后摩智能 develops storage-compute (存算一体) AI chips aimed at endpoint and edge large-model deployments. The company completed a strategic financing of several hundred million RMB from China Mobile-affiliated funds. It has signed a strategic cooperation agreement with China Mobile Research Institute to jointly advance R&D and mass-production application of its chips. China Mobile Research Institute will support product demand analysis, end-side large-model research, new product design, scenario exploration, and promotion of software toolchain standards. 后摩智能 will provide diversified compute support for deployment and application of endpoint and edge large models. The partners plan to jointly explore product forms for government and enterprise large-model appliances, Xinchuang AI PCs, home computing boxes, robots, and industrial inspection terminals, and to drive commercialization. Houmo Intelligent develops high-compute AI chips using a storage-compute integrated (存算一体) architecture to overcome memory bottlenecks and boost performance and energy efficiency. The company's chips target large-edge and cloud inference scenarios, including intelligent driving and pan-robotics. Houmo says its architecture yields substantially higher compute and far better power efficiency compared with conventional designs. Founded at the end of 2020, the team exceeds 100 people, with over 70% of R&D staff holding master’s degrees or PhDs. The company completed a first-chip validation tape-out in August 2021 and plans to accelerate product commercialization and mass production. Funds from the recent round will be used to increase R&D on storage-compute integrated large-compute AI chips and to expand in intelligent driving and robotics markets. Houmo Intelligence focuses on original compute-in-memory (存算一体) intelligent computing chips and a software–hardware integrated platform designed to overcome the memory-wall and performance bottlenecks of traditional von Neumann architectures. The company develops high-compute, low-power, high-energy-efficiency chips and solutions aimed at large-edge scenarios (robotics, unmanned vehicles) as well as cloud inference and training. Its technology and storage-process innovations target substantial improvements in AI chip energy efficiency. The founding team is composed of international academics and senior chip-industry experts; founder Dr. Wu Qiang wrote an award-winning Princeton PhD thesis on high-efficiency compute chips and compilers and has worked at AMD, Facebook, and served as CTO at Horizon. Team members have long research experience in advanced memory devices and compute-in-memory techniques and have published in ISSCC, ISCA and HPCA. The company completed a financing round of tens of millions of dollars (angel) to support product development and commercialization. Houmo Intelligence (后摩智能) designs high‑energy‑efficiency, large‑compute AI chips based on in‑memory (compute‑in‑storage) architecture to address performance and power bottlenecks. The company targets single‑chip compute performance in the hundreds to thousands of TOPS while keeping power consumption low, enabling both edge scenarios (autonomous vehicles, robotics) and cloud workloads (recommendation, image analysis inference and training). Founded at the end of 2020 by Dr. Wu Qiang and a team of international academics and industry experts, Houmo claims deep technical and engineering capabilities across circuit design, AI core design, SoC, compiler and toolchain. The business says it is the first domestic team pursuing in‑memory architecture at scale for very large compute chips in China. Houmo positions its product roadmap to accelerate AI adoption by delivering “ten‑fold effect” chips for ultra‑large compute demands. The company completed a RMB 300 million Pre‑A financing and plans to use the funds to accelerate chip R&D, expand the team, and advance early market deployment and commercial landing.

Total raised

$46M

Funding rounds

4

Latest round

Series A

Latest activity

Aug 2021