International Online (Reporter Hu Xiaowei): Inside the server room, rows of servers are neatly arranged, with indicator lights flashing continuously. In the Suzhou Computing Power Industrial Park, a data center appears calm on the surface but actually bears computing demands from various regions. On September 3, reporters joined the "Looking at China in the First Year of the New Era: Daily Renewal in Jianghuai" online thematic promotional campaign to visit Suzhou, Anhui, and explore the development trajectory of the city's computing power industry. Today, from cloud computing to intelligent computing, and from data storage to artificial intelligence model training, the infrastructure that once served "cloud migration" is becoming the digital foundation supporting the new round of industrial upgrading.
5 Milliseconds: Bringing Suzhou Closer to the Yangtze River Delta in "Computing"
Behind the quiet server rooms lies the "new computing power" accumulated by Suzhou over more than a decade of digital industry layout. The competition in the computing power industry is not just about hardware scale, but also depends on network speed, supporting energy, and response efficiency. "Currently, the one-way transmission latency from Suzhou to Shanghai can be as low as approximately 5 milliseconds," said Liang Xixi, a staff member at the Technical Support Center of Suzhou Huarui Network Information Service Co., Ltd. Compared to the previous transmission latency of 20 to 30 milliseconds, this breakthrough has significantly narrowed the distance between Suzhou and the core computing power zone of the Yangtze River Delta, enabling computing resources to respond more quickly to business needs in the Yangtze River Delta and other areas. It fully adapts to applications requiring real-time data interaction, such as bank real-time settlement, online services, and artificial intelligence, allowing more computing resources that previously required remote invocation to be deployed locally.
Currently, the Suzhou Computing Power Industrial Park has formed a multi-level computing power system including intelligent computing, supercomputing, and general-purpose computing. The park plans to build five T3+ level data centers with a total construction area of approximately 100,000 square meters. Two are currently in operation, one is undergoing electromechanical installation, and two have completed the construction of their main structures.
Over a Decade of Layout, Catching the New Trend of "Computing Power"
The rise of Suzhou's computing power industry is not a short-term windfall from a trend, but the result of long-term strategic planning. In 2013, the Suzhou High-tech Zone was established, explicitly identifying cloud computing and intelligent manufacturing as the "two primary" industries, and率先 initiated the layout of core digital infrastructure such as data centers, networks, and electricity. After more than a decade of deep cultivation, a number of standardized data centers have been established, laying a solid hardware foundation for the subsequent iteration and upgrading of the computing power industry.
"Data centers involve heavy asset investment and have a relatively long initial construction cycle," said Zhu Kuiyuan, head of the Computing Power Industry Center. Suzhou's forward-looking infrastructure layout allows related enterprises to directly utilize existing resources, achieving a "move-in ready" status, which can save at least six months of construction time.
This "first move" is becoming an advantage for Suzhou in undertaking the development of the computing power industry. Zong Qiong, a technician at the Computing Power Intelligent (Suzhou) Computing Center, introduced that an important reason for enterprises choosing to settle in Suzhou is that the local area already possesses relatively complete data center infrastructure conditions. "We can come in directly to deploy equipment and build platforms, without needing to build data centers from scratch."
Not Just About "How Much to Compute," But "What to Compute"
The true value of computing power does not lie in how many servers are stacked in the server room, but in which industrial scenarios these computing capabilities ultimately enter.
At the Bianshui Riverside Scientific Computing Center within the park, approximately 1,550 servers form a supercomputing capacity of about 15P, serving fields such as scientific computing, artificial intelligence, urban computing, industrial simulation, meteorology and oceanography, new energy, and biomedicine.
On the other hand, intelligent computing is more oriented towards applications such as artificial intelligence model training and autonomous driving. Liang Xixi introduced that the National Innovation Intelligent Computing Center provides computing support for fields such as intelligent connected vehicles, Internet of Vehicles (IoV), and autonomous driving. Among them, a new energy vehicle insurance claims model, after obtaining user authorization, can retrieve vehicle-related information, identify the time, location, and damaged components of an accident through model recognition, and provide references for insurance claims. Related technologies can also be used to analyze the causes and probabilities of vehicle accidents, providing data support for subsequent vehicle design and upgrades.
"What to compute" is becoming an important dimension for measuring the value of the computing power industry. Zhu Zangyuan has a straightforward judgment: "Our scale is not the advantage; the diversity of type supply is the advantage." Currently, the park has laid out five main computing centers, including one supercomputing center and four intelligent computing centers, with an intelligent computing capacity exceeding 4,000P, accompanied by services such as general-purpose computing, cloud computing, and data storage. Different types of computing resources can be combined and scheduled according to customer needs, providing services for different scenarios such as scientific research, manufacturing, and government platforms.
From a Data Center to an Industrial Chain
As computing power demand continues to grow, the customers of Suzhou's computing power industry are no longer limited to the local area. Zhu Zangyuan introduced that the rack utilization rate of data centers in the park currently exceeds 90%, with customers covering government platforms, data storage, scientific research and development, manufacturing, and other fields. Some new server rooms already have customers waiting to move in. In his view, Suzhou's advantage in undertaking the computing power industry lies not only in the already built data centers but also in the infrastructure and service capabilities gradually formed over the past decade. "We compete on service," Zhu Zangyuan said.
On one side, servers are continuously expanding in capacity; on the other, application demands are growing steadily. According to the park's plan, the overall computing power scale will reach approximately 8,000P by 2026. With the successive construction and commissioning of new intelligent computing centers, Suzhou is moving further from its early cloud computing industry foundation towards becoming a cluster for the computing power industry.
The data centers laid out more than a decade ago now undertake new applications such as artificial intelligence, autonomous driving, and scientific computing; the infrastructure once used for storing and transmitting data is also transforming into a new digital base driving industrial upgrading. As rows of servers continue to flash, what Suzhou is "computing" is no longer just data, but a new industrial space facing the future.
