DAU已死，TPD永生

Core Insights - OpenAI is reportedly shifting focus from the traditional DAU (Daily Active Users) metric to a new metric called TPD (Tokens Per Day), which measures the value generated by users through AI interactions rather than just user engagement [2][3][25][60]. Group 1: Shift in Metrics - DAU has been a key performance indicator in the internet industry for the past 20 years, primarily measuring user attention [4][5]. - The traditional logic of DAU is based on capturing user time to drive advertising and sales, but this is being challenged by AI agents that can perform tasks without direct user interaction [6][7][62]. - OpenAI's product lead emphasized that DAU does not reflect the value created by users, leading to the adoption of TPD as a more relevant metric [3][25]. Group 2: The Role of AI Agents - AI agents can automate tasks that previously required significant user attention, fundamentally changing how productivity is measured [8][10][22]. - Users can now delegate complex tasks to AI agents, resulting in high token consumption without the need for constant app interaction, thus diminishing the relevance of DAU [12][24][63]. - The emergence of AI agents suggests that many traditional applications may become obsolete, as users will rely on agents to perform tasks rather than manually interacting with multiple apps [17][20][24]. Group 3: TPD as a New Metric - TPD measures the computational resources a user drives through AI, indicating a shift from time-based productivity to value-driven productivity [25][27]. - Companies like Midjourney and Cursor exemplify high TPD, achieving significant outputs with minimal personnel, showcasing the efficiency of AI in driving productivity [30][32][43]. - The value of a user is now determined by their ability to leverage AI agents effectively, rather than the time they spend on applications [38][39]. Group 4: Implications for Business Models - The transition to TPD will alter business models, moving from user-based revenue to consumption-based revenue, where companies charge based on token usage rather than user counts [46][47]. - OpenAI's projected annual recurring revenue (ARR) for 2025 is expected to exceed $20 billion, largely driven by token consumption from enterprise clients [47]. - Future SaaS models will likely charge based on the computational power required, rather than per user, reflecting the shift in how value is generated [48][49]. Group 5: Future of Software and User Engagement - The rapid adoption of AI agents is expected to lead to a decline in traditional software applications, as users will prefer to interact with agents that can complete tasks efficiently [54][66]. - The competitive landscape will shift from user acquisition to enhancing the efficiency of user-driven AI interactions, fundamentally changing how platforms are valued [45][62]. - While DAU remains relevant for content and social products, its importance is diminishing in productivity tools and enterprise services, where TPD is becoming the primary measure of success [61][63].