Over the past decade, Weibo has continually advanced in technological evolution, with Alibaba Cloud witnessing every step of this national-level application's journey.

"All apps deserve to be rebuilt with AI." Since the start of the year, major apps and various applications have accessed DeepSeek, reaffirming this assertion from two years ago. The era of intelligent app upgrades is now fully upon us.

From apps to intelligent apps, this transformation is a natural result of national-level applications adapting their business IT architecture and capacity reserves to the times, ensuring better user service. Whether in the mobile internet era or amidst the wave of intelligence, these large applications and national-level platforms have consistently pursued technological evolution and iterative exploration.

Take Weibo, a national-level application, as an example. Over the past decade, it has enriched the collective memory of Chinese people with diverse content forms and richer interactive methods. Behind the scenes, the company's IT architecture and technical support have evolved, encompassing cloud migration, intelligent exploration, and comprehensive large model application innovation.

This year marks the tenth anniversary of Alibaba Cloud and Weibo's collaboration. Their partnership has been a journey of mutual achievement and shaping. While Weibo records the public life memories of Chinese people, Alibaba Cloud has been a steadfast witness to Weibo's technological evolution.

01

Weibo Embraces Its Own Large Model

As numerous apps compete to announce their integration with the DeepSeek large model, many may not realize that Weibo's interactive comment robot is actually a product developed using a large model.

In late July 2023, an account named "Pinglun Wow Robot" (later changed to "Pinglun Robert") posted its first tweet. Initially, it did not attract much attention, with only a few users who had been replied to by it engaging in curious interactions. Many did not understand the technical principles behind it; they simply found the robot somewhat intelligent.

For instance, under the robot's first tweet, a user in Beijing asked, "Can you tell jokes?" The robot responded with a cold joke about a snail and a turtle, amusing a large group of people and attracting more comments and onlookers. Some even came to thank it for its reply. A user from Singapore commented, "I saw your comment today, thank you."

In addition to "Pinglun Robert," the "MBTI Xiaohangjia" account is another classic example of Weibo's large model application in interactive scenarios. It accurately calculates the personality type of the poster based on their already posted tweets. Due to its intriguing nature, this account quickly sparked an interaction boom with influencers and users after its launch, amassing a fan base of nearly one million.

Currently, a large number of apps are integrating large models, and companies are acting with urgency. Observers can sense their strong FOMO (Fear of Missing Out) mentality. Amidst the wave of intelligence, how to avoid using AI merely for AI's sake and find entry scenarios that align with user habits has become a major challenge for major application companies.

Based on product and data characteristics, as well as user habits, Weibo has developed its own "Zhiwei" large model. For example, Weibo users have unique expression styles, often preferring short, colloquial, and interesting phrases to express their opinions. "Pinglun Robert" is the most representative product trained by the "Zhiwei" large model using these data. It is frequently mentioned when discussing the AI-ization of various media content platforms.

The interaction between users and robots is visible to followers, creating a demonstration effect. Additionally, the content of the robot's comments may stimulate the discussion and sharing desires of bloggers and their fans. The application of large models can effectively activate the Weibo ecosystem and user atmosphere, enhancing user activity.

These new explorations are all centered around serving users and clients, leveraging the capabilities of new technologies, and combining users' needs in scenarios to explore and develop new products and interaction modes. AI seamlessly integrates into these scenarios without showcasing its prowess.

As a long-term partner of Weibo, Alibaba Cloud's public cloud and intelligent capabilities have provided crucial support for Weibo's intelligent exploration, helping it create the "Zhiwei" large model with unique Weibo characteristics.

02

From the Internet to the Mobile Internet: Weibo Embraces Cloud Computing

Weibo's evolution in the era of large models is a microcosm of the company's consistent open stance towards new technologies across different technological waves.

Before the wave of intelligence arrived, over a decade ago, people's content consumption habits and main carriers also transformed from the internet to the mobile internet. In that wave, Weibo was a pioneer. It aligned with users' content consumption needs and technological development trends, becoming an early adopter and practitioner of cutting-edge cloud computing technology.

If you are a long-time Weibo user, you might remember the saying, "Watching changes China." In those years, during significant events such as Weibo's anti-human trafficking campaign, the London Olympics, the Ya'an earthquake, and the disappearance of Malaysia Airlines Flight 370, many people felt the impact of internet products like Weibo on how Chinese people participate in public life.

It was also during those years that a new era began to emerge. In 2010, Steve Jobs released the iPhone 4, heralding the mobile internet era. Smartphones became a new platform for information acquisition, socializing, and entertainment after computers, significantly lowering the threshold for internet access.

On mobile phones, people could share their lives, exchange information, and follow trending topics anytime and anywhere. Social activities were no longer restricted by time and location, and the number of internet users surged during this period. At the end of June 2010, there were approximately 277 million mobile internet users in China, and by June 2015, this number had reached 594 million.

A plethora of well-known national-level applications, such as WeChat and Alipay, thrived during this wave. They gradually infiltrated all aspects of life, transforming people's content consumption, social, and payment habits.

Those who used to follow and engage with Weibo trending topics on computers now had an additional habit – "scrolling through Weibo." From sitting upright in front of a computer to holding a mobile phone and constantly "scrolling through Weibo," the sudden surge in users, unexpected trending topics, and their eagerness to stay updated during major events suddenly became a challenge for Weibo's IT architecture and technical framework as a national-level platform.

Weibo first experienced this challenge in 2015. That May, Fan Bingbing and Li Chen proactively announced their relationship on Weibo with the simple words "We," sparking a huge wave of public opinion with an astonishing 2 million reposts and comments in just one hour. The leveraged promotion by numerous brand merchants kept the heat fermenting.

The sudden influx of traffic caused Weibo's servers to crash for a while. The new technological wave, the consumption habits of the new generation of users, and the IT architecture of national-level products converged at a critical juncture.

It was also during this period that the nascent cloud computing industry was booming. Initially, a group of mobile internet application and game companies chose cloud computing, and many enterprises valued its benefits for development. Enterprises no longer needed to build complex development environments and deploy resources locally, reducing the cost of mobile internet entrepreneurship. It can be said that the support of this emerging industry facilitated the rapid spread of the mobile internet industry across China in just a few years.

For Weibo, it valued the characteristics of elastic computing, massive storage, and high availability, which were well-suited to the unpredictable traffic spikes of mobile users during sudden trending topics.

Cloud elasticity is highly beneficial for internet platforms like Weibo, which experience obvious peaks and troughs. Enterprises do not need to bear the high procurement cost of a large number of servers, and when traffic subsides, they can quickly switch, avoiding idle resources. As a result, Weibo and Alibaba Cloud hit it off, and Weibo became one of the first large applications to embrace cloud computing.

With the support of Alibaba Cloud, traffic spikes are no longer a significant challenge for Weibo. Even as users' consumption needs further evolve, and Weibo's content form transitions from images and text to videos, live broadcasts, etc., such as the Bai Baihe incident in 2017 that ignited public opinion with videos as the trending topic carrier, and the Weibo Night event featuring live broadcasts, traffic is no longer a burden for Weibo.

Simultaneously, with the development of Weibo's product form and the increase in business complexity, Weibo's IT technology architecture has also evolved from the earliest distributed platform architecture to an elastic hybrid cloud architecture, and then to an intelligent cloud-native architecture.

In recent years, the number of Weibo trending topics has remained a criterion for judging the popularity of an event, but trending topics causing Weibo crashes have become less common. This is not because netizens are no longer interested in celebrity gossip or social trending topics, but because the new IT infrastructure has become robust enough.

03

Towards AI: Embarking on the Next Decade

"When something major happens, go to Weibo" has become a user mindset. For audiences of various groups, Weibo has become the platform they turn to first during significant events. Simultaneously, you will find that on Weibo, groups from different fields and circles can find the in-depth and three-dimensional information they seek.

This also makes Weibo's user demographic very broad. The age group of the audience includes not only those focused on serious social news but also a large number of younger users. For example, according to data from Weibo's "2024 Weibo User Consumption Trend Report" released in December 2024, nearly 80% of Weibo users are post-90s and post-00s, with users from first-, second-, and third-tier cities accounting for more than 70%.

This is a group with considerable spending power and a wide range of interests. The topics of content consumption are diverse, and they also have relatively higher content aesthetics and taste. For instance, in multiple vertical fields such as sports events, games and animations, automobiles and digital products, fashion and beauty, vertical online social circles have obvious consumer demand for trending topics in related fields.

To serve these high-quality users well, in addition to optimizing channels, product operations, and other aspects, the platform also needs to invest in the technical framework to ensure user activity and stickiness. This is a transformation at the level of the content distribution mechanism. Mainstream internet and mobile internet platforms have been undergoing this process over the past few years. The cooperation between Alibaba Cloud and Weibo has also continued in this area.

Looking back over the past decade, with the development of the times and technological progress, Weibo has consistently used in-depth and three-dimensional content in various forms to serve the platform's users and record the public life of Chinese people. In this process, Alibaba Cloud's cloud computing technology and AI capabilities, as one of the underlying infrastructures, have supported the evolution of Weibo's product form.

Interestingly, both companies were founded in 2009. The former is a national-level product spanning the internet and mobile internet, while the latter has become a leading IT infrastructure provider supporting these enterprises. They have jointly witnessed the most glorious mobile internet era. Their cooperation is a process of mutual achievement and shaping, as well as a two-way journey.

On one hand, Alibaba Cloud supports Weibo in multiple dimensions, such as managing traffic peaks and troughs, enhancing user stickiness, and exploring large models amidst the wave of intelligence.

On the other hand, with large national-level applications like Weibo migrating to the cloud, the cloud computing industry has also transitioned from small and medium-sized users utilizing the cloud to large enterprises deeply integrating the cloud. Nowadays, migrating to the cloud is a widely accepted practice, and enterprises are beginning to evolve towards deeply integrating the cloud, pursuing IT-driven business innovation and leapfrog development. Simultaneously, cloud computing players are also moving towards the era of cloud + AI.

Since last year, generative AI has accelerated its transition from general-purpose technology to industry applications, and the demand for model invocation in different industries and application scenarios has grown rapidly. In addition to basic scenario exploration, many enterprises have started attempting to reconstruct their workflows in Agent mode, and the application of large models is evolving from single-point breakthroughs to systematic applications.

Since the turn of the year, domestic industry leaders have enthusiastically adopted open-source strategies, propelling AI applications into a state of explosive growth akin to a prairie fire. Projections indicate that by 2025, a staggering 500 million AI applications will have emerged. Industry experts concur that on the consumer side, large models will significantly enhance application experiences, while in the ToB realm, they promise to revolutionize efficiency.

To facilitate the seamless integration of large models within enterprise settings, industry insiders widely agree that businesses must undertake comprehensive enhancements, encompassing application architecture design, model service platform capabilities, and infrastructure support. These efforts will lay the groundwork for unlocking substantial business and societal value from large models.

Amidst this wave of innovation, the collaboration between Alibaba Cloud and Weibo stands out as visionary and pioneering. The two entities have engaged in profound cooperation at the infrastructural and architectural levels, paving the way for the successful deployment of large models. As AI continues to redefine content creation, consumption, and interaction paradigms, transforming the content ecosystem and intelligent application scenarios, this partnership promises to unleash a myriad of new possibilities.

Alibaba Cloud stands as a testament to the ongoing evolution of Weibo, a national social media platform poised for continued growth as a new decade dawns.