Moore Threads Set to Unveil Its Next-Generation GPU Architecture

IT Home reported on December 9 that, according to Moore Threads’ official WeChat account, the company will host its first-ever MUSA Developer Conference (MDC 2025) from December 19 to 20 at the Zhongguancun International Innovation Center in Beijing.

Moore Threads stated that MDC 2025 will serve not only as a comprehensive showcase of its MUSA technology system and full-stack capabilities, but also as a strategic platform connecting technological innovation, developer collaboration, and industrial applications. Centered on the unified MUSA system architecture, the conference will present Moore Threads’ full-stack ecosystem progress—from hardware to software, from core technologies to real-world application scenarios—accelerating the large-scale adoption of domestic, fully functional GPU technologies across industries and strengthening the fundamental computing infrastructure for the intelligent era.

Zhang Jianzhong, Founder, Chairman, and CEO of Moore Threads, will for the first time deliver a systematic explanation of the company’s full-stack development strategy and long-term vision built around MUSA. During the conference, Moore Threads will officially unveil its next-generation GPU architecture and introduce a complete roadmap covering its product portfolio, core technologies, and industry solutions. The event will also feature real-world deployment cases across multiple sectors and updates on ecosystem development.

Earlier, on December 5, 2025, Moore Threads Intelligent Technology (Beijing) Co., Ltd. (stock code: 688795.SH) officially debuted on the STAR Market of the Shanghai Stock Exchange, becoming China’s first full-function GPU company to enter the capital market.

Founded in 2020, Moore Threads is built upon its self-developed MUSA unified system architecture, achieving a major technological breakthrough by enabling a single chip to simultaneously support AI acceleration, graphics rendering, physics simulation, scientific computing, and ultra-high-definition video encoding and decoding. To date, the company has successfully mass-produced five chips and iterated through four generations of GPU architectures and intelligent SoC products. Its offerings now span from individual chips and computing cards to intelligent computing clusters, forming a comprehensive acceleration product matrix for artificial intelligence, scientific computing, and graphics rendering—fully supporting cloud, edge, and endpoint computing scenarios.