Chinese AI tactile sensing developer Tashan Technology recently announced the completion of a multi-million Series B funding round. Backed by strategic investors including Joyson Electronics, Taiping Innovation, and AUX Group, the company has established a full-stack technological ecosystem encompassing chips, #sensors, algorithms, and scene-specific applications.
Tactile perception has evolved into a critical requirement for the #robotics industry. Data indicates that the penetration rate of tactile sensors in robotic dexterous hands has surged from roughly 20% to over 60% in 2025. Tashan Technology’s core innovations include the world’s first hybrid analog-digital AI tactile chip, built on a spiking neural network architecture, enabling low-latency, low-power edge processing. Its flagship products, such as the TS-F fingertip sensor, offer a force resolution of 0.01N and are supported by a vision-tactile fusion training platform.
Notably, Tashan Technology is the world’s first tactile simulation partner for NVIDIA’s Isaac Sim. By the end of May, the company’s orders on hand had reached four times its total revenue for the previous year. It expects robotics-related business to account for two-thirds of its annual revenue by 2026. Currently, the firm collaborates with over 180 industry players, including Agibot and Galbot, commanding over an 80% market share in the humanoid robot tactile sensor niche.
[AgentUpdate Depth Analysis] Tactile perception serves as the critical "final mile" for embodied AI to transition from controlled lab settings into the chaotic physical world. Currently, most AI agents remain overly reliant on visual feedback, lacking the fine-grained haptic understanding of material texture, friction, and force required for dexterous manipulation. Tashan Technology's full-stack approach—integrating hardware sensors with simulation platforms like Isaac Sim—essentially builds a nervous system for AI agents. Unlike fragmented component suppliers, Tashan provides a closed-loop data collection and algorithmic standard. By leveraging spiking neural networks to minimize computational overhead and collaborating with Turing Award winner Richard Sutton on a "robot kindergarten," the company is pioneering a new paradigm for autonomous tactile learning. In the long term, as tactile data emerges as the next frontier for multimodal LLM training, companies mastering the intersection of physical sensing and high-fidelity simulation will define the foundational infrastructure for the entire robotic ecosystem. This "hardware + data + simulation" flywheel creates a significant barrier to entry that will dictate the competitive landscape of the humanoid era.
