Alibaba is entering the race for the AI super traffic gateway.

Author | Huang Yu

Editor | Wang Xiaojuan

During the global AI arms race over the past three years, Alibaba has focused primarily on AI infrastructure for B-side customers, investing little in AI-native applications for consumers. Goldman Sachs even put forward the view that “for China’s AI industry, Alibaba leads in infrastructure, Tencent leads in applications.”

However, Alibaba is not about to miss the chance to compete for the next super traffic gateway, and as 2025 approaches its finale, has dropped another bombshell on the industry.

Four days after news broke of Alibaba launching its "Qianwen" project, on November 17, Alibaba officially announced the public beta of the Qianwen APP, stating that it aims to “leverage the world’s top-performing open-source model Qwen3, introduce free services, and integrate various life scenario ecosystems, to comprehensively compete with ChatGPT.” 

Clearly, as the capabilities of the Qianwen Qwen model continue to improve, Alibaba believes the time has come to enter the front lines of AI-native application competition.

The Qianwen APP was formerly Alibaba’s first AI-native application released in September 2023—Tongyi Qianwen APP (later renamed Tongyi APP). This “fresh appearance” signals Alibaba’s ambition to fully enter the AI to C market and, to some extent, hints at their dissatisfaction with the APP’s previous monthly active users of just over 3 million.

The Qianwen APP has now taken center stage in Alibaba’s AI strategy and is the vanguard for grabbing the next super traffic gateway. Alibaba’s core management even views the “Qianwen” project as “the future battle of the AI era.”

However, this battle will not be easy. The competition among AI-native applications has been ongoing for more than two years, and this year has entered a fever pitch. As things stand, ChatGPT leads globally with 700 million monthly active users, while Doubao and DeepSeek hold temporary leads in the domestic market with over 100 million monthly actives each. 

Tech giants are zeroing in with full firepower on the super gateway of the AI era, and the launch of Qianwen will undoubtedly intensify the competition. Whether Alibaba, a latecomer, can use Qianwen to achieve a breakthrough and reshape the landscape of AI-native applications remains highly uncertain. 

Benchmarking ChatGPT

This year, Alibaba has shown remarkable fighting spirit. The "Qianwen" project, following AI infrastructure and Taobao Instant Purchase, is another high-profile, group-level strategic initiative announced for 2025.

As early as November 13, market sources reported that Alibaba had secretly launched a strategic-level project named “Qianwen,” using the Qwen’s strongest model as the core to build a personal AI assistant—Qianwen APP—with a direct target on ChatGPT.

On that day, Alibaba’s stock price soared, with gains reaching up to 6%.

Now, the dust has settled. According to Wallstreetcn, the Qianwen APP quietly appeared in major app stores on November 16.

To take on ChatGPT, Alibaba has also announced a global version of the Qianwen APP will be launched soon, seeking to leverage Qwen’s overseas influence to compete for international users.

Alibaba states that the current release of the Qianwen APP is an initial version, aiming to use the most advanced model to create a “chat-friendly and capable personal AI assistant.”

Beyond smart chatting, being “capable” will be a major focus for Qianwen APP’s future development, especially in everyday life areas where the APP has already shown strong potential.

According to insiders speaking to Wallstreetcn, Alibaba is planning to integrate maps, food delivery, ticketing, office, learning, shopping, health, and other life scenarios into Qianwen APP, thereby greatly enhancing its practical abilities.

Alibaba dares to challenge ChatGPT directly because it holds two trump cards: complete free access and globally recognized underlying technical capability.

It’s worth noting that ChatGPT, as the “top-tier,” now offers both free and paid subscription plans. The free version is open to all but has limited messages, response speed, and functionality. Paid plans are priced in US dollars, supporting monthly and annual billing, with the Plus plan costing $20 per month, clearly a barrier to user growth.

Alibaba’s investment in AI infrastructure is widely seen. Building on its February announcement of a three-year, 380 billion RMB AI infrastructure plan, Alibaba CEO Wu Yongming doubled down at September’s Yunqi Conference, promising even larger investments. He also claimed that by 2032, Alibaba Cloud’s global data center energy consumption will be ten times that of 2022.

At the model level, Alibaba's open-source Qwen model has now achieved global impact.

Stanford University’s AI Research Institute “2025 AI Index Report” indicates that the performance gap between top AI models in China and the US has shrunk dramatically to just 0.3%, almost vanishing, and Alibaba’s Qwen contributed the third highest impact globally. 

Furthermore, NVIDIA CEO Jensen Huang revealed at the 2025 GTC conference that Qwen has the largest and still growing global market share among open-source models.

Qwen has become the first choice for many developers and enterprise markets. According to a Sullivan report from September, Alibaba Tongyi ranks first in China’s enterprise-level large model invocation market.

At present, Alibaba Qianwen Qwen has open-sourced over 300 models, with downloads across major global model communities exceeding 600 million, and 170,000 derivative models—surpassing Meta’s Llama to become the world’s largest open-source model family.

It is evident that Alibaba’s external messaging now consistently uses the name Qianwen Qwen; renaming Tongyi APP to Qianwen APP may also help global users more easily connect the app with the Qwen model.

AI Assistant Free-for-all

In this AI large model craze, Alibaba’s release of its first AI-native app actually wasn’t late; it had launched Tongyi Qianwen APP two years ago, at a time when only Baidu and ByteDance among internet giants had launched AI-native apps. Tencent launched Yuanbao last May, and DeepSeek emerged at the beginning of this year.

After launching, Tongyi Qianwen APP’s performance was lukewarm, with monthly active users trailing well behind the market leaders.

QuestMobile data shows that the leading AI-native app landscape is relatively stable. In September, Doubao surpassed DeepSeek with 172 million and 145 million monthly active users, respectively; Tencent's Yuanbao ranked third with 32.86 million, and Tongyi Qianwen APP ranked tenth with 3.06 million monthly users.

Compared with Tongyi Qianwen APP, internally within Alibaba, Quark was actually the product they hoped to develop as the super gateway of the AI era. In March this year, Quark was upgraded to Alibaba’s flagship AI app, becoming the star product for "AI to C."

According to Wallstreetcn, as AI capabilities improve, Alibaba has decided that dialogue-based AI assistants are a better approach, and going forward will focus on developing Qianwen APP and integrating it into Quark.

Quark’s product positioning is both AI search and AI browser; like Taotian, 1688, Gaode, and overseas e-commerce, it is an AI transformation of existing business. Qianwen APP, on the other hand, is now Alibaba’s only native AI application.

Currently, both Qianwen APP and Quark fall under the Alibaba Intelligent Information Business Group, managed by group president Wu Jia.

The Qianwen team has revealed that for the Qianwen APP, product and algorithm preparations have been ongoing, with more frequent core management discussions around the summer of 2025, and the most intensive company-wide deliberations happening before National Day.

Ultimately, it was Wu Yongming who made the judgment and decision—Alibaba must have an AI-native, C-side super gateway.

Alibaba’s management believes that the development of AI will go through three stages: “learning humans,” “assisting humans,” and “surpassing humans.” Currently, the capabilities of large models have entered the “assistant” Agentic AI era, so the timing for Alibaba to forcefully enter the consumer market is ripe.

Alibaba is making significant investments in AI-native applications. Reportedly, after September, Alibaba mobilized over a hundred core engineers, setting aside two floors at its Hangzhou headquarters as a dedicated base for the “Qianwen” project.

However, Alibaba faces not only a slew of powerful competitors but also the increasingly challenging industry growth situation.

QuestMobile data shows that as of March 2025, DeepSeek topped with 194 million monthly actives, followed by Doubao and Yuanbao with 116 million and 41.64 million, respectively. It’s apparent that by September, the leading AI-native apps hadn’t seen a significant rise in monthly activities.

Moreover, as of September 2025, mobile AI-native APPs, In-APP AI, and smartphone manufacturer AI assistants had overall scales of 287 million, 706 million, and 535 million, with quarterly compound growth rates of 3.4%, 9.3%, and 1.2%, respectively, indicating clear growth bottlenecks.

Currently, as giants compete for the “general platform,” some vertical-track AI-native apps are instead performing well. For example, AQ, Xiaoyunque, and Xinghui had quarterly compound growth rates of 83.4%, 90.8%, and 33.3%, respectively.

For years, Alibaba has been questioned for its inability to create C-end blockbuster apps outside e-commerce, with almost all C-side traffic entrances taken by WeChat and Douyin. With the AI large model opportunity, Alibaba has rekindled its consumer-side ambitions, welcoming another chance to break through, but with Qianwen APP as its general AI assistant, the challenge to become a super AI application remains formidable. 

Risk Warning and DisclaimerThe market contains risks; investments should be made cautiously. This article does not constitute personal investment advice, nor does it take into account any individual user's specific investment goals, financial circumstances, or needs. Users should consider whether any opinions, views, or conclusions in this article suit their particular situation. All investment decisions made accordingly are at the user’s own risk.