Domestic sparse-computing AI-chip company MoXing Artificial Intelligence announced on November 3 the completion of a nearly 1 billion yuan Series C+ round, with participants including Renai Capital, Begonia Fund and several state-owned and financial investors, Cailian Press and Xinzhiwei reported. The round follows a similar near-1-billion-yuan raise just five months earlier, totaling nearly 2 billion yuan across the two.
The financing reflects rising capital conviction in sparse computing's value for inference economy. As agent applications spread, sustained inference load makes running cost a permanent line item; the National Data Administration said China's daily Token calls exceeded 140 trillion in March, up more than 40% from end-2025. With power, energy and infrastructure constraints tightening, the metric has shifted from peak performance to effective inference throughput per cost.
MoXing's dual-sparse optimization skips unnecessary computation, and its S30 and S40 cards have won MLPerf Inference three consecutive times; a remote-sensing collaboration showed inference throughput up to 2.88x higher with almost no accuracy loss. Proceeds will fund a next-generation SparsePrime card and a sparse-computing industry-academic alliance.
Source: Cailian Press / Xinzhiwei
