2026/01/23

The key is to let users have a go first and then let them make their own choices.

Recently, the topic of the "race for AI entry points" has ignited intense debate.

The market always outpaces public opinion, with tech giants continuously taking the baton in succession. Early last year, Tencent's Yuanbao made a rapid surge, investing heavily in promotion and once topping the free app chart in Apple's China region. By mid-year, ByteDance's Doubao began to gain momentum, expanding from software to hardware and drawing significant attention with its "disruptive innovation." At the end of the year, Alibaba's Qianwen rose to prominence, with advertisements flooding the screens and even surpassing Yuanbao in some aspects. Recently, it has been upgraded to an intelligent agent.

In terms of data, besides DeepSeek, AI applications from ByteDance's Doubao, Alibaba's Qianwen, and Baidu's Wenxin Assistant have all surpassed monthly active users of over 100 million, indicating a trend towards a tripartite balance of power. But does this mean the game is over? Clearly not; it's more like celebrating prematurely.

Whether it's Doubao integrating or developing more and more intelligent terminals, or Qianwen's "all-in-one" intelligent agents, it only suggests that this AI marathon has just become more intriguing. The real competition is yet to come, and new dark horses could emerge at any time during the race. There are precedents for this as well. Ten years ago, WeChat Pay staged a remarkable comeback against Alipay during the Spring Festival, achieving in one week what Alipay had accumulated over ten years. More recently, last year, DeepSeek's low-cost comeback and overnight success saw it reach 100 million users in just seven days.

However, after a year of development, the performance of each tech giant's AI strategy has varied. For instance, Alibaba has been more aggressive, ByteDance has taken a unique approach, Baidu has made steady and layered progress, and Tencent has been taking one step at a time, not falling behind but moving at its own pace. These are all new annotations to this year's AI competitive landscape.

01

Tech Giants Sharpen Their AI Strategies

At the turn of the year, tech giants are eagerly announcing the latest progress in their AI businesses, further intensifying the discussion around the "AI super entry point."

On January 22, Baidu released and launched the official version of its native full-modal large model, Wenxin 5.0. This model boasts full-modal understanding and generation capabilities, supporting the input and output of various types of information, including text, images, audio, and video. Previously, Baidu's Wenxin Assistant had already surpassed 200 million monthly active users. Currently, Wenxin Assistant can directly access numerous services, fully covering e-commerce, health, local life, finance, and other fields.

A week ago, Alibaba's Qianwen App announced its full integration with Alibaba's ecosystem businesses, including Taobao, Alipay, Taobao Deals, Fliggy, and Gaode. It upgraded and launched over 400 AI-powered service functions, ranging from daily conversational Q&A to executing complex tasks in real life. The Qianwen intelligent agent has also officially entered the physical world.

Notably, within less than two months of its launch last November, the monthly active users of Qianwen's consumer-facing end have exceeded 100 million. This rapid development has been hailed as a remarkable victory by Alibaba CEO Wu Yongming.

Earlier still, ByteDance's Volcano Engine was announced as the exclusive AI cloud partner for the 2026 Spring Festival Gala, marking the beginning of the tech giants' Spring Festival marketing battle. During this year's Spring Festival Gala, Volcano Engine will be deeply involved in the CCTV Spring Festival Gala programs, online interactions, and live video streaming. The intelligent assistant Doubao will also introduce various interactive features. Meanwhile, Doubao's smart glasses have entered the pre-shipment stage, with a total planned production of approximately 100,000 units for the first generation, primarily targeting senior users of "Doubao." Additionally, there are rumors that ByteDance is developing a new generation of AI earphones.

Clearly, Doubao, which leads the industry in monthly active users, is not content with just the mobile phone industry. It hopes to extend the potential of this large model application to more intelligent terminals, including AI glasses, AI earphones, as well as tablets and smart cars, achieving a software-hardware integrated ecosystem and seeking more entry points.

In contrast, although Tencent has not launched new AI products, organizational and personnel changes reflect a shift in its AI strategy. For instance, a month ago, Tencent announced an upgrade to its large model research and development architecture, establishing the AI Infra Department, AI Data Department, and Data Computing Platform Department to comprehensively strengthen its R&D system. Meanwhile, former OpenAI researcher Yao Shunyu officially joined Tencent as Chief AI Scientist and Head of the AI Infra Department. Yao's arrival is seen as Tencent's effort to strengthen the integrated management of models and infrastructure, particularly enhancing the fusion between algorithm R&D and underlying computing power, and improving the infrastructure's ability to adapt to model evolution.

In his first public appearance, Yao Shunyu also stated that his recent focus is on models, products, and the field of artificial intelligence. He also mentioned the differentiated characteristics and demand characteristics of the B-end and C-end markets, especially emphasizing that the revenue brought by the ToB market will be greater in the future.

02

Diverse Intentions, Varied Outcomes

The actions of tech giants in developing their AI businesses also reflect their respective survival situations. Industries such as advertising, e-commerce, and search may be the first to be disrupted by AI, which is an important reason why Alibaba and Baidu are more proactive and enthusiastic in embracing AI.

Alibaba Chairman Joseph Tsai once pointed out that the e-commerce industry is facing increasingly fierce competition, with new entrants constantly emerging, especially social e-commerce platforms like Douyin, which adopt new business models and pose a significant threat to traditional e-commerce giants. Therefore, Alibaba plans to reconstruct its e-commerce ecosystem through AI technology. In the future, its e-commerce business will no longer be limited to traditional shelf-style e-commerce but will create a smarter and more efficient consumption experience through innovative models such as AI-driven personalized recommendations, social shopping, and virtual shopping assistants.

Last May, Alibaba clearly defined its core strategic direction for the next three to five years as "e-commerce" and "cloud + AI," emphasizing that AI technology will become the core driving force for all its businesses. On the one hand, it provides computing power services based on Alibaba Cloud and, with Tongyi Qianwen, offers large model services to numerous developers, which will also drive the growth of sales revenue for other cloud-based products. On the other hand, as the largest e-commerce platform in China, Alibaba regards AI as a key driving force to enhance its competitiveness and applies it in its own ecological scenarios to create more breakthroughs in the AI-to-C space. Therefore, whether from the perspective of business transformation or the company's new narrative, Alibaba cannot afford to lose the AI battle.

Throughout 2025, Alibaba has been frequently active around AI, forming a three-dimensional offensive with Alibaba Group's core products advancing head-on and ecological partners such as Ant Group responding on the flanks. This includes the Qianwen App taking the lead in becoming an AI intelligent agent, the transformation and upgrading of Kuake, and the extension and future layout of AI glasses as flanks, while Ant Group's "Lingguang" and "Afu" have achieved differentiated breakthroughs and in-depth penetration in tool creation and the health vertical field.

Compared to Alibaba's aggressive push into AI, Baidu has been an early mover with a steady and layered approach, also aiming to prove itself. This is evident from Baidu being the first global tech giant to launch a large language model product, Wenxin Yiyan, in March 2023. Although Wenxin Yiyan took the lead, it was later caught up and surpassed by competitors due to various reasons such as technical performance, ecological support, user experience, and marketing promotion.

In fact, Baidu's technical strength is not weak, and its "computing power-framework-model-application" full-stack layout is itself a long-term barrier and self-contained ecosystem. For example, Baidu has built a full-stack AI system covering self-developed chips (Kunlun Core), deep learning frameworks (PaddlePaddle), the Wenxin large model, and AI applications such as Apollo Go, AI search, and digital humans. This enables Baidu to not only provide technical components but also output systematic AI internalization methodologies.

It can be said that Baidu has a strong customer base in the B-end, especially among government and enterprise clients, as its monetization efficiency is higher. However, it is slightly lacking in the C-end, lacking a corresponding commercial closed loop. In the future, it needs more entry points to connect with the physical world.

Baidu's current situation is similar to Tencent's but also different because Tencent has gaming as its cash cow and WeChat as its powerful weapon, so Tencent is not in a hurry. As previously mentioned, after the release of ChatGPT, Lu Shan, President of Tencent TEG, stated at an internal strategic meeting that GPT took three years to productize, and Tencent's large model will undoubtedly encounter many challenges, so there is no need to rush.

Tencent's value creation in AI lies, on the one hand, in integrating Yuanbao's ecosystem by connecting it with dozens of internal applications such as WeChat, QQ, Tencent Meeting, and Tencent Docs, covering core scenarios such as social interaction, office work, and consumption, allowing C-end users to engage with it, while independently testing various AI functions within WeChat. On the other hand, it involves systematically integrating AI into various core businesses such as gaming, advertising, and enterprise services to drive business efficiency and revenue growth, as reflected in last year's financial reports.

For Tencent, its biggest future AI lever is still WeChat. As a Tencent executive said, WeChat will eventually launch an AI intelligent agent that can understand user intentions and use WeChat's ecosystem of mini-programs, payments, and content closed loops to execute tasks.

Among the tech giants' AI layouts, ByteDance is the most intriguing due to its unconventional approach, as seen in Doubao's mobile phone strategy. As a company that rose to prominence with algorithmic recommendations, ByteDance excels at transforming cutting-edge technologies into user-perceptible, high-frequency products. Leveraging billion-user entry points such as Douyin and Toutiao, its C-end AI products have quickly gained popularity, forming a three-dimensional product matrix of "daily tools + creative platforms + emotional companionship."

Meanwhile, ByteDance has not bet on a single model but has built a "model supermarket" centered around the Doubao large model family, covering multiple dimensions such as general-purpose, vertical, and multimodal models. In addition to the C-end, through Volcano Engine, ByteDance has also opened up its AI capabilities to various industries, forming a service system of "models + platforms + solutions." Most importantly, ByteDance has not gone it alone but has promoted AI ecosystem co-construction by establishing industry alliances. For example, in the mobile phone sector, it has formed a smart terminal large model alliance with OPPO, Vivo, and Xiaomi, and in the automotive sector, it has collaborated with Tesla and other automakers to create an automotive large model ecosystem alliance. All of these have broadened its AI path.

The latest news indicates that ByteDance plans to increase its investment in the artificial intelligence field this year, with an initial planned capital expenditure of 160 billion yuan, about half of which will be used to purchase advanced semiconductors for developing artificial intelligence models and applications. The flexibility of a non-listed company in strategic investment is also an aspect of its competitiveness.

03

It's Premature to Talk About the Competition for AI Entry Points

From the "hundred-model battle" three years ago to the current competition for the "AI super entry point," the AI landscape has seen its ups and downs. Some players have dropped out midway and faded into obscurity, while new players have risen against the odds and become overnight sensations. Some continue to invest heavily, struggling to find landing scenarios, while others have successfully commercialized and entered the capital market. In the end, the AI competition still comes down to technology, funding, and ecological strength, and after the shakeout, it has become a game of competition and cooperation among tech giants.

As industry insiders have said, the current short-term focus for tech giants competing for AI entry points is to capture user mindshare and scale, getting users to form the habit of "turning to AI for their needs." The medium-term goal is to translate this into commercial returns by empowering core businesses through AI entry points. The long-term objective is to control the industry ecosystem, establish dual barriers in technology and commerce, and ultimately form an ecological pattern of "super entry point dominance + vertical player attachment," controlling core traffic and data discourse, and transforming AI capabilities into a sustainable profit engine and industry pricing power.

However, despite the emergence of four AI applications in China that each boast over 100 million users, it would be premature to claim that they have definitively secured their positions as primary gateways to the AI landscape. This assertion can be dissected from three key perspectives: Firstly, generative AI is inherently prone to issues such as AI hallucinations (where the AI generates false or misleading information) and flawed decision-making processes, necessitating the establishment of user trust. Particularly when AI-generated content (GEO) is commercialized, it may foster the proliferation and intrusion of lower-quality content, thereby elevating the rate of harmful AI outputs. Secondly, in terms of the richness and scalability of application functionalities, the cross-application operations of Doubao Mobile previously ignited considerable social debate concerning safety and privacy concerns. While the Qianwen intelligent agent can directly execute AI-driven shopping functions, such as ordering takeout, making purchases, and booking flights without redirecting users, its capabilities are confined to Alibaba's ecosystem applications and do not extend to other external platforms. Consequently, its usable scope remains constrained. If Qianwen fails to expand its reach beyond the current ecosystem, it will merely constitute an additional app on users' phones rather than evolving into a genuine AI entry point. Thirdly, the cultivation of user habits is paramount. The current user base is transient; the crux lies in genuinely motivating users to not only use but also habitually rely on AI, a feat contingent upon product stability, demand dependency, and even hardware compatibility.

An IDC report also underscores that, concerning software and ecosystems, there exists substantial room for enhancement in the profound empowerment of AI capabilities across more intricate and varied scenarios. This limitation is primarily attributed to the progress of software ecosystems and content innovation, coupled with the ongoing need to cultivate user habits and conduct scenario education.

Nevertheless, one point remains abundantly clear: this year, the competition within the AI application sphere will further intensify its focus on ecological synergy capabilities, particularly the extent of integration between AI technology and existing businesses, as well as the efficacy in enhancing business efficiency. Moreover, compliance and security will gradually emerge as a consensus across the entire industry, representing the fundamental benchmark of industry competition. Both regulatory mandates and user expectations compel companies to prioritize data governance and privacy protection.

Therefore, it is premature to engage in discussions about the "competition for AI entry points." The priority should lie in enhancing product functionality and user experience, allowing individuals to experiment with the technology firsthand before making informed choices. This, indeed, is what truly matters.

- END -