Weifan Intelligent (维泛智能)
Category: AI Chips / Semiconductors
China's first native robotics 'brain chip' company, a Peking University spin-off developing a unified neuromorphic-GPU architecture (BiGPU) for embodied intelligence, targeting replacement of Nvidia Jetson in Chinese robotics. Weifan Intelligent (维泛智能) was founded in 2025. The company is led by 殷积磊 (Yin Jilei). Based in Beijing, China. Team size: 11-50. Total funding raised: $68.6M (Seed, May 2026) + RMB 100M+/~$14M+ (Seed+, Aug 2026) — cumulative ~$82M+. Latest round: Seed. Key investors include 中关村资本 (Zhongguancun Capital), 启航投资 (Qihang Investment), 上海未来产业基金 (Shanghai Future Industry Fund), 石溪资本 (Shixi Capital), 佰维存储 (Biwin Storage), 燕创集团 (Yanchuang Group), 海益投资 (Haiyi Investment), 探元创投 (Tanyuan Venture Capital), 中金资本旗下基金 (CICC Capital fund), 九坤创投 (Ubiquant Ventures), 泽羽资本 (Zeyu Capital), 汕韩潮芯 (Shanhan Chaoxin).
- Founded
- 2025
- Headquarters
- Beijing, China
- Team size
- 11-50
- Total funding
- $68.6M (Seed, May 2026) + RMB 100M+/~$14M+ (Seed+, Aug 2026) — cumulative ~$82M+
Value proposition
First native Chinese robotics 'brain chip' that fuses neuromorphic (brain-inspired) computing with general-purpose GPU compute in a unified architecture (BiGPU), enabling low-power, high-efficiency AI inference for embodied intelligence while maintaining full software compatibility with mainstream AI frameworks.
Products and solutions
BiGPU (Brain-Inspired GPU) architecture — unified neuromorphic-GPU chip for embodied intelligence 'brain'; targets Nvidia Jetson replacement for Chinese robotics market; supports Attention Transformer, VLA, world models, and spiking neural networks
Unique value
Proprietary Brain-Inspired GPU (BiGPU) architecture that homogeneously fuses ANN and SNN computation into a single instruction set and toolchain, eliminating the complexity of heterogeneous solutions while achieving dramatic power reduction for robotics edge AI.
Target customer
Chinese humanoid robotics companies; embodied intelligence OEMs; edge AI and industrial automation providers seeking domestic alternatives to Nvidia Jetson
Industries served
Embodied Intelligence / Humanoid Robotics; Edge AI; Industrial Automation
Technology advantage
Homogeneous fusion of ANN (GPU) and SNN (neuromorphic) computation in a single architecture with shared instruction set and toolchain; converts >80% of GEMM operations from ANN to SNN-style accumulate computation for dramatic power reduction; incubated from Peking University PAICORE Lab with deep neuromorphic research heritage; first-mover in native Chinese robotics 'brain chip' market
How they differentiate
Unlike competitors using heterogeneous (separate SNN + NPU) approaches, Parthus uses a homogeneous fused architecture (BiGPU) that shares a single instruction set and software toolchain, deeply compatible with mainstream AI ecosystems. This reduces development complexity and ecosystem integration costs while achieving the power benefits of neuromorphic computing.
Main competitors
Nvidia Jetson (Jetson AGX Orin / Thor) — dominant incumbent; 天数智芯 (Iluvatar CoreX) with '彤央' (Tongyang) edge AI chips; SynSense (时识科技) — neuromorphic chip company; Lynxi Technologies (灵汐科技)
Key partnerships
Peking University PAICORE Lab (incubation and technology heritage); Collaborating with leading Chinese robotics companies (some projects in active cooperation stage)
Major milestones
Founded May 2025 as Peking University PAICORE Lab spin-off, completed several hundred million RMB seed round (May 2026), established municipal-level AI chip lab with Peking University and BUPT (May 2026), PAICORE 2.5 test chip completed tape-out and return, BrainOS operating system framework development completed, team expanded to ~100 people within six months, completed additional 'Seed+' round (>RMB 100M) in Aug 2026 to fund productization/commercial delivery, OmniRT deployment tool planned for release to strategic partners Sept 2026, mass-production chip tape-out still targeted Q2 2027
Market positioning
First-mover and currently the only native Chinese startup developing a purpose-built robotics 'brain chip' that fuses neuromorphic and GPU compute, positioned as a domestic alternative to Nvidia Jetson for the Chinese embodied intelligence market.
Geographic focus
China (domestic market for robotics brain chips; competing against Nvidia Jetson dominance and other Chinese GPU/neuromorphic chip startups)
Patents and IP
Patent filed for unified ANN-SNN software/hardware method and device (unified instruction format and address encoding for SNN and ANN); 5 patent filings (per Qichacha); 22 trademark registrations
About 殷积磊 (Yin Jilei)
Ex-COO & R&D VP at 知存科技 (Witmem); Chip R&D Director at IBM and GlobalFoundries; Chip R&D roles at MTK (MediaTek) and VIA Technologies; Peking University MS in Microelectronics; 20+ years semiconductor industry experience
Latest news about Weifan Intelligent (维泛智能)
More AI Chips / Semiconductors companies
Official website: https://www.36kr.com/p/3821371042877575