Core Viewpoint - The emergence of AI smartphones, particularly represented by ByteDance's "Doubao Phone" nubia M153, signifies a significant shift in the mobile internet landscape, contrasting with the API standardization approach taken by Apple and Google, leading to a redefined relationship among smartphone manufacturers, app developers, and users [5][7][12]. Group 1: AI Smartphone Development - On December 19, ByteDance announced collaborations with hardware manufacturers like vivo, Lenovo, and Transsion to advance AI smartphone technology following the launch of the "Doubao Phone" [5][6]. - The "Doubao Phone" utilizes GUI technology to enable AI assistants to perform complex tasks across applications, which has prompted a defensive response from mainstream apps [5][10]. - The core technology of the "Doubao Phone" is based on a deep integration of GUI and system-level permissions, allowing AI to execute tasks like ordering food and comparing prices seamlessly [7][10]. Group 2: Technical Route Comparison - The competition between the GUI paradigm represented by Doubao and the API paradigm led by Apple and Google highlights a fundamental divergence in AI smartphone strategies [12][13]. - The GUI approach allows for broader application compatibility without requiring developer cooperation, while the API approach emphasizes stability, privacy, and efficiency but relies on app developers to integrate their functionalities [12][13]. - Apple and Google are currently focusing on API development, which is seen as a more conservative and slower approach compared to the rapid advancements in GUI technology [17][20]. Group 3: Market Implications - According to Canalys, the global AI smartphone shipment share is expected to rise from 16% in 2024 to 54% by 2028, with a compound annual growth rate of 63% from 2023 to 2028, driven by major players like Samsung and Apple [20][21]. - The introduction of AI assistants in smartphones raises concerns among major app developers about potential disruptions to their business models, as AI could bypass traditional app functionalities [21][22]. - The future ecosystem of AI smartphones is anticipated to evolve into a "layered governance" structure, where different players will have varying degrees of influence and control over AI operations [22][23].

Core Viewpoint - The emergence of AI smartphones has created a clear division in technology routes, with ByteDance's "Doubao Phone" leading a GUI (Graphical User Interface) approach, while Apple and Google maintain a more conservative API (Application Programming Interface) standardization route, resulting in a significant shift in the mobile internet landscape [1][4][8]. Group 1: AI Smartphone Development - ByteDance, in collaboration with ZTE Nubia, launched the "Doubao Phone" nubia M153, which allows AI assistants to perform complex tasks across applications using GUI technology [1][4]. - The Doubao Phone's AI assistant can execute tasks like ordering food and comparing prices through voice commands, showcasing its cross-application capabilities [4][6]. - The technology behind Doubao Phone involves deep integration with system-level permissions, enabling the AI to simulate user interactions with various apps [6][7]. Group 2: Competitive Landscape - The competition between the GUI approach of Doubao and the API approach of Apple and Google highlights a fundamental clash in the mobile internet's business logic and interests [4][8]. - Apple's API approach focuses on building a standardized framework for developers to integrate AI capabilities, which is seen as more stable and privacy-conscious but requires cooperation from app developers [9][13]. - Google is also pursuing an API strategy, emphasizing cloud collaboration and prioritizing desktop applications, while still in the early stages of mobile GUI implementation [13][14]. Group 3: Market Trends and Projections - According to Canalys, the global AI smartphone shipment share is expected to rise from 16% in 2024 to 54% by 2028, with a compound annual growth rate of 63% from 2023 to 2028, driven by advancements in chip technology and increasing consumer demand for AI features [14]. - The introduction of AI assistants in smartphones is anticipated to disrupt traditional business models, as they may directly intervene in transactions, raising concerns among major internet companies [14][15]. Group 4: Future Ecosystem Dynamics - The future of AI smartphones is likely to see a "layered governance" structure, where major apps like WeChat and Taobao may develop their own AI agents to maintain control over user interactions, while smaller apps may be directly managed by system-level AI [16][17]. - The shift towards AI-driven ecosystems is expected to transform the competitive landscape from a focus on traffic acquisition to value co-creation, with smartphone manufacturers taking the lead [16][17].

Investment Rating - The report maintains an "Accumulate" rating for the computer industry [4]. Core Insights - The mobile interaction paradigm is transitioning from GUI to Agentic interaction, allowing users to express their intentions in natural language, which the mobile agent then executes [1][12]. - Two main technical routes for mobile agents are identified: API paradigm and GUI paradigm, each with distinct advantages and challenges [1][2][24]. - The rise of mobile agents signifies a reshuffling of mobile internet traffic among mobile manufacturers, large model manufacturers, and application developers, leading to complex interactions among these parties [3][26]. Summary by Sections Mobile Agent and Interaction Paradigm - The shift from GUI to Agentic interaction is driven by the increasing complexity of applications and the need for more efficient user interactions [1][12]. - Users can now communicate their needs through natural language, with mobile agents handling the execution of tasks across different applications [1][12]. API Paradigm Analysis - The API paradigm involves creating standardized semantic interfaces that require app developers to adapt and expose functionalities for agent use [16][18]. - Apple's App Intents framework exemplifies this approach, emphasizing privacy and structured integration [16][17]. GUI Paradigm Analysis - The GUI paradigm operates without developer cooperation, using visual models to simulate user actions on the screen [2][19]. - Recent advancements in multi-modal models, such as Google's Gemini 3 Pro, have significantly improved the ability to understand and interact with UI elements [19][21]. Comparison of API and GUI Agents - GUI agents offer higher generality, allowing them to operate across various applications without developer adaptation, while API agents excel in reliability, performance, and privacy [2][24]. - API agents can complete complex tasks in a single call, whereas GUI agents may require multiple steps, leading to higher computational costs and potential delays [24]. Evolution of Business Models - The emergence of mobile agents is reshaping the competitive landscape, with mobile manufacturers seeking to leverage traffic entry points and large model manufacturers aiming to create comprehensive applications [3][27][28]. - Application developers face a dual challenge of collaborating with mobile and model manufacturers while protecting their own interests [31]. Recommendations for Attention - Key players in the GUI agent space include ByteDance, Google, Alibaba, and ZTE, while Tencent, Alibaba, and Google are notable in the API agent domain [7][33].