GuShiio.com Wen| photon planet
Whenever there is a big event, what Ali needs is not necessarily calm, but there is a high probability that it will be Jack Ma.
Late at night on February 11, news spread that Apple, together with Ali, would jointly develop localized AI functions for China iPhone users. Earlier, the appearance of Alibaba founder Jack Ma in Ali’s Hangzhou Park also increased the credibility of this news to a certain extent. He inspected Area C of Ali Xixi Park, where Quark and other AI to C businesses are located.
As early as early as last year, since Apple CEO Cook visited China, Apple Intelligence’s cooperation has triggered long-term discussions in the industry. At that time, it had long become a consensus that under the considerations of data privacy and localization, Cook was only one of the few vendors that could provide complete AI+ cloud services, and Ali was one of them.
Although it is still a little short, the market has responded strongly to this, especially Alibaba (NYSE:BABA)’s share price has fluctuated sharply in the past two days.
After all, this is the alliance of two frustrated middle-aged people who
Regardless of whether Apple Intelligence can become a lifeline for Apple’s sales in China, if the cooperation can succeed, it means that Alibaba, a big ship, puts users first, and the AI-driven strategic transformation will eventually reap the fruits.
Get the second ticket?
Perhaps it is because things are troublesome. Since Apple began to screen China AI partners, there have been only constant rumors for more than a year, and there has been no news of actual implementation.
For Apple, whose sales in Greater China continue to decline, it is very important to quickly finalize cooperation with domestic AI manufacturers and speed up the implementation of AI functions. Putting aside the lingering aftermath of TikTok, Tencent’s AI progress is relatively slow. As for Baidu, which Cook intended to flop last year, there are reports that the two sides have differences on technical and privacy issues.
Ali, whom he approached, seemed to be Apple’s only choice until the recent appearance of DeepSeek.
“The truth that flowers bloom outside the wall and the fragrance inside the wall has been once again verified by DeepSeek. Under the tap water of major overseas media and KOL, domestic users have quickly accepted the recognition that DeepSeek represents the forefront of AI technology, and its user base has also grown rapidly.
But it is not enough to justify Apple ‘s flop. After all, Apple’s contact and evaluation with domestic AI companies far precedes the time when DeepSeek really emerged. Secondly, as one of the few options, Ali is far superior to DeepSeek in terms of major customer service capabilities. Finally, compliance as a core element of implementation also requires a risk control system that understands domestic characteristics.
In terms of model capabilities, the Qwen2.5-Max released by Ali during the Spring Festival has been well matched with DeepSeek V3 in various major lists. Moreover, DeepSeek is still on the way to train multimodal capabilities such as graphics and video, and it is somewhat difficult to require it to implement in smartphone scenarios that emphasize multimodal interactions.
More importantly, for AI service companies, the base model is a face, and most of the time it is used to score points. What Apple wants is products that can be used and supporting service output.
From the perspective of service capabilities, DeepSeek’s hundreds of people’s team size and lack of cloud service capabilities are difficult to ensure the localized adaptation of Apple’s AI and the concurrent needs of a large number of users. On the other hand, Ali has provided technical support for large-scale events such as the Paris Olympics and Spring Festival Gala many times in the past year. This may be an important reason why Apple has made up its mind.
According to QuestMobile data, as early as the second quarter of 2023, domestic iPhone users reached 250 million. Looking back at the grand occasion of the entire network when Siri came out many years ago, it is not difficult to imagine that the launch of the Chinese version of Apple Intelligence will be a high-concurrency scenario of hundreds of millions, and this is the unique skill of Ali, who started its business as an e-commerce company.
At the Paris Olympics, Alibaba Cloud’s technical support allowed cloud computing to surpass satellites and become the mainstream broadcast method of the Olympics for the first time. More than two-thirds of the signals were distributed to the world based on Alibaba Cloud; last year, Alibaba Cloud provided more than 1 million core CPU resource support for Tmall Double 11, setting a new record for the scale of its elastic scheduling resources.
Last year, Apple, which had not cooperated well with Baidu, successively contacted leading domestic manufacturers such as Byte, Tencent, and Alibaba. Judging from this time point, perhaps many of the large-scale activities Alibaba Cloud has supported in recent months are stress tests it conducted for Cook, an interviewer.
When cloud computing was in the ascendant, Ali, which was able to withstand the peak of Double 11 traffic, proved the potential of cloud computing to the market. The business model of public cloud was thus realized, and it also won the first public cloud ticket.
At a time when cloud computing has become the first entry point for AI applications, Ali’s ticket to the next era seems to fall on its ability to handle high-concurrency scenarios.
Uncertain shouts
One of the gray lines is that looking back on Ali’s earlier changes in the AI strategic layout, its cooperation with Apple seems to have long been harboured.
At the beginning of this year, the UALink Alliance, an AI server chip interconnection organization that aims to break Nvidia’s monopoly on computing power, announced the latest board of directors. Alibaba Cloud was successfully selected as the only China company. This means that it has a certain degree of ecological synergy with Apple, which has long become a director.
More obvious changes have occurred internally. For example, it was originally used as an exhibit on the shelves. The Tongyi to C application originally packaged by Alibaba to B services has been split into Alibaba Intelligent Information Business Group with Alibaba Cloud. It is the isolation of toB services and toC products.
Judging from the currently open Apple Intelligence product form, its cooperation model with third-party AI is cloud call rather than end-side deployment. Tongyi App itself cannot reap any benefits from this cooperation. More importantly, C-end users do not care about infrastructure. They only care about the possibility of cross-platform AI applications with good services. This possibility represents the bargaining power of Ali and Apple.
Take WeChat as an example. As the largest super App in China, it has engaged in many rounds of games with Apple in terms of public account rewards, Mini programs, etc. due to Apple tax issues. In Apple’s closed-loop ecosystem, this is true for app stores that are almost exclusively used for distribution, let alone model service providers that Apple hides under AppleIntelligence services?
To a certain extent, Apple’s core need for China’s AI partners is to make them Guizhou on the cloud in the AI era.
On the other hand, cooperative service providers, whether it is Baidu, which was rumored earlier, or Alibaba today, face a buyer’s market. Posting an apple eaten by Turing on a customer’s LOGO wall is undoubtedly a golden sign, but the price is that their own models need to be hidden under Apple Intelligence. Not only can they not obtain the traffic they lack most, they also face competition from mature manufacturers such as Baidu and Tencent.
The core reason for the conflict between Baidu and Apple mentioned above is the issue of user data usage and ownership in Apple’s closed-source system, that is, Baidu cannot borrow relevant data from iOS users for training its own models. Interestingly, the time when Ali began to comprehensively speed up the development of AI to C applications was almost synchronized with this accident.
Furthermore, after reorganizing the to C business, Ali’s cooperation with Apple has actually raised new path questions.
The first thing to bear the brunt is whether Tongyi Application and Quark, two products that represent Ali AI’s implementation in C, will choose to follow the old path of bean buns and Kimi, and quickly explode through input and new products. In Ali ‘s own organizational structure, there is bound to be resource tilt for different applications. We have previously learned that the budget resources of the Alibaba Intelligent Information Business Group are relatively limited. For now, Quark and Tong Yi are attacking two scenarios: mobile and PC respectively, and there is no fear of horse racing after meeting up for the time being.
Since I am determined to do a good job in toC, I have to mention the developer community that has been working with Tongyi for a long time.
Some people believe that the gap between traditional to C and to B is bridged by the atomization and lightweight characteristics of AI applications. MODA itself has already accumulated a large number of individual independent developers and a community base represented by lightweight applications. If the community is fully opened to C-end users and entered the mobile end, this may not be the next application square in the era of big models.
“Stand in line moment
The explosion of DeepSeek is regarded by some people as a new GPT moment. In recent times, investment and financing around the AI frontier has become hot again.
Whether Masayoshi Son entered OpenAI or Musk stole Ultraman, the investment and financing surrounding the creation of a company is nothing more than dividing camps. It is like not accepting investment so far, but the external valuation is between US$1 billion and US$150 billion. Magic square that fluctuates around US$150 billion. A similar situation may appear in different markets after Ali and Apple reach a cooperation.
The first person to face the demand for side selection is Ali. In the AI era, the business model of cloud computing relying on computing power has not changed fundamentally. Lightweight applications developed by large customers like Apple and small and medium-sized developers use the cloud as the first entry point. Ali’s takeover of Apple as a major customer is undoubtedly a landmark event in the commercialization of the big model, which may lead to a new round of controversy over the growth path.
As early as a few years ago, when the growth of the cloud computing industry slowed down, there was a path struggle between KA and small and medium-sized enterprises in the industry. This even gave rise to the fact that major cloud manufacturers continued to reduce the weight of profit margins in order to compete for project-based orders under share competition. At that time, Alibaba once made a major sales organization to adapt to the pace of KA’s customers, and also introduced Cai Yinghua, former president of Huawei’s corporate business China. However, it lost a lot of share due to multiple reasons such as the Xinchuang environment and the rise of operator cloud.
Alibaba’s 2025 fiscal year semi-annual report shows that as of September 30, 2024, Alibaba Cloud Intelligent Group’s revenue was 29.610 billion yuan, a year-on-year increase of 7%. As model calls continue to reduce prices, major customers are likely to become important support for management’s high-profile announcement that cloud business growth will return to double digits in the second half of 2024.
In addition, the choice of choosing sides is also placed in front of the extremely sophisticated smartphone manufacturers.
From a user’s perspective, the model capabilities of cloud access are not important in today’s daily use. In contrast, what they need more is sufficiently intelligent end-side capabilities. Currently, search maps and AI operations open to domestic smartphones are only relatively basic interactions and applications.
Recently, Glory, vivo, OPPO, and Huawei have announced that their mobile smartphones will be connected to DeepSeek. If it weren’t for Apple, maybe each family would maintain a similar pace, focusing on self-developed models, and choose a day to fight again after adding money to DeepSeek’s capabilities. However, after Apple teamed up with AliTongyi, the situation may change the data assets accumulated in the e-commerce business to be biased towards user behavior analysis, which is precisely the most scarce ability of mobile phone manufacturers ‘end-side AI.
Companies need to take sides, but technology does not. Tongyi, which was born in Dharma Institute, has had the temperament of a man of science and engineering since its birth. We do not need to worry about its strategic determination. The Qwen series, which has long been at the forefront of major lists, can prove this. But what needs to be a question mark is whether Alibaba can concentrate its resources to develop new AI businesses and products after Apple’s cooperation.
Perhaps this is the main reason why Jack Ma appeared in Xixi Park.
