The Chinese New Year of 2026 will undoubtedly be a battle for the Chinese internet companies, with AI applications taking center stage.
So far, ByteDance (Doubao, Volcano Engine), Tencent Yuanbao, Alibaba (Qianwen, Alipay), and Baidu Wenxin have all invested massive resources, using red envelopes as leverage to break into the AI-driven Spring Festival season.
Previously, ByteDance announced that Volcano Engine would be the exclusive AI cloud partner for China's national broadcaster CCTV’s Spring Festival Gala, with Doubao deeply integrated into the show’s interactive segments. Tencent has put up 1 billion yuan in cash red envelopes to promote the Yuanbao App, embedding it into group chats and leveraging its social function. While Qianwen has yet to reveal if it will join the red envelope rivalry, media reports indicate it has announced sponsorship of Jiangsu TV’s Spring Festival Gala, suggesting more Spring Festival initiatives will follow.
The Spring Festival is the most important holidays for the Chinese. In 2014, WeChat red envelopes made their debut, opening the door for WeChat Pay and marking a milestone in the popularization of mobile payments in China.
Over the past decade or so, these kinds of promotions have not only added a technological touch to traditional holidays, but have also become critical battlegrounds for internet giants, evolving from simple social interactions to fierce competition for users and industry positioning.
This year, with ByteDance, Alibaba, and Tencent all gathering for the Year of the Horse Spring Festival, the message is clear: the competition for AI application entry points on the consumer side has reached a fever pitch. Especially for these three giants, each is leveraging its own strengths and resources to make different strategic choices. As they stage an AI “Three Kingdoms” race, they also face challenges.
Strategy never emerges out of thin air; it is deeply rooted in a company’s history, culture, and core capabilities—what is often called the company’s DNA. One thing is certain: their core goal is to secure a leading position in the AI era, or even take the lead, ultimately winning this AI endurance race.
ByteDance is Staking out the AI Entry Point
Among the three companies, ByteDance is the most aggressive. This continues its tradition of “miracles through sheer effort,” enabling it to avoid limiting itself to a single direction. Instead, it is pushing forward on multiple fronts simultaneously—applications, hardware, and developer platforms—adopting a multi-pronged approach.
On the consumer product front, ByteDance has adopted an intensive deployment strategy. According to incomplete statistics, ByteDance has launched more than a dozen AI applications, covering areas such as chat, social networking, and tools, with Doubao as its core AI application.
Doubao’s Agent capabilities are deeply integrated with ByteDance’s content and entertainment ecosystem, and its positioning leans more towards entertainment. Users can generate fun images and videos, and interact through voice conversations, among other features.
On the AI hardware side, ByteDance has also been active. By the end of 2025, the “Doubao Mobile Assistant” was launched on ZTE smartphones, causing a stir in the industry. Unlike traditional mobile voice assistants, Doubao Assistant, through deep collaboration at the operating system level, has gained the ability to directly operate various phone functions.
However, to achieve this capability, it must read and understand information on the user interface (GUI), which could cross the red line of user privacy and data security, sparking controversy within the industry. Many smartphone manufacturers, especially those in banking and payment applications, generally resist or block such features due to security and risk control concerns.
In addition to smartphones, ByteDance is also rumored to be launching Doubao AI earphones, AI glasses, and other hardware.
At the platform level, ByteDance’s AI Agent tool “Kouzi” recently announced its 2.0 brand upgrade, now integrating capabilities such as Agent Skill and Agent Coding. Through Kouzi, developers and ordinary users can create Agents covering a variety of vertical scenarios. These Agents can be integrated with one click into ByteDance platforms such as Douyin and Feishu.
This is a combined strategy of “application + hardware + platform,” continuing ByteDance’s “rapid and aggressive” style since the mobile internet era: ByteDance rose to prominence through algorithms, from personalized news recommendations on Toutiao to immersive short video feeds on Douyin. Each phenomenal success has stemmed from the extreme application of recommendation algorithms and the precise capture of user interests.
Over time, this has also become a path dependency, or even a part of ByteDance’s DNA, leading to a deep-seated anxiety during this period: when users can directly and clearly express their needs to Agents using natural language, the “guessing + recommendation” information feed model that ByteDance relies on for success may be disrupted.
The core of ByteDance’s AI strategy is to “seize new gateways,” aiming to develop almost every software and hardware product with gateway potential. Therefore, ByteDance must move fast, leveraging its expertise in product matrix strategies and rapid iteration to bet on every possible form that could become the next super gateway.
At present, although Doubao has already gained a considerable user base on the consumer side, it is far from time to celebrate. With competition intensifying, how long its first-mover advantage can last and how to establish a sustainable new business model remain significant challenges.
Meanwhile, the “entry point” positioning strategy is also a double-edged sword. As Doubao becomes more deeply integrated at the hardware and operating system levels, it has sparked industry concerns about user privacy and data security.
For example, in recent days, doubts have resurfaced regarding the security risks of Doubao phones. @Doubao Mobile Assistant responded promptly and directly, stating that all feature calls are strictly based on user authorization, and drawing a clear line on the core points of contention: insisting on cloud processing, “no storage, no training,” with end-to-end encrypted data transmission and multiple layers of protection to ensure privacy.
This means that as ByteDance attempts to take the technological lead, it must also face tough tests in terms of ecosystem compatibility and industry trust. Striking a balance between rapid expansion and building trust will be crucial in determining whether its AI ambitions can ultimately be realized.
Alibaba’s Aggressive AI-to-Consumer Push
Alibaba’s pivot has been nothing short of swiftness. In the first half of 2025, its main AI-to-consumer focus was still on Quark. From the second half of 2025 to now, it has realigned its strategy and offerings, positioning Qianwen as its new flagship and launching an “AI blitzkrieg”—first by releasing the standalone Qianwen App, and then rapidly integrating Alibaba’s core business ecosystem into the Qianwen App.
Currently, Alibaba’s AI-to-consumer strategy centers on “Qianwen + Alibaba Cloud,” deeply integrating with e-commerce businesses like Tmall and Taobao, as well as other Alibaba services such as Amap and Alipay. The goal is to seize the AI shopping and industry service entry point, reshape the “people-goods-scenario” model, and pursue a “full-stack technology + industry penetration” approach.
The release of Qianwen 6.0 marks a key milestone, as Alibaba ushers in the era of Agent-based services—bridging the Agent capabilities that Doubao Mobile failed to extend across systems within its own business ecosystem.
At the model level, Tongyi Qianwen has adopted an open-source strategy from the outset. In the early stages of a technological paradigm shift, open-sourcing core technologies can greatly accelerate adoption, unite a broad developer community, and quickly build an ecosystem around its own platform.
This is a strategy at which Alibaba excels—leveraging organizational strengths to rally resources and win major battles. At the start of the new year, Alibaba Group CEO Wu Yongming wrote in his “New Year’s Letter” to all employees that several major battles had been fought beautifully this year, including the launch of the Qianwen App.
Alibaba’s DNA is “business” and “infrastructure.” From its early days in the B2B sector with the mission of “making it easy to do business anywhere,” to Taobao’s C2C e-commerce, and then to Alipay, Alibaba Cloud, and Cainiao Logistics, Alibaba has spent the past two decades building the foundational infrastructure for the business world.
This DNA shapes the distinct core of its AI strategy: building a solid foundation for the B-side (business), while breaking through on the C-side (consumer), operating on both fronts. This seems to mirror ByteDance in another way: one is the C-side feeding back into the B-side, while the other is the B-side supporting the expansion of C-side business.
In Alibaba’s view, AI is not an isolated product, but the underlying operating system for the next generation of cloud computing, commerce, and all enterprise services—essentially, a form of infrastructure. To this end, Alibaba has invested heavily in AI infrastructure, even planning to invest over $53 billion (about 380 billion RMB) in cloud computing and AI over the next three years.
For Alibaba, AI models or applications themselves may not directly generate profits, but by leveraging Alibaba’s full-stack technical capabilities (from chips, computing power, and models to real-world scenarios), the company aims to build a thriving AI business ecosystem. This ecosystem connects developers, merchants, and consumers, channeling token consumption to Alibaba Cloud and other foundational infrastructure, thus forming a complete, bidirectional “C—B” closed loop.
Among major AI players, Alibaba arguably has the most comprehensive capabilities. However, it lacks high-exposure channels like WeChat or Douyin within its own ecosystem, so its AI-to-consumer traffic relies mainly on network-wide distribution and the organic word-of-mouth “self-driven” traffic generated by its own large models.
At the same time, building infrastructure is a long-term, capital-intensive process with relatively slow returns. As Alibaba embraces the new AI paradigm, it needs to further find solutions to rebuild and balance its business model.
Tencent: Steady Progress in AI-driven Business
Tencent’s DNA is “social” and “connection.” It is closest to Chinese users and understands their social habits better than anyone. From QQ to WeChat, Tencent has built the underlying “relationship chain” of China’s internet, and, using this as a core, has constructed a business ecosystem that encompasses nearly every aspect of Chinese people’s digital lives—making it the platform with the most scenarios and entry points on the Chinese internet.
This also sets Tencent apart from ByteDance and Alibaba. Its core mission is not to externally "seek out" or "seize" new entry points and scenarios, but rather to address AI-related anxieties internally—that is, to upgrade and transform its various products with AI without compromising user experience, and to leverage AI to empower and reconstruct its own vast and complex business ecosystem.
This means that Tencent’s AI strategy is not about making a sudden breakthrough overnight, but rather about undertaking a systematic project akin to "turning an elephant around"—a process that truly tests patience and internal coordination.
The "Yuanbao" App launched by Tencent has shown notable user growth over the past year through rapid iteration. It has not only added numerous native AI features, but also allows users to conveniently search for and access music, films, and other content from the Tencent ecosystem within Yuanbao.
But Yuanbao’s real distinction lies in its integration with Tencent’s product ecosystem. It serves as the Yuanbao contact in WeChat, the Yuanbao avatar in Tencent Meeting, the Yuanbao plugin in QQ Browser, and more. Yuanbao is also active in the comment sections of various Tencent products, where users can simply @Yuanbao to chat and interact with it.
These efforts aim to make Yuanbao an atomic AI capability—a fundamental module that can be integrated into more Tencent products. As Yuanbao’s own capabilities iterate and improve, the AI capabilities and user experience across the entire Tencent ecosystem will also be elevated.
From standalone task-oriented tools to content platforms and then to core social products, Tencent’s various offerings are quietly undergoing AI-driven transformations step by step. Tool-type products like Sogou Input Method and QQ Browser have undergone comprehensive AI upgrades.
Content service products such as Tencent News and Tencent Video are incorporating AI capabilities in a subtle, unobtrusive manner, tailored to their unique user scenarios.
Communication and social products like WeChat and QQ, being closest to users and massive in scale, face greater challenges in AI transformation, and Tencent continues to explore this area.
The prevailing industry consensus is that AI applications are shifting from chatbots to Agents. The usability and effectiveness of Agents depend, to a certain extent, on how many real, high-quality tools and products are available within the ecosystem. Tencent’s steady push to AI-enable its various products is laying a solid foundation for its Agent ecosystem.
At the end of last year, Tencent made significant organizational adjustments for its large model business, bringing in former OpenAI researcher Yao Shunyu as Chief AI Scientist in the CEO/President’s Office, who is also responsible for the Large Language Model Department and the newly established AI Infra Department.
Regarding his work direction after joining Tencent, Yao Shunyu once mentioned, "Tencent is a company with a strong To-C DNA. We will consider how to enable AI to deliver value to more users and implement AI across a wide range of scenarios."
At the latest 2025 Tencent all-hands meeting, Pony Ma emphasized that the core of Tencent’s AI lies in the long-term competitiveness of its products and user experience. On the much-discussed topic of AI integration within the WeChat ecosystem, Pony Ma made it clear that an all-in-one AI suite is not necessarily what everyone wants. In the future, Tencent will continue to adhere to a decentralized approach, planning the intelligent ecosystem of WeChat in a way that balances user needs and privacy protection.
Recently, several companies including Google and OpenAI have predicted AI development trends for 2026. There is a consensus that "Agents will become the absolute core, and multi-agent systems are expected to demonstrate truly visible industrial impact." Agents will undoubtedly be a key focus for Tencent in the AI arena.
Tencent has a solid foundation in product ecosystems, a wealth of high-quality Chinese-language content in its content ecosystem, and contextual scenarios generated in social interactions. Coupled with an increase in talent density, Tencent’s advantages will become even more prominent in the Agent era.
In comparison, what Tencent needs to improve is its large model capabilities, as well as the degree of integration between large models and products. The introduction of Yao Shunyu and organizational restructuring also demonstrate Tencent’s determination.
Ma also recently mentioned that the Hunyuan large model team and the Yuanbao product team have already begun cross-functional collaboration, secondments, and joint design efforts. This organizational structure ensures that the foundational model capabilities are closely coupled with upper-level product needs, preventing a disconnect between technology and the market.
Tencent’s goal is not to create a viral hit, but to achieve AI transformation of its entire business system and ecosystem through AI architecture and system design—what it calls "system-level intelligence." This poses even greater challenges to the execution and strategic resolve of Tencent’s teams. For now, Tencent is maintaining its own pace; it may not seem aggressive, but its business penetration is progressing rapidly.
As noted in a Goldman Sachs report, Tencent’s strategy is "to remain restrained" and "to achieve ecosystem penetration." This strategy requires massive internal coordination and profound organizational transformation, which is extremely challenging. However, once achieved, the resulting competitive moat will be even deeper.
Going forward, Tencent is expected to continue focusing on its areas of expertise. For example, the Yuanbao App has recently begun internal testing of a new feature called "Yuanbao Party," where users can create or join a "Party," allowing AI to participate in chats, moderate the atmosphere, and execute tasks. This is seen as Tencent’s key attempt in the field of "AI + Social."
Caption: Yuanbao Pai currently in internal testing
Judging from Yuanbao’s launch of 1 billion yuan in red envelopes and its social gameplay features, we may see more visible moves from Tencent in 2026.
Big Tech’s AI Race
Overall, the differences in AI strategies among ByteDance, Alibaba, and Tencent fundamentally stem from their respective core business models, resource endowments, and commercial ecosystems.
Specifically, ByteDance relies on consumer-side traffic to quickly achieve breakthroughs through a “traffic-content-commerce” closed loop; Alibaba leverages its e-commerce and cloud ecosystem to build a “technology-commerce” moat with full-stack capabilities; Tencent, with its highly sticky social ecosystem, adopts a steady and incremental penetration strategy, vigorously advancing product AI integration and continuing to focus on its areas of expertise.
2026 is shaping up to be a highly anticipated year for AI development in China. In the foreseeable future, China’s AI narrative will continue to be led by companies like ByteDance, Alibaba, and Tencent, while major model players such as DeepSeek, Kimi, and Zhipu, as well as internet giants like Baidu, JD.com, and Meituan, will enter niche tracks based on their respective strengths, creating a diverse and competitive industry landscape.
However, it’s important to recognize that AI is not a race where first-mover advantage guarantees victory. Instead, it’s a comprehensive contest requiring technological accumulation, scenario understanding, ecosystem integration, and a global perspective. All the major players have entered the field, but none can confidently claim that a single breakthrough will secure overall success.
The ultimate winner may not be the one making the most noise today, but rather the one that can endure cycles and continuously evolve. Just like a game of chess in the middle stages: the outcome is still undecided, and only those with patience and perseverance will have the last laugh.
