Donghua Software Plans AI Computing Clusters, in Talks for Ascend 384 Supernode Business
Donghua Software said it plans to build domestically controlled AI computing infrastructure, including clusters ranging from 1,000 to 10,000 GPUs. The company is in talks with Huawei regarding Ascend 384 supernode computing cluster business, though the outcome remains uncertain. It will disclose further details if a concrete plan materializes and meets disclosure standards.
Donghua Software said on an investor interaction platform that, in the area of domestic computing power, it plans to build independently controlled intelligent computing infrastructure, laying out intelligent computing clusters at the scale of thousands to tens of thousands of cards.
On the ecosystem front, the company has established deep cooperation with companies such as Huawei, Moore Threads, and Enflame, forming a collaborative model where scenario implementation and computing power infrastructure support each other.
The company maintains deep ecosystem cooperation with Huawei and is currently actively engaging in discussions regarding the Ascend 384 supernode computing cluster business. This commercial negotiation involves certain uncertainties. If the business forms a clear implementation plan and meets information disclosure standards, the company will promptly issue relevant announcements to fulfill its disclosure obligations.