Chongqing - Chongqing Zhisuan Technology Co., Ltd., located in Xiyong Micro-electronics Industrial Park, recently was recognized as one of China's first 50 exemplary cases of city-wide digital transformation for its cross-regional computing power scheduling system.

Amid the growing challenges of allocating and utilizing computing power resources in Western China and the urgent need to develop AI, the system provides an exemplary model for integrating regional and inter-regional computing power while coordinating data, algorithms, and resources.

Computing power resources consist of chips and energy. The western region has significant potential due to its abundant energy reserves and high energy consumption. However, challenges like technology, costs, and regulations limit resource utilization efficiency.

The system aims to optimize and manage computing resources in Chongqing and western China by integrating data centers, improving resource utilization by 12%. With real-time monitoring and dynamic allocation, idle rates have been significantly reduced.

Chongqing Artificial Intelligence Innovation Center. (Photo/Xiyong)

The system draws on surplus computing power from neighboring provinces, establishing an efficient cross-provincial scheduling platform. A high-speed, direct connection network with a latency of just 41 milliseconds has been created, facilitating a seamless blend of computing power with green energy and lowering enterprise transformation costs.

Data, algorithms, and computing power form the three pillars of AI: data serves as the foundation, algorithms guide data processing, and computing power drives efficiency. Together, they propel AI's continuous evolution. As the sole operator of the Chongqing Artificial Intelligence Innovation Center, Chongqing Zhisuan Technology Co., Ltd. has incubated 96 AI applications and trained over 1,000 AI professionals.

Beyond coordinating computing resources, the system unifies data, algorithms, and computing power. Chongqing Zhisuan has established an AI model service zone, developing platforms that support model development, training, and inference. Integrating data from various industries and public sources has made AI development more accessible.

Furthermore, the development of universal algorithms like large language models and computer vision (CV) enables a cohesive application of computing power, data, and algorithms, significantly boosting AI efficiency for enterprises.

The system has developed hundreds of algorithm models and data resources, successfully applied across multiple sectors. Notably, its integration with petrochemical explosion-proof robotics has reduced safety risks by 50%. The cloud computing system now precisely manages over 10,000 hardware units and 24,000 cloud instances, ensuring comprehensive lifecycle management.