Global large model iteration continues to accelerate with no signs of slowing down. AI model invocation volume is expected to maintain high growth next year, and this growth is not linear; once breakout applications like phone assistants and multimodal tools emerge, they could trigger a leap in AI model usage and computational demand. It is recommended to closely monitor the development trends of the AI industry both domestically and internationally, focusing on investment opportunities in AI computing power, domestic AI hardware, and AI applications. The main views of Huaxin Securities are as follows.
On December 18, 2025, ByteDance's Volcano Engine hosted the 2025 FORCE Momentum Conference. Volcano Engine released updates including the Doubao large model 1.8 and the video generation model Seedance 1.5 pro, and announced that the daily token usage of the Doubao large model has exceeded 50 trillion, ranking first in China and third globally. According to media reports, Volcano Engine has raised its annual revenue target of hundreds of billions of yuan, originally set for achievement between 2029 and 2031, by several hundred billion yuan.
The Doubao model is accelerating its update iteration, upgrading its enterprise application platform and overall savings plan. At this conference, Volcano Engine officially released the Doubao large model 1.8 (DoubaoSeed-1.8), specifically optimized for multimodal and Agent scenarios. In terms of evaluation metrics, the Doubao large model 1.8 still lags behind advanced overseas models like Gemini 3 Pro, but this release is not a major version update. The President of Volcano Engine stated in a media interview that the Doubao large model 2.0 is expected to launch soon, potentially bringing significant improvements in model capability, which could be a major event for the development of domestic AI models.
Volcano Engine also released the video generation model Seedance 1.5 pro, which supports synchronized audio-visual output and multi-person, multi-language dialogue dubbing, possessing stronger cinematic-level narrative power to meet the high-end creative demands of film, television, comic series, e-commerce, and advertising. To facilitate enterprise AI applications and Agent deployment, Volcano Engine comprehensively upgraded the enterprise-level AI Agent platform AgentKit, released the enterprise version of the AI-native programming tool TRAE, and launched the HiAgent "1+N+X" intelligent workstation. Regarding model pricing, Volcano Engine further innovated its pricing model by introducing the "AI Savings Plan," the industry's first large model savings plan, covering all pay-as-you-go large model products on Volcano Engine, including the Doubao model family and various third-party open-source models.
ByteDance is building a strategic ecosystem around AI large models and has revised its future revenue growth target upwards. ByteDance's Volcano Engine has now established a complete set of AI cloud-native architecture, from models to MaaS, Agent development, and Agent operations. The current daily token invocation volume of the Doubao large model has reached 50 trillion, an increase of approximately 67% from the 30 trillion disclosed in September, maintaining high quarterly growth, and an increase of over 200% compared to the 16.4 trillion disclosed six months ago. Behind the high growth in model invocation is ByteDance's ecosystem strategy. Currently, 9 out of the top 10 global mobile phone manufacturers have chosen Volcano Engine services, covering over 500 million smart terminals. The Doubao large model is also widely used in fields such as automotive, finance, universities, and consumer sectors. According to media reports, Volcano Engine has adjusted the hundred-billion-yuan annual revenue target set in 2021: the original achievement timeline of 2029-2031 remains unchanged, but the target amount has been increased by several hundred billion yuan. Based on this, it is projected that ByteDance's AI and cloud business will continue to experience rapid growth in the future.
Risk warnings include: (1) Pressure from AI technology investment and cost control; (2) Intensifying market competition; (3) Challenges to sustainable user growth; (4) Exchange rate fluctuations and internationalization risks.
