China Edges Closer to Competing with Global GPU Giants
China appears poised to challenge the dominance of NVIDIA, AMD, and Intel in the graphics market. A little more than a year and a half after its founding, Moore Threads introduced the MT Unified System Architecture, known as MUSA, the backbone for its upcoming graphics cards. The architecture is built around a 12 nm process technology, signaling a focus on efficiency and performance balanced with manufacturing practicality.
At a single event, the company unveiled two graphics accelerators intended for different segments: the MTT S60 designed for desktop PCs and the MTT S2000 aimed at servers. Early performance figures place the MTT S60 around 6 TFLOPS, while the MTT S2000 reaches about 12 TFLOPS. Those numbers put them in a similar league to current mid to high tier consumer GPUs and comparable server solutions, illustrating Moore Threads’ ambition to offer compelling alternatives in both consumer and enterprise spaces. Comparisons with other generations show the MTT S60 edging close to the performance level of mid range cards such as the RTX 2060 and a portion of the performance of higher end solutions, while the MTT S2000 targets workloads that benefit from double precision and steady throughput in data center environments.
MTT S60 comes with 8 GB of memory, while the S2000 is equipped with 32 GB. The architecture supports a broad software ecosystem, including OpenCL, OpenGL, Vulkan, and NVIDIA CUDA compatibility. In addition, hardware encoding and decoding support spans H.264, H.265, AV1 for both encoding and decoding, with the decoding stack extending to VP8 and VP9. This mix of APIs and codecs is designed to ease integration for developers and to broaden application compatibility across gaming, media, and professional workloads.
During the presentation, demonstrations included League of Legends running in Full HD, offering a glimpse of real world performance in a popular esports title. While detailed performance tables were not released at the time, the visuals and responsiveness suggested a capable experience for competitive gaming and casual play alike. The event highlighted practical usage scenarios and emphasized the potential for future software optimization and driver support from Moore Threads and partners.
Details on launch timing and price remain scarce. The company indicated that product availability would be announced later, with the expectation that pricing information will accompany a formal release window. In parallel, Rossgram, another player in the region, showcased its own application interface, underscoring the broader industry activity around new GPU architectures and platform ecosystems in the market.
The broader implication of the introduction is a renewed emphasis on domestic semiconductor capabilities and regional supply chains. Moore Threads’ steps to deliver a complete stack—from silicon design to software compatibility and use case demonstrations—illustrate an ambition to compete at scale. Industry observers will watch closely how the MUSA architecture evolves, how developers adopt the platform, and how Moore Threads extends its footprint beyond initial markets. As the company continues to refine drivers, optimizations, and developer tools, the coming months could reveal a clearer path to competitive pricing and broader gaming and data center adoption. The balance between performance, efficiency, and ecosystem support remains the key lever for any emerging GPU platform aiming to challenge established incumbents on a global stage.