Chongqing - At Jinfeng Laboratory in Western (Chongqing) Science City, an AI-assisted diagnostic system has cut the time needed to analyze a pathology slide from at least five minutes to about 50 seconds.
Behind that 50-second analysis are large clinical datasets, specialized AI models and the millisecond-level computing network Chongqing is developing.
Over the past year, Chongqing has accelerated efforts to provide computing power on demand with near-instant response times. Its recent selection as one of China’s first regional nodes for national computing-power interconnection strengthens the city’s role in rapidly deploying computing resources across regions.
Pathology slides are crucial to diagnosing tumors and other complex diseases, but interpreting them has long depended heavily on experienced pathologists.
China faces an estimated shortage of about 100,000 pathologists. The gap is particularly acute at primary-level medical institutions, where many county and township hospitals lack specialists capable of independently diagnosing brain tumors and rare cancers. Patients often have to travel to major urban hospitals, adding time and cost to an already difficult medical journey.
Manual slide review is also labor-intensive. A single patient may have more than 10 slides, each of which must be examined carefully. Heavy workloads can slow diagnosis and increase the risk of fatigue-related errors.
To help address the problem, Jinfeng Laboratory developed an AI-assisted pathology system that analyzes slides and supports doctors in making diagnoses.
To improve the system’s accuracy, Jinfeng Laboratory partnered with 46 top-tier hospitals across China to collect nearly 4.8 million pathology images totaling 3 petabytes. It also created China’s first standardized clinical case database for brain tumors, supporting the training of medical AI models.
But training those models and using them to analyze vast amounts of data require substantial computing power.
The laboratory initially faced two choices, neither of them ideal. Building an in-house computing cluster would require heavy investment and could leave expensive equipment underused. Renting cloud computing resources outside Chongqing would be more flexible, but network latency could make them too slow for real-time medical use.
China Telecom Chongqing addressed the bottleneck by providing Jinfeng Laboratory with high-performance computing resources and a dedicated 100 Gbps network. In effect, the company built an information superhighway between the laboratory and computing centers. Each leg of the data transmission has one-way latency of no more than 1 millisecond, making the delay virtually imperceptible to users.
The results are already reaching local hospitals. The system now serves 43 primary-level medical institutions across 34 districts and counties in Chongqing. Its AI model has achieved 90% accuracy in diagnosing brain tumors and has completed nearly 1,000 real-world clinical tests, helping hospitals with limited specialist resources handle complex cases more efficiently.
Researchers conduct experiments at Jinfeng Laboratory in Western China (Chongqing) Science City. (Photo/Chongqing Daily)
Jinfeng Laboratory shows how Chongqing is applying computing infrastructure in the real economy, where fields such as medicine, manufacturing, finance and research increasingly require both greater computing power and faster access.
According to China's Ministry of Industry and Information Technology, the country's intelligent computing capacity had increased by 177% year on year as of June 2026. More businesses and research institutions are becoming dependent on high-performance computing.
Yet many face the same dilemma as Jinfeng Laboratory: buying and maintaining their own equipment is expensive and often inefficient, while computing resources in other regions may not respond quickly enough for real-time production and research.
China launched a national “millisecond computing” initiative in October 2025 to connect, deploy, and access computing resources almost instantly. The country aims to establish low-latency, widely accessible urban computing networks by 2027.
In March 2026, Liangjiang New Area was selected as one of the first pilot regions. Chongqing has since upgraded network equipment and optimized transmission routes to make computing resources faster and more flexible to deploy.
The municipality is also advancing the “Xinjiang Computing Power to Chongqing” project, which brings computing resources supported by Xinjiang's abundant green energy to Chongqing. The model is designed to lower costs while reducing the environmental impact of energy-intensive computing.
Telecom operators are expanding the infrastructure. China Telecom Chongqing has upgraded its all-optical high-speed network to create a distributed urban computing system. China Mobile Chongqing has reconfigured the municipality's optical network and established a dedicated high-speed link between Chongqing and Hami, Xinjiang, enabling computing resources to be deployed rapidly across provincial borders and over long distances.
After nearly a year of construction, Chongqing's millisecond-level computing network is taking shape and being put into use.
According to the Chongqing Communications Administration, China Telecom's high-speed computing network now covers key towns, industrial parks and major business areas across the municipality.
China Mobile's “Xinjiang Computing Power to Chongqing” project can deploy up to 5,000P of computing capacity over distances of up to 6,000 kilometers. Cross-provincial data transmission takes as little as 15.5 milliseconds. Dedicated 500 Gbps links have also been established to support the interconnection and sharing of computing resources across western China.
In Liangjiang New Area, a core pilot zone, data-transmission latency between computing centers has remained below 1 millisecond. This supports highly time-sensitive applications in industrial production, intelligent trading and advanced scientific research. The area is also working to keep latency below 10 milliseconds when ordinary devices access computing resources.
Leveraging the Chengdu-Chongqing national computing hub, Chongqing has formed a multi-tier network with latency of about 1 millisecond in the central urban area, 3 milliseconds across the municipality and 5 milliseconds across the Chengdu-Chongqing region. It can connect directly to 53 key cities nationwide, making Chongqing one of China's leading computing hubs for network performance.
The automotive sector offers another example. Through the “Xinjiang Computing Power to Chongqing” project, China Mobile has built a customized solution for Changan Automobile: green computing resources in Xinjiang train large AI models, while Chongqing handles tasks requiring real-time decisions.
The system supports data training and road simulations for Changan’s Tianshu intelligent-driving platform, accelerating the development of autonomous driving, collision avoidance and connected-vehicle functions while improving safety and reliability.
A delegation attending the international symposium on “The Global Significance of Chinese Modernization” visits Changan Global R&D Center in Liangjiang New Area, Chongqing, June 16, 2026. (Photo/Chongqing Daily)
(Li Jiayan, as an intern, also contributed to this report.)
By continuing to browse our site you agree to our use of cookies, revised Privacy Policy and Terms of Use. You can change your cookie settings through your browser.
For any inquiries, please email service@ichongqing.info