RM新时代有限公司

2026Year3In March, the government work report reviewed and approved at the fourth session of the 14th National People's Congress proposed for the first time"creating new forms of intelligent economy", clearly requiring"deepening and expanding'artificial intelligence+', promoting the accelerated adoption of new-generation intelligent terminals and agents""implementing new infrastructure projects such as ultra-large-scale intelligent computing clusters and computing-electricity synergy"This top-level design marks the transformation of artificial intelligence from a"technical tool"into the"underlying logic"of economic development. Local governments urgently need to build new infrastructure integrating computing power, algorithms, data, and scenarios to support large model R&D and industrial application deployment. In this context, Dongxihu District of Wuhan City actively responds to the national strategy and takes the lead in launching the construction of a large model intelligent computing center.

Dongxihu District is the location of the National Cybersecurity Talent and Innovation Base, and has formed a "1+4" modern industrial system with cybersecurity and big data as the core, and electronic information, intelligent manufacturing, modern commerce and logistics, and food and health as the leading industries. It has gathered over 220 cybersecurity enterprises, and has the Cybersecurity Schools of Wuhan University and Huazhong University of Science and Technology, as well as the planned Wuhan Cybersecurity University. To seize the opportunities of the AI large model industry and accelerate the construction of a collaborative ecosystem integrating "data, computing power, algorithms, and scenarios," Dongxihu District, together with CloudWalk Technology and Huawei, is jointly building a large model intelligent computing center for training and inference around the "One Center, Four Platforms" (Dongxihu Training and Inference Integration Center and AI Public Computing Platform, Content Review Large Model Application Platform, Intelligent Computing Center Operation Platform, and Industrial Ecosystem Collaboration Platform). The aim is to leverage independent and innovative software and hardware infrastructure to create an "Industry AI Application Innovation Center," achieve unified management of computing resources, efficiently incubate industry large models, and inject powerful momentum into the high-quality development of the regional digital economy.The "1+4" modern industrial system brings together over 220 cybersecurity enterprises, and includes the School of Cyber Science and Engineering at Wuhan University and Huazhong University of Science and Technology, as well as the planned Wuhan Cybersecurity University. To seize the development opportunities of the AI large model industry and accelerate the construction of a collaborative ecosystem integrating "data, computing power, algorithms, and scenarios," Dongxihu District, together with CloudWalk Technology and Huawei, is jointly building a large model training and inference intelligent computing center centered on "one center and four platforms" (Dongxihu Training and Inference Integration Center and AI Public Computing Platform, Content Review Large Model Application Platform, Intelligent Computing Center Operation Platform, and Industrial Ecosystem Collaboration Platform). This initiative aims to leverage independently innovated software and hardware foundations to create an "Industry AI Application Innovation Center," achieving unified management of computing resources and efficient incubation of industry-specific large models, injecting strong momentum into the high-quality development of the regional digital economy.

New Landscape of Intelligent Computing:

DongxihuAIThe Question of Industrial Leap!

Facing intensified global AI competition and the in-depth implementation of domestic"artificial intelligence+"strategies, Dongxihu District has diverse new demands in promoting the deployment of the large model industry. As theAIindustry accelerates its upgrade, it is necessary to both consolidate a trustworthy computing power foundation and achieve model"ownership"; to support both scientific research and industrial aggregation. This requires the intelligent computing center to be not only a computing power provider, but also an ecosystem builder and value creator.

Adaptation of Computing Power Foundation to Industrial NeedsAdaptation for demand: The demand for digital and intelligent transformation among enterprises in cybersecurity, intelligent manufacturing, and electronic information within the district is strong, but large model training places extremely high requirements on the scale, performance, and reliability of computing clusters. Small and medium-sized enterprises face the dilemma of""wanting to use but not daring, able to use but cannot afford"". Dongxihu District urgently needs a public computing infrastructure that is intensively built, elastically supplied, and integrated for training and inference, making computing power as readily available as water and electricity.

Integration of Cutting-edge Technology and Characteristic Scenarios:Dongxihu District, home to the cybersecurity base, has unique scenario advantages in content security, data security, and other fields. However, large models face security challenges such as"hallucination", bias, and non-compliant content generation during deployment. High-investment, low-output basic large models can be"utilized", but what truly empowers local industries are industry large models incubated from data in government affairs, manufacturing, energy, and other fields within the district, realizingAIassets"ownership"and accumulating regional intelligent capital.

Aggregation of Industrial Factors and Innovation Ecosystem:Although a large number of cybersecurity enterprises have gathered in the district, the large model industry chain is long and finely divided, requiring the collaborative coexistence of algorithm providers, computing infrastructure, data service providers, application developers, and other diverse entities. How to build a platform that effectively links"government-, industry-, academia-, research-, and application", enabling efficient connection between the research resources of Wuhan University and Huazhong University of Science and Technology and the needs of local enterprises, and allowing leading enterprises to drive collaborative innovation among small and micro enterprises, is key to forming industrial cluster synergy.

Transformation of Application Incubation and Business Closed Loop:The ultimate value of AI lies in empowering thousands of industries. Dongxihu District has traditional advantageous industries such as food and health, and modern commerce and logistics, and urgently needs to transform large model technology into replicable and scalable industry solutions. However, from technology prototype to commercial application, there is a long chain involving model fine-tuning, application development, security evaluation, compliance filing, etc. Dongxihu District needs an operation platform integrating model services, data openness, enterprise aggregation, and talent cultivation to promote collaborative innovation among AI enterprises, technology providers, and research institutions.

Intelligent Computing Integration: CloudWalk×Huawei

created"One Center, Four Platforms"New Industrial Ecosystem

To meet practical development needs, Dongxihu District, together with CloudWalk Technology and Huawei, has jointly built a comprehensive solution centered on"One Center, Four Platforms"from multiple dimensions including computing power, models, ecosystem, and security, achieving a closed loop from computing power supply to industrial empowerment.

?? Training-Inference Integrated Architecture Unleashes Full Lifecycle Value of Computing Power

The core of the project is the construction of the Dongxihu Training-Inference Integration Center, planned with125P Flops@FP16theoretical peak computing power, adopting the full-stack independently innovative AscendAIcomputing platform. This platform is based on the high-performance computing capabilities of AscendAIprocessors, combined with400GE RoCElosslessRDMAnetworking and all-flash+hybrid flash intelligent tiered storage system, building a"cluster capability that leads in three dimensions: computing power scale, utilization rate, and availability."Notably, the platform adheres to the"same-architecture training and inference"design philosophy. Model retraining and online inference require no complex conversion; training and inference resources can be time-shared and flexibly interchanged, maximizing computing power utilization efficiency and protecting government investment.

?? Industry Model Incubation: ForgingModels as Tools to Build Distinctive Capabilities

Leveraging the rich industrial data of Dongxihu District, the project focuses on six major fields including cybersecurity, smart energy, and intelligent manufacturing, incubating industry-specific large models that can be sold and promoted. Through"incubation with district dataL1model'ownership'", it accumulatesAIassets with independent intellectual property rights and Hubei brand identity, helping traditional enterprises complete digital and intelligent transformation. Relying on CloudWalk's independently developed"CongRong Large Model", which integrates multimodal processing, security assessment, and public opinion detection, it efficiently captures and filters non-compliant content through in-depth analysis of text, images, video, and audio data. This is not only an output of technical capabilities but also a strategic move to transform Dongxihu District's cybersecurity industry advantages into core competitiveness in the era of large models.

?? Full Lifecycle Intelligent Computing Operations System Drives Industrial Upgrading

The intelligent computing center operation platform provides comprehensive AI industry operation services for enterprises and institutions through computing power leasing, model services, and managed operations. Based on computing power operations and unified user management, the platform achieves full lifecycle management of computing resources through a visual cockpit.——From resource scheduling and task monitoring to metering, billing, and permission control, everything is visible and manageable. More importantly, the platform focuses on building an industrial ecosystem through mechanisms such as node authorization, industry matchmaking, and joint R&D, promoting enterprise aggregation and traditional industry upgrading, and driving the efficient operation of the Dongxihu AI industry cluster.

?? Platform Synergy: Building an Ecosystem for Industrial Aggregation and Innovation Incubation

Centered on the"Industrial Intelligent Computing Service Platform, National Security Large Model Innovation Incubation Platform, Industrial Aggregation Production Factor Collaborative Development Platform, and Scientific Research Innovation and Talent Training Platform", the project gathers partners including AI enterprises, data service providers, and research institutions."Among the Four Platforms", the Industrial Ecosystem Collaboration Platform aims to gather AI enterprises, cybersecurity companies, large model technology providers, data providers, computing power operators, and research institutions to create a collaborative development platform for production factors. This cooperation model not only improves resource utilization efficiency but also promotes close collaboration across the industrial chain, injecting new vitality into the development of the intelligent computing center.

The construction of this project is an important practice of CloudWalk Technology and Huawei jointly creating the firstAI 2.0template in a provincial capital city, and a key measure for Dongxihu District to undertake the national responsibility of building the National Cybersecurity Talent and Innovation Base. Looking ahead, with the125Pfull operation of intelligent computing power, the official release of the content review large model, and the deep operation of the four platforms, Dongxihu District will become an important growth pole for AI industry innovation and development in central China, contributing to Wuhan's goal of becoming the top computing power hub in central China and Hubei's construction of a national digital economy highland, writing a new chapter in AI empowering high-quality development of the real economy.