XeeVision Secures Nearly $70M to Mass Produce Brain-Inspired Chips

The big picture: Chinese brain-inspired vision chip company XeeVision has consecutively completed three rounds of financing, raising a cumulative total of nearly 500 million yuan (approximately $70M USD). The funding will primarily support the mass production and industrial deployment of its second-generation brain-inspired vision chips, next-generation technology iteration, customer adaptation, and team expansion.

Why it matters:

• Breaking the “impossible triangle”: XeeVision’s core technology, derived from Tsinghua University’s “Tianmou chip,” fundamentally redefines visual perception by breaking the traditional trade-offs between resolution, frame rate, and dynamic range, while significantly reducing data transmission and computation.
• Pioneering physical AI vision: The company’s brain-inspired complementary vision chips are designed to provide high-performance, low-power visual perception and understanding for the physical AI era, enabling intelligent systems to perceive and comprehend the world more like humans.
• Strong academic and industrial backing: XeeVision’s technology originates from globally recognized research published in Nature, and its diverse investor base includes government industrial funds, leading financial institutions, Tsinghua-affiliated capital, and strategic investors from the embodied AI and automotive sectors.

How it works:

• Complementary dual-pathway paradigm: The Tianmou chip utilizes a novel paradigm that mimics the retina’s multi-pathway mechanism, performing differentiated acquisition and complementary encoding of visual information (light intensity, spatial structure, temporal changes) at the perception front-end.
• Sensor-algorithm integration: Beyond the sensing chip, XeeVision integrates complementary vision with learnable algorithms, allowing the system to form more reliable visual representations in complex scenarios without perfect ground truth, evolving sensors into intelligent input layers for large and world models.
• Machine vision nervous system: XeeVision is extending its single-chip capability into a complete machine vision nervous system, where the Tianmou chip acts as the “retina,” a specialized transmission protocol as the “optic nerve,” and visual models as the “visual cortex” for understanding and reasoning, enabling native visual perception for physical world interaction.

The catch: While XeeVision boasts cutting-edge technology and strong academic credentials, the brain-inspired computing and AI chip market in China is intensely competitive, with numerous startups and established players vying for market share. Scaling mass production for a novel chip architecture and securing broad adoption across diverse industries like robotics and autonomous driving will require robust execution and significant capital, especially amid global supply chain complexities and potential export controls.