en.Wedoany.com Reported - On June 29, the National Supercomputing Internet launched the SCNet Ten-Thousand-Card Co-Creation Incentive Plan. Leveraging the scaleX ten-thousand-card supercluster, it opens up to AI developers, startup teams, and research institutions worldwide up to hundreds of millions of Tokens in quota and thousands of card-hours of Chinese computing resources. The plan will run until December 31, 2026, focusing on two main themes: AI and AI4S. It will carry out co-creation projects including best practices for Chinese computing power, developer competitions, paper reading sessions, live training broadcasts, and AI application incubation, covering scenarios such as livelihood applications, industrial simulation, and biomedicine. Selected participants can use over 1,500 open-source models at zero cost and gain benefits including technical empowerment, brand exposure, and cash sharing.
The focus of this plan is to transform ten-thousand-card-level computing power from a mere infrastructure into an innovative resource directly usable by developers. AI teams, research institutions, and startups can apply for projects to gain support in computing power, models, and platforms, lowering the barriers to large model training, inference validation, and AI4S experiments.
The scaleX ten-thousand-card supercluster serves as the underlying resource support for this incentive plan. Large model training, scientific intelligent computing, and high-throughput inference all require the coordination of large-scale heterogeneous computing power, distributed storage, task scheduling, network communication, and model service platforms. In the past, many small and medium-sized teams, even with model ideas, industry data, or application scenarios, were often constrained by computing costs, deployment barriers, and engineering environments. The SCNet Ten-Thousand-Card Co-Creation Incentive Plan packages and opens up token quotas, thousands of card-hours of computing resources, open-source models, and technical support, allowing developers to focus more on model tuning, application validation, and scenario implementation, rather than investing heavily in building the underlying environment first.
AI and AI4S are the two main lines of this plan. The AI direction leans more towards intelligent applications, industry models, agents, content generation, knowledge Q&A, office tools, and the developer ecosystem; the AI4S direction serves scientific research and engineering computing, covering high-computing-power scenarios such as industrial simulation, biomedicine, materials computation, complex system modeling, and scientific data processing. By placing both themes within the same plan, the National Supercomputing Internet indicates that supercomputing resources are expanding from traditional high-performance computing tasks to a comprehensive platform shared by large models, scientific intelligence, and industrial applications. Livelihood applications can verify the inclusiveness of AI services, industrial simulation can improve engineering R&D efficiency, and biomedicine requires stronger capabilities in model training, molecular simulation, and data analysis.
Selected teams can use over 1,500 open-source models, which is a highly practical benefit of the plan. For developers, computing resources are just the first step; the model library, runtime environment, technical support, and application incubation also affect whether a project can succeed.
Open-source model resources can help teams quickly build prototypes, reducing the cost of training models from scratch. Developers can choose language models, multimodal models, industry models, or inference models based on task requirements, and then use the platform's computing power for fine-tuning, evaluation, and deployment. Technical empowerment will address issues of model adaptation, task parallelism, inference acceleration, data processing, and engineering deployment; brand exposure and cash sharing will help outstanding projects gain attention from clients, investors, and industry resources. For startup teams, this combination of benefits is more comprehensive than a one-time computing voucher, as it covers multiple stages from resource acquisition to result demonstration, and from application incubation to commercial transformation.
This plan will also drive China's computing ecosystem from "resource construction" into "resource operation." After a computing center is built, what truly determines its value is the utilization rate, number of applications, developer activity, and the ability to transform industrial projects. Through the SCNet Ten-Thousand-Card Co-Creation Incentive Plan, the National Supercomputing Internet connects developers, research institutions, startup teams, and computing platforms, forming an operational mechanism of computing sharing, content co-creation, and brand co-building. The platform will complete qualification review and project matching within 5 working days, and after approval, distribute computing resources to user accounts. This process design is closer to developer ecosystem operations rather than traditional project-based resource allocation.
Opening up to global developers also gives this plan an international collaborative nature. Competition in the AI ecosystem does not rely solely on model companies and large enterprises; startup teams, research labs, and industry developers also contribute to application innovation. If the National Supercomputing Internet can consistently provide stable computing power, model tools, and low-barrier services, it has the opportunity to attract more teams in AI4S, industrial simulation, biomedicine, and livelihood applications to the platform, generating more reusable cases. What needs to be monitored subsequently is the quality of selected projects, the actual utilization efficiency of computing resources, developer retention, the commercialization outcomes of co-creation projects, and the stability and scheduling capability of the scaleX ten-thousand-card supercluster in large-scale tasks.
The launch of the SCNet Ten-Thousand-Card Co-Creation Incentive Plan marks that China's supercomputing infrastructure is more proactively serving the AI industry and scientific innovation. The ten-thousand-card cluster provides underlying computing power, the hundreds of millions of Tokens quota lowers the barrier to model invocation, the over 1,500 open-source models address the starting point of development, and application incubation and cash sharing target result transformation. As the plan continues until December 31, 2026, the National Supercomputing Internet will test the carrying capacity of China's computing platform in the developer ecosystem, scientific intelligence, and industrial applications through a batch of AI and AI4S co-creation projects.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
