ByteDance Redefines AI’s New Paradigm: Comprehensive Push

ByteDance Redefines AI’s New Paradigm: Comprehensive Push

ByteDance Redefines the New Paradigm of AI Development: Comprehensive Pushing Forward. ByteDance’s all-round layout and application-pushing approach has shown the world a new stance. This move not only makes Baidu nervous but also puts pressure on Tencent, Alibaba, and various cloud computing software companies. With an all-in determination, ByteDance aims to quickly obtain an AI admission ticket.

The daily average of 4 trillion tokens far exceeds expectations, indicating that some to-B scenarios that consumers don’t perceive are rapidly consuming tokens. Clues can be seen from this release. The fields of automobiles and local life are growing rapidly. This raises the question: Will the volume of AI to B far exceed that of to C? Can AI succeed where China’s software digitization has not in the past? Can charging by AI usage撬动 enterprise paid subscription business models? According to estimates, ByteDance’s daily average tokens may exceed a quadrillion next year.

AI is regarded as a super entrance and connector. This press conference has made domestic internet giants once again experience FOMO in AI, especially Tencent. AI is not only chatbots and search but also a unified super entrance that can encompass various past concepts and become a tool for ByteDance to expand its empire’s boundaries, connecting thousands of industries, online and offline, software and hardware. There is a powerful commercialization engine in the middle, and at the bottom is a model base supported by computing power and massive data.

ByteDance has confirmed that model leadership has no moat. Finding the “data flywheel” between applications and models is the moat. There are advantages to being a latecomer, achieving better training and inference at a lower computing power cost. ByteDance shows an industry-wide “pushing forward” stance, covering all areas that can be covered and quickly filling scenarios. One is to occupy territory, and the second is similar to the AWS model. The fuller the computing power is used, the lower the cost. In the current stage of AI applications, computing power cost may be the core competitiveness, and the next step may be a price war. Price reductions are good for computing power chips and can achieve software and hardware collaborative optimization.

All this stems from Zhang Yiming’s determination to invest in AI far exceeding that of other domestic internet companies. ByteDance has provided a template for the world this time, attempting to quickly get the ticket with “application pushing forward” and “high-profile actions”. We are fortunate to have such enterprises and entrepreneurs in China participating in the AI wave with a “gamble” attitude. Planet Robo has updated the detailed capex breakdown of ByteDance and uploaded six expert meeting minutes of Doubao and glasses.