With the application of large models, the development of artificial intelligence has entered the 2.0 era. At this stage, intelligent computing cloud services have accelerated their evolution and become the core driving force for the development of a new generation of artificial intelligence, showing unlimited potential. According to the report "China's intelligent Computing Service Market (the first half of 2024) Tracking" released by IDC, the overall market of intelligent computing services in China increased by 79.6% year-on-year in the first half of 2024, and the market size reached 14.61 billion yuan. According to IDC forecast, China's intelligent computing services market 2023-2028 compound growth rate of 56.2%. This reflects the huge potential of China's intelligent computing services market.

At the annual Tech Talk held by Jinshan Cloud on January 7, 2025, Liu Tao, senior vice president of Jinshan Cloud, said that Jinshan Cloud has comprehensively upgraded the intelligent computing service system to help the application of large models. Sun Xiao, assistant president of Jinshan Cloud and head of public Cloud Product research and development Center, and Xu Yinfei, assistant president of Jinshan Cloud and head of AI product research and development Center, respectively introduced the product technology evolution and landing practice of Jinshan Cloud in intelligent computing cloud and industry model at the meeting.

With AI resonance intelligent computing cloud to platform and Severless evolution

With insight into the trend of intelligent computing cloud, Jinshan Cloud has built intelligent computing cloud 1.0 capability system around basic cloud, intelligent computing network and intelligent computing platform, and is moving towards intelligent computing cloud 2.0, which is evolving into platform and Severless. In terms of basic cloud, the company has comprehensively upgraded computing, storage, and network by integrating software and hardware integration capabilities, and built high-performance infrastructure, which has laid a solid foundation for the development of intelligent computing cloud.

In the process of the development of AI large models, intelligent computing network has become the key to the energy efficiency of new infrastructure in the era of intelligent computing. In terms of intelligent computing network optimization, Jinshan Cloud not only provides a variety of networking architectures, but also provides comprehensive and intuitive network views for operation and maintenance personnel through the self-developed RoCE network monitoring platform, and helps them allocate and utilize resources more reasonably through the computing power platform, which is convenient for customers to use and improve their business continuity.

Facing the challenges of large model customers at the data level, Kingshan Cloud has developed and released a number of Serverless data products, including Serverless Spark, by integrating the capabilities of cloud servers, cloud networks, containers and high-performance object storage, focusing on customers' demand for flexibility and simplicity. Among them, Serverless Spark's performance in the AIGC customer data cleaning scenario is particularly outstanding, and it has achieved landing and gained recognition in many head general artificial intelligence companies.

Generative AI technology continues to mature, and application scenarios continue to expand. Jinshan Cloud will focus on the data platform, training platform, reasoning platform and storage platform to build the overall capacity of the intelligent computing platform, to provide large model customers with an easy-to-use and stable platform within the complete life cycle, to ensure its continuous running time and performance.

With the same frequency with customers to help the industry model intelligent transformation

2024 is known as the "first year" of "big model application landing" and "industry big model development". According to the "China Industry Large Model Competitiveness Research Report (2024)" released by CCID, by combining industry knowledge and intelligent technology, industry large models can provide customized solutions for specific fields, and effectively solve the limitations of general large models in industry applications.

In order to meet the needs of customers in the vertical field of digital intelligent transformation, Jinshan Cloud focuses on the large model of the light boat industry, and builds the full-stack capability of platform, model and application to help customers in law, government affairs, finance and other industries realize intelligent transformation.

In terms of model capability, Jinshan Cloud released qzhou series 13B/32B/70B multi-size industry base model. Jinshan Cloud industry model not only reached the industry advanced level in CEval and CMMLU two authoritative standard evaluation sets, as well as mathematical ability, code ability, instruction following and reasoning ability, Moreover, the fine-tuned qzhou-law model based on high-quality data in the legal industry has surpassed the industry's first-class level in many indicators on the lawbench standard evaluation set. Based on the light boat base model, Jinshan Cloud also launched models applicable to various scenarios for government, legal and financial scenarios, and have been implemented to help customers achieve quality and efficiency.

At the platform and application level, in order to meet the needs of artificial intelligence application development in different scenarios, Jinshan Cloud released a one-stop Hanhai large model development and application platform, covering multiple scenarios such as intelligent dialogue and content creation, based on long-term to provide customers with stable and efficient large model training and promotion, and help them efficiently achieve zero code/low code artificial intelligence application construction. Continue to reduce the user's large model application threshold, the platform has continued to deliver government, enterprise, financial and other industries customers. At the same time, the Jinshan Cloud Hanhai training platform released by the company provides customers with a stable, easy-to-use and extreme performance deep learning training environment for large-scale distributed training of large models. Specially designed for legal consulting and smart contract review, Jinshan Cloud Xiaozhou Law application has served a number of leading law firms and enterprises.

In terms of application landing, Jinshan Cloud has reached a strategic cooperation agreement with Xinyonghe Group in enterprise services. In the cooperation, Jinshan Cloud will build a one-stop platform integrating large model training reasoning, application development and long-term operation capabilities, and create a closed-loop capability of "big data + artificial intelligence" two-wheel drive. At the same time, Jinshan Cloud will also cooperate with Xinyongzhong and professional experts in the field, use the knowledge and data of field experts, continue to iteratively optimize the field of large models including audit and compliance, and quickly build the native application of artificial intelligence in the industry and empower thousands of industries.

The wave of artificial intelligence is surging forward, and large model applications continue to land, which bring unlimited imagination and unprecedented development opportunities to cloud computing. Jinshan Cloud will continue to actively embrace artificial intelligence, join hands with ecological partners, and achieve resilient development with continuously enhanced product technology capabilities and continuously thickened service capabilities.