Parthus (维泛智能), a spin-off from Peking University's PAICORE Lab, has raised hundreds of millions of...
The AMW Read
Novelty 2: introduces a new Chinese entrant with a differentiated neuromorphic-GPU fusion approach, a meaningful update to the robotics chip landscape. Significance 2: segment-level potential to challenge Nvidia's Jetson dominance in Chinese robotics, but pre-tape-out risk caps impact.
Parthus (维泛智能), a spin-off from Peking University's PAICORE Lab, has raised hundreds of millions of yuan in seed funding for a native robot "brain chip" that fuses neuromorphic computing with GPU architecture. The round was co-led by Zhongguancun Capital and Qihang Investment, with participation from Shanghai Future Industry Fund, Shixi Capital, Biwin Storage, Yancheng Group, Haiyi Investment, and Tanogen Capital. Founded in May 2025, Parthus is developing a Brain-Inspired GPU (BiGPU) architecture that unifies spiking neural network (SNN) and artificial neural network (ANN) computation in a single instruction-set framework, aiming to serve as a drop-in replacement for Nvidia's Jetson series in embodied intelligence workloads.
Why it matters: Parthus enters the AI chip market at a moment when the robotics segment is almost entirely dependent on Nvidia Jetson modules, creating a classic hyperscaler-distribution moat that no Chinese startup has yet challenged. The company's differentiation — a unified SNN/ANN architecture rather than the heterogeneous SND+NPU approach attempted by earlier neuromorphic startups — directly updates the open debate about whether neuromorphic computing can achieve practical commercial deployment outside research labs. If Parthus delivers on its BiGPU roadmap, it could become one of the first native robot-brain chip makers to break the Nvidia grip on embodied AI, potentially reshaping the capital-compression arc for Chinese robotics OEMs facing high-cost GPU imports. The company's ability to maintain software ecosystem compatibility while reducing power consumption through neural coding conversion is the key technical claim that will be tested at tape-out, currently scheduled for Q2 2027.
Parthus co-founder Yin Jilei brings over 20 years of semiconductor experience from IBM, GlobalFoundries, MediaTek, and VIA, along with a prior startup exit as COO of Zhicun Technology. The team's pedigree from both Peking University academic research and commercial chip design suggests a plausible path from lab to product, though the two-year development cycle means the company will not face market validation until late 2027 at earliest. The seed round size — undisclosed but described as "hundreds of millions of yuan" (likely RMB 200-500M, or ~$28-70M) — places this below the threshold for cross-substrate capital-cycle significance, but the technology direction carries structural implications for the robotics chip segment and the broader China AI hardware ecosystem.