Wan Weixing, Head of Qualcomm's AI Product Technology in China: The Key to the Next-Generation AI Experience | 2025 AI Partner Conference

2025 is the first year of the explosion of AI applications. As the global AI competition enters the "China Moment," a profound technological transformation is quietly rewriting the industrial landscape. At this critical juncture, the industry faces core propositions: How to bridge the gap between AI technology and large - scale applications? Where will the next disruptive AI super - application be born?

On April 18th, the 2025 AI Partner Conference hosted by 36Kr grandly opened at the Shanghai Modu Space. The theme of this conference is "The Super App is Coming," focusing on the disruptive transformation of AI applications in all industries. The conference is divided into two major chapters: "The Super App is Coming" and "Who is the Next Super App," covering seven major topics such as "Growing Up in the AI World" and "In 2025, Focus on AI Super Applications." It includes more than 10 keynote speeches, 3 round - table discussions, and two sessions for releasing the lists of excellent AI case enterprises, deeply analyzing how AI technology reconstructs business logic and reshapes the industrial landscape, and exploring the infinite possibilities brought by AI super - applications.

On that day, Wan Weixing, the person - in - charge of Qualcomm's AI product technology in the Chinese market, gave a keynote speech titled "The Key to the Next - Generation AI Experience."

The following is the content of Wan Weixing's speech, edited by 36Kr:

Dear guests and friends online, good morning! I'm Wan Weixing from Qualcomm Technologies, Inc. I'm very honored to discuss with you the core elements of the next - generation AI experience. Nowadays, AI has been deeply integrated into daily life, triggering a revolutionary wave of user experience on various terminal devices.

Wan Weixing, the person - in - charge of Qualcomm's AI product technology in the Chinese market

In the field of smartphones, AI portrait processing and landscape optimization functions enable ordinary users to take professional - level photos easily. The voice assistants of mainstream mobile phone brands support natural language interaction. Whether it's system settings, APP operations, or even complex multi - task instructions, users can easily complete them just by speaking. In the scenario of intelligent vehicles, it has become a standard feature for drivers to communicate smoothly with the in - vehicle system through voice. In the field of AI PCs (Artificial Intelligence Personal Computers), AI functions such as Copilot significantly improve work efficiency - from generating meeting minutes, organizing materials, to making PPTs and editing videos, complex tasks can be completed with one click. After XR glasses are equipped with multi - modal models, they can analyze the information within the user's field of vision in real - time and provide accurate personalized recommendations.

Behind these innovative applications, the operation of edge - side AI is playing a key role. Compared with cloud - side AI, edge - side AI has significant advantages:

First, it provides better privacy and security protection. As the most frequently used personal device, the mobile phone stores sensitive data such as photos, documents, and chat records. Users often don't want to upload them to the cloud. Edge - side AI can efficiently complete data reasoning and processing locally, reducing the risk of privacy leakage.

Second, it offers a better personalized experience. The usage habit data generated daily by devices such as mobile phones and PCs can be captured and analyzed in real - time by edge - side AI, providing more flexible and efficient customized services based on user preferences. In addition, edge - side AI also has multiple advantages in terms of performance, energy efficiency, response speed, practicality, and cost control. It can quickly respond to user needs in cooperation with the cloud.

In terms of interaction methods, AI has brought revolutionary breakthroughs. From the early mechanical keyboard and mouse operations to today's multi - modal data input (voice, text, image, sensor signals, etc.), AI is redefining the human - computer interaction interface. We believe that AI is the new - generation UI (User Interface) - users input instructions through multi - modal data, and the edge - side intelligent agent can analyze and understand the intentions in real - time, complete task distribution and planning, and finally call various services such as entertainment, e - commerce, and navigation.

To achieve this end - to - end full - link, Qualcomm has carried out a systematic technical layout in three major aspects: hardware, software, and ecosystem:

At the hardware level, Qualcomm provides developers with powerful AI acceleration capabilities through customized system - level IP design and heterogeneous computing architecture. Among them, the CPU is responsible for processing tasks with high timeliness requirements; the GPU is good at pixel - level rendering of images and videos; the NPU (Neural Processing Unit), with its high - performance and low - power consumption characteristics, is particularly suitable for running continuous and ubiquitous tasks such as large - language models and multi - modal models; the sensor hub is designed for low - power, all - day - running applications such as environmental perception, voice wake - up, and real - time monitoring.

At the software level, for diverse product lines such as mobile phones, automotive cockpits, AI PCs, and IOT, Qualcomm has created a unified AI software stack, which is fully compatible with mainstream development frameworks such as Onyx, TensorFlow, and PyTorch, as well as open - source inference engines such as Onyx and TensorFlow Lite. In addition, Qualcomm's self - developed AI EngineDirect SDK can directly call hardware modules, further accelerating the development and deployment of AI applications.

At the ecosystem and tools level, Qualcomm launched the Qualcomm AI Hub, allowing developers to download optimized models, enabling them to easily and quickly integrate pre - optimized AI models into applications, thereby creating a faster, more reliable, and more privacy - friendly user experience.

The edge - side intelligent agent planner is the core of Qualcomm's promotion of the next - generation AI experience. Aiming at an end - to - end solution, from natural language understanding and intention recognition to task orchestration and local tool invocation, it uses local device data (such as personal knowledge graphs and usage habit records) throughout the process to make task planning more accurate. Subsequently, the planner also supports diverse services such as music playback, weather queries, navigation, and ordering food, and realizes cross - application linkage.

Take the birthday scenario as an example: After the user discusses birthday arrangements with their mother, the edge - side intelligent agent can automatically analyze local data such as chat records and calendar schedules, and quickly generate birthday suggestions, including the time and location for restaurant reservations. Moreover, the intelligent agent can actively identify user needs - if it detects that the user has a birthday plan in the near future, it will actively push recommended plans; during the itinerary execution stage, the intelligent agent can link applications such as the calendar, navigation, and catering, adjust the reservation time according to real - time traffic conditions, or recommend ordering flowers from a nearby flower shop based on the user's and their family's preferences, making the entire birthday arrangement more intelligent and warm.

In the future, edge - side AI will continue to break boundaries, bringing users a safer, smarter, and more personalized experience. Qualcomm will also continue to deepen technological innovation and work hand in hand with ecosystem partners to promote the in - depth implementation of AI in various fields.

That's all for my sharing. Thank you for listening!