Workflow
46.4%!字节火山引擎,夺冠
Zheng Quan Shi Bao·2025-04-11 04:38

Core Insights - The IDC report indicates that the volume of large model calls on China's public cloud is expected to reach 114.2 trillion tokens in 2024, with a significant growth trend starting in the second half of 2024 due to rapid advancements in cloud vendors' large model capabilities and a price reduction strategy for model calls [1][5][6] Group 1: Market Share and Growth - ByteDance's Volcano Engine leads the market with a 46.4% share, followed by Baidu Smart Cloud and Alibaba Cloud, each holding 19.3% [4][6] - The call volume for large models has increased nearly tenfold in six months, from an average of 96.3 billion tokens in June to 952.2 billion tokens in December 2024 [5][6] - The rapid growth in model calls is attributed to the enhancement of large model capabilities and the reduction in calling costs by major cloud vendors [5][6] Group 2: Industry Applications and Trends - The report highlights that 80% of enterprises plan to adopt more generative AI use cases in the next 18 months, indicating a strong growth trajectory for AI applications across various industries [10][12] - The main expected benefits of generative AI for enterprises include improved operational efficiency and increased profits and revenue [12] - The public cloud large model services are expected to see diverse applications in sectors such as e-commerce, automotive, and finance, with significant growth anticipated in image and video model calls starting in late 2024 [12][14] Group 3: Future Outlook and Recommendations - The report suggests that 2025 will be a pivotal year for AI application explosion, with the infrastructure for generative AI expected to reach 35.6 billion RMB and the software market around 4.9 billion RMB [16] - Industry consensus emphasizes the need for a "public cloud first" development strategy to avoid fragmentation and promote effective data sharing [15][16] - The integration of models like DeepSeek into various cloud platforms is expected to enhance awareness and deployment of optimized large models, benefiting the cloud industry significantly [14][15]