MemSilicon Raises Over 400M Yuan in Series A, Launches LPU Development for LLM Inference

On October 19, MemSilicon announced the completion of its Series A financing round of over 400 million yuan. The round was led by Boyuan Capital, with participation from Bosch Ventures, Chuangling Capital, miHoYo, Zhongguancun Science City, GigaDevice, and others. Existing shareholders including Furong Capital, Guokai Chuangxin, Henghui Capital, Jinqiu Fund, Shunxi Fund, Xiaomi Strategic Investment, and Zhongke Chuangxing also increased their stakes.

MemSilicon originates from the team of academician Liu Ming at the Chinese Academy of Sciences, with more than two decades of accumulated expertise in RRAM (resistive random-access memory) materials, devices, processes, and chip design. The team's work covers reliability, process integration, IP, and chip design, and has advanced to tape-out verification and application. Beyond RRAM, the team also has strong research foundations in 3D integration, near-memory computing, and hybrid compute-in-memory architectures.

The company also announced that Lu Tao, former president and Greater China general manager of Graphcore, has joined as CEO and co-founder. Lu previously served as vice president at Lightelligence, general manager at Cavium China, and held senior roles in AI infrastructure components for more than 20 years.

With the closing of this round, MemSilicon unveiled a strategic upgrade: while continuing the productization of RRAM, it will start development of its first-generation LPU (Language Processing Unit) for large language model inference, aiming to build a next-generation inference engine with "full-capability models, high-speed tokens, and disruptive cost-performance." The company said its LPU is based on memory technology and memory architecture innovation, jointly optimizing computation, storage, and interconnection at the system level to address LLM inference bottlenecks.