Simate
Category: Robotics / Embodied AI
Chinese Physical AI startup building an AI-native research stack (AutoResearch) and general-purpose physical fast system (Simate-beta/Sipai) for zero-shot embodied robot manipulation. Simate was founded in 2026. The company is led by 张颖 (Zhang Ying). Based in China. Team size: 11-50. Total funding raised: $40M. Latest round: Venture Round. Key investors include Undisclosed — reported to include autonomous-driving/embodied-AI focused VCs, major internet companies, and industrial capital.
- Founded
- 2026
- Headquarters
- China
- Team size
- 11-50
- Total funding
- $40M
Value proposition
Building "Physical RSI" (Recursive Self-Improvement for Physical AI) — an AI-for-Physical-AI research stack where AI agents autonomously propose hypotheses, run experiments, and iterate robot foundation models, aiming for zero-shot/few-shot general physical manipulation.
Products and solutions
Simate-beta (first general-purpose physical fast system model, topped RoboDojo leaderboard with avg score 33.95 / SR 27.96%), Sipai (pluggable model framework supporting World Model, World Action Model, VLA, VLM architectures), Sinfra (AI-native training/simulation/inference infrastructure), AutoResearch (autonomous research platform, in beta with researchers from MIT, Caltech, Tsinghua, Peking Univ, HKUST), RoboScientist (AI scientist for experiment-to-insight).
Unique value
World's first company to systematically implement RSI (recursive self-improvement) into Physical AI; combines top-tier autonomous-driving mass-production engineering (FSD-level end-to-end) with frontier world-model research and humanoid control talent.
Target customer
Robotics/embodied AI developers, research institutions, and enterprises deploying general-purpose physical manipulation systems.
Industries served
Robotics / Embodied AI, Physical AI research, autonomous systems
Technology advantage
4D physical perception + hierarchical temporal memory for a "general physical fast system"; AI-native infra enabling parallel execution of dozens of independent research lines with high-frequency filtering (world model + simulation pre-screening before real-robot testing); SiPAI pluggable model framework making architectures "AI-researchable"; open-sourced AutoResearch platform.
How they differentiate
Focuses on automating the research process itself (AI for Physical AI) rather than hand-crafted task demos; leverages FSD-level end-to-end mass-production engineering experience; open-sources core internal research tooling to attract external researchers.
Main competitors
Galaxea AI (星海图), SynapX (章鱼动力), other Chinese embodied-AI foundation model startups (e.g., 自变量机器人, 银河通用)
Key partnerships
AutoResearch beta program with researchers from MIT, Caltech, Tsinghua University, Peking University, HKUST, open-sourced AutoResearch research platform
Major milestones
Founded July 2026, completed multiple rounds of hundreds-of-millions-RMB financing within ~2-3 months, topped RoboDojo physical AI leaderboard (Sept 2026) with Simate-beta, launched AutoResearch beta platform, plans staged open-source releases and year-end 2026 milestone results.
Growth metrics
Founded July 2026; ~3 months to #1 on RoboDojo; multiple funding rounds totaling hundreds of millions RMB; AutoResearch beta with top-university researchers.
Market positioning
Emerging Chinese Physical AI challenger; self-reported #1 on RoboDojo physical AI benchmark (surpassing GPT-6-Astra and DeepMind entries), achieved within 3 months of founding.
Geographic focus
China (primary); global Physical AI / embodied intelligence race
About 张颖 (Zhang Ying)
Ex-Horizon Robotics (地平线) core technical lead of HSD autonomous driving system; one of the youngest technical experts at Alibaba P10-equivalent level; led Horizon's intelligent driving through three generations (map-based → mapless → pure-vision one-stage end-to-end), the domestic system closest to Tesla FSD; two-time consecutive O-level (top) performance in Horizon's 1000-person HSD team.
Latest news about Simate
More Robotics / Embodied AI companies
Official website: https://mate-robot.cn/