Tongxin Intelligent raises hundreds of millions RMB to build AGI-native inference chips
量子位 · wechat · 2026-11-16
Tongxin Intelligent, founded in April by AI professor Zhu Songchun and incubated by BIGAI, has raised seed and angel rounds totaling several hundred million RMB to build a new inference chip for AGI algorithms.
- Why now: models iterate faster than chip cycles; world models and embodied agents need graph reasoning, probabilistic/causal inference and action evaluation that GPUs handle poorly (context: AMD's $8.2B WorldLabs deal, NVIDIA pairing GPUs with Groq).
- Theory-driven design: based on Zhu's CUV framework (cognition, utility, values) and "small data, big tasks," the chip adds an AGI Core alongside Vector/Tensor units for parse-graph-based logic, causal and value reasoning.
- Engineering: CEO Wang Haiyang was a Loongson founding member and chief architect of Hygon CPUs; hybrid compute-storage near memory, reconfigurable units, Chiplet variants for cloud/edge/robot.
- Progress: architecture defined within 3 months, FPGA prototype validation in December; first product targets cost-effective cloud and edge inference, positioned as coexisting with GPUs.
More from Embodied
- Grid Robotics Proposes Agentic Robotics Paradigm Built on Embodied System 2 — Grid Robotics · 2026-11-16
- Shenzhen hosts APEC 2026, showcasing humanoid robots and its design-to-mass-production cycle — CyberRobooo · 2026-11-16
- Qiming's Shen Jin: Safety is the final hurdle for PhysicalAI mass adoption — 创业邦 · 2026-11-16
- Analog chip AX100 tape-out solves matrix equations in under 100ns; humanoid capacity scales — 创业邦 · 2026-11-16
- Fable 5.2 learns to run: 30-second 3D skeleton animation rendered in agent-focused Vork engine — FinanceYF5 · 2026-11-16
- Google's weekly blitz: multiple Gemini models, XR glasses, TPUs and agent tools — Saboo_Shubham_ · 2026-11-16