Wang stated that Baichuan will advance its AI application journey without engaging in such marketing tactics, viewing them as unhealthy. He expressed confidence in Baichuan's ability to pursue a sustainable approach in its AI endeavors.