Amidst a cooling capital market, Moore Threads held its first-ever 'MUSA Developer Conference,' capturing widespread interest. During the event, Moore Threads revealed its latest full-featured GPU architecture, dubbed 'Huagang,' and introduced the AI training and inference integrated chip 'Huashan,' along with the high-performance graphics rendering chip 'Lushan,' both built on this architecture. Moreover, the company unveiled the 'Kua'e Ten-Thousand-Card Intelligent Computing Cluster' and the MTT C256 super-node architecture plan. Moore Threads also presented cutting-edge technologies like the intermediate language MTX. MTX, a component of the MUSA 5.0 software stack, offers compatibility with GPU instruction architectures across various generations, thereby cutting down on developers' adaptation expenses. It is anticipated to be accessible to developers in the first half of the upcoming year.

The 'Huagang' architecture has markedly boosted computational capabilities, supporting asynchronous programming and ultra-large-scale interconnectivity. It features built-in AI generative rendering structures and is compatible with DirectX 12 Ultimate. The 'Lushan' and 'Huashan' chips, based on this architecture, also exhibit distinctive characteristics. The 'Lushan' chip supports multiple protocols, enables super-node expansion, and has greatly enhanced image performance.

Furthermore, the conference highlighted the 'Kua'e Ten-Thousand-Card Intelligent Computing Cluster' and made strides in ecological expansion, introducing the 'Yangtze River' SoC, MT Robot Embodied Intelligence Solution, MTT AIBOOK AI Computing Notebook, MTT AI Cube Miniature Computing Device, and the MT Lambda Embodied Intelligence Simulation Training Platform.

Moore Threads is striving to broaden its ecological horizons through a strategic deployment of 'full-stack software and hardware architecture + full-scenario products.' Nonetheless, the true test of whether it can successfully carve out a domestic ecological niche will unfold over time.