Alibaba will invest heavily in the AI C-end market

After September, over a hundred core engineers transferred from Beijing, Guangdong and other places gathered at the C4 floor of Alibaba’s Hangzhou Xixi Park. Afterwards, two floors of the C4 building were closed off, and employees needed to swipe their work badges and obtain special security approval before entering.

More than a hundred core engineers are involved in Alibaba’s secret project, the “Qianwen APP,” aimed at C-end users. Tiger Sniff learned that the project was approved by the top management of the group and composed of core engineers from business lines such as Quark and Alibaba Cloud. Within Alibaba, this is a strategic level project aimed at entering the C-end model market.

Abroad, the most well-known market for this is OpenAI’s ChatGPT series, while in China, from the second half of 2024 to the first half of 2025, there will be fierce competition between Doubao, Yuanbao, and DeepSeek.

Add weight to C-end

Within Alibaba, throughout 2024, Quark will be the most important product within the group for the C-end model market and AI search portal market. At the same time, Alibaba Cloud’s generic Qwen model did not directly enter the C-end market, but focused more on the download volume and usage in the open source community (professional developers in the industry).

Tiger Sniff exclusively learned that Alibaba’s internal assessment criteria for the basic model team are guided by model influence, including the total number of derivative models and popularity in the open source community and developer community (download volume, discussion heat, etc.). It is worth noting that as of now, the Qwen model has become the top model in the global open source community in terms of download volume and the number of derivative models.

In the current Alibaba system, the basic model Qwen team belongs to Alibaba Cloud, while the C-end product belongs to Quark. And both product teams report directly to Alibaba CEO Wu Yongming at the highest level. Tiger Sniff learned that the main team of the “Qianwen APP” is composed of Quark, but the core basic model of the product is based on Qwen.

From a collaborative perspective, the senior management of the group has promoted the participation of talented personnel from multiple departments. As of now, the “Qianwen APP” is operating as a project team, and there may be new team system adjustments after the product is officially launched. Tiger Sniff learned that the current division of labor in cooperation is basically similar to the operation mode of the Tongyi APP under the jurisdiction of the Shuzhi Group (where Quark also belongs).

According to another source, since the fourth quarter of 2024, Alibaba’s senior management has visited scholars in the fields of large models and multimodality at home and abroad, and successfully invited several top scientists from around the world to join the team. And some of them are also involved in the development of the “Qianwen APP”.

From multiple sources, Tiger Sniff has learned that the popularity of DeepSeek in the C-end market at the beginning of 2025 has had some impact on Alibaba’s internal basic model capabilities and product ideas. In terms of capabilities, the team has further strengthened its resources invested in reasoning and multimodal abilities. On the product side, the team and senior management have begun to seriously consider the broad space for “entering the C-end AI product market”.

The upcoming “Qianwen APP” is mainly based on the Qwen model, but covers some of the capabilities of quarks (such as multimodal recognition, embedding of some functional modules, etc.). It can be said that the “Qianwen APP” is more like a “heavy composite brigade” deployed by Alibaba, which has gathered the top forces of various AI C-end capabilities within the group. Its tactical goal is to gain a sufficiently broad market share in the C-end market and improve Alibaba’s mindset in the AI C-end market.

After the news was released, Alibaba’s stock price benefited significantly, with Hong Kong stocks rising nearly 6% in the afternoon of the day and closing with an increase of over 3%.

You may need to re evaluate the value of Alibaba

After this news was released, my first reaction was that the outside world may reassess the long-term value of Alibaba.

As of October 2025, OpenAI’s latest valuation has exceeded $500 billion. However, it is worth noting that OpenAI actually exists based on the “OpenAI Microsoft Nvidia” alliance. In this mode, Microsoft and Nvidia are actually playing the “ammunition supplier” of OpenAI. OpenAI itself does not have the infrastructure and necessary capabilities of “production computing”.

According to the latest collaboration framework, OpenAI plans to deploy 20GW of computing power in the next decade, and will invest over $1 trillion in real money around this framework. The perspective provided by this series of information is: 1. OpenAI’s biggest pain point is computing power. 2. OpenAI relies on a platform that includes Microsoft, Nvidia AMD、 The American AI ecosystem, including “FAANG” forces such as Oracle.

In contrast, Alibaba’s competitive advantage is that it is closed-loop.

It has the ability to produce computing power and infrastructure. According to Wu Yongming’s announcement, Alibaba is preparing to continue investing 380 billion yuan in AI infrastructure. If the “Qianwen APP” successfully continues to expand its market share in the C-end market, then when we observe this product, the closed loop of computing resources behind it must be a key element that cannot be ignored.

Another invisible dimension is commercialization. As early as 2023, OpenAI began experimenting with e-commerce monetization, but this path has yet to be fully realized. For Alibaba, there is a “broad” imagination space: if they have an AI product that is at the “mental entrance” level in the C-end market, it will bring huge traffic entry effects and support to their e-commerce.

The above two dimensions are actually global variables that may arise from the move of “Qianwen APP”.

However, the challenge facing the “Qianwen APP” is enormous: if we refer to the path of the rise of two friendly merchants, Doubao and Yuanbao, in 2024 and 2025, both of them rely on the existing huge traffic pools within the ecosystem. If the “Qianwen APP” wants to rise to prominence, traffic must be a key test question.

And according to DeepSeek’s plan, their latest model is also about to be released, so the “Qianwen APP” may compete head-on with DeepSeek in terms of time cycle, which will be a fierce product and mental war.

But the story doesn’t end here.

We need to see Alibaba’s urgent need for C-end AI mindset: almost at the same time as the news of the “Qianwen APP” appeared, Alibaba’s other AI C-end entrance product – Quark AI Glasses – is also about to be officially launched. From soft to hard, Alibaba clearly hopes to open up some opportunities in the C-end market within the year. Considering that March 31st is the milestone of Alibaba’s new fiscal year, we have reason to draw two conclusions:

·Alibaba hopes to verify the actual effectiveness of both software and hardware in the C-end AI market by March 31, 2026, and plan the overall strategy for the new fiscal year based on this;

·Alibaba may invest resources in these two C-end AI mental entry points in a saturated manner, similar to its support for Taobao flash purchase business this year.

From this perspective, the next two to three months will be crucial for Alibaba.