Designed to scale across consumer graphics and data centre workloads, this platform promises large performance jumps and modern graphics API support.
As gaming graphics and AI workloads grow more demanding, GPU developers face increasing pressure to deliver higher performance, advanced ray tracing, and modern software support while reducing dependence on foreign silicon. In China, this challenge is intensified by the push for domestic alternatives that can compete with established players such as Nvidia, AMD, and Intel. This has created the need for next generation GPU architectures that scale across both gaming and AI computing.
Addressing this demand, Moore Threads unveiled its next generation GPU architecture, Huagang, at the MUSA Developer Conference. Set to debut in 2026, the architecture will power both consumer gaming GPUs and high performance AI accelerators, promising major performance improvements across graphics, ray tracing, and compute workloads.
The first gaming product based on Huagang is the Lushan GPU, which will succeed the current MTT S80 and S90 series. According to the company, Lushan is designed to deliver up to 15 times higher performance in AAA gaming workloads and up to 50 times improvement in ray tracing performance compared to its predecessors. The company also claims substantial gains in AI compute, memory capacity, and rendering efficiency, alongside full support for modern graphics APIs such as DirectX 12 Ultimate.
In parallel, the company introduced Huashan, a next generation AI GPU featuring a chiplet based design with HBM memory. The company claims Huashan offers performance comparable to Nvidia Hopper and Blackwell class accelerators, with higher memory bandwidth, improved compute density, and support for precision formats ranging from FP4 to FP64. Designed for large scale AI deployments, Huashan supports high speed interconnects for clustering tens of thousands of GPUs.
Key highlights of the Huagang platform
- Lushan gaming GPU with up to 15x gaming and 50x ray tracing performance gains
- Full DirectX 12 Ultimate and hardware ray tracing support
- Increased AI compute, memory capacity, and rendering efficiency
- Huashan AI GPU with chiplet architecture and HBM memory
- Support for advanced low precision and mixed precision AI formats
The company is positioning itself as a serious domestic contender in both gaming and AI GPUs, signalling China’s growing ambition to build a self reliant high performance computing ecosystem.
