Moffett AI, a Shenzhen-based developer of AI inference chips, has completed a Series C+ round of nearly CN¥1 billion. Investors include Renai Capital, Haitang Fund, Dongheng Petroleum, Biqiu Capital, Paila Capital, and Zhongcai Capital. The company builds chips that use sparse computing to lower AI model inference costs.
Funding so far
The round follows another nearly CN¥1 billion financing about five months earlier. Together, the two rounds total close to CN¥2 billion. The new capital will fund upgrades to its sparse-computing inference chips and early research on next-generation processors.
Deployment and customers
Moffett AI has installed inference clusters in four regions of China. It has also partnered with major telecommunications operators. A 1,000-card cluster is among these deployments.
Products and software
Its next-generation SP200 accelerator card is scheduled to arrive before the end of the year. Moffett AI also plans to release an upgraded SECA software stack built for sparse computing. The software includes an operator library, a compiler, and tools redesigned around its architecture.
Moffett AI is also building a sparse computing research alliance with universities such as Fudan, Tsinghua, and Shanghai Jiao Tong. The alliance will run joint research topics across algorithms, mathematics, and software tooling. These efforts are intended to support wider adoption of sparse computing.