On October 22, Huixi Intelligence, a Beijing-based developer of high-computing-power chips for embodied AI, said it had closed Pre-B+ and B rounds, bringing its total equity financing in 2026 to more than 2.5 billion yuan.

The round was led by the Beijing Artificial Intelligence Industry Investment Fund, with participation from government-backed and market-oriented investors including the Beijing Robotics Industry Development Investment Fund, Hidden Hill Capital, Loyal Valley Capital, Golden Harbor Investment, Jiantou Investment, Huatai Baoli, Huimu Yuanlun, the Shanghe Momentum Fund and the National Artificial Intelligence Industry Investment Fund. Existing backers also increased their stakes, among them Shunwei Capital, Vision Plus Capital, Cathay Capital, SenseTime's Guoxiang Capital, Bohua Industrial Investment, Guoqi Investment, Longmafeng Venture Capital, 37 Interactive Entertainment, Shengyi Capital, Xinglian Capital and Zhuoyuan Asia Capital.

The company said the proceeds will go mainly toward developing its next generation of high-computing-power chips for embodied AI, speeding up volume deliveries and building out its industry ecosystem.

A Hard Résumé Behind the 2.5 Billion Yuan

Founded in April 2022 and headquartered in Beijing's Yizhuang district, Huixi Intelligence positions itself as a supplier of computing platforms for embodied AI.

Founder, chairman and CEO Xu Ningyi earned his bachelor's, master's and doctoral degrees from Tsinghua University's Department of Electronic Engineering. He went on to lead the hardware computing group at Microsoft Research Asia, where the custom accelerator system he spearheaded became Microsoft's first such deployment in hyperscale data centers. He later served as chief architect and chair of the technical committee at Baidu's AI chip unit, where he led the design of Kunlun, China's first cloud AI chip to support both training and inference, and drove its first large-scale industrial deployment.

Xu subsequently served as chief scientist of SenseTime's custom computing team in an advisory capacity, and as a tenured professor at Shanghai Jiao Tong University's Tsingyuan Research Institute, where he heads the custom computing center. In April 2022, he founded Huixi Intelligence with fellow Tsinghua alumni Zhang Jianyong, He Guanghui and others.

Co-founder Zhang Jianyong was previously assistant vice president of autonomous driving at NIO, where he led development of the automaker's first-generation driver assistance features and served as platform engineering lead for its second-generation autonomous driving system. He oversaw the world's first production launch of EyeQ4-based features and took Orin-platform capabilities from engineering samples to the world's first production vehicle launch.

The other co-founder, He Guanghui, is a professor at Shanghai Jiao Tong University's School of Electronic Information and Electrical Engineering, with research spanning AI chips and computer architecture.

Three Building Blocks: RPU, RISE and RhinoVLA

The company's technology stack rests on three building blocks, under a design philosophy it sums up as "the data closed loop defines the chip."

The RPU is the compute core, supplying the computing power for multimodal perception, task understanding, reasoning and decision-making, and action generation.

RISE is a computing platform built around the company's in-house Guangzhi R1 chip and rated at 500 TOPS. The Guangzhi R1 itself is a high-performance AI SoC built on a general-purpose programmable architecture.

RhinoVLA supplies standardized pre-trained foundation models and a complete development pipeline.

The company's production-ready solutions currently address three markets: embodied AI, L4 autonomous logistics vehicles, and intelligent driver assistance for passenger cars.

Chips on Robots First: 17,625 Operations in Six Days

On the engineering side, Huixi Intelligence has built a software stack covering model frameworks, compilers, operator libraries, runtime, middleware, operating system, simulation and real-time control. Through hardware-software co-optimization, it aims to lower the barrier for robotics companies moving from algorithm development to deployment on physical machines.

In 2025, a computing platform based on the Guangzhi R1 chip was deployed in AgiBot's Genie G2 robot, which the company describes as a decisive step from chip development and validation to commercial mass production.

In June 2026, several AgiBot Genie G2 robots running on Guangzhi R1 chips completed 17,625 operations during six days of continuous, livestreamed work on a production line, with a task success rate of 99.99%.

Huixi Intelligence has now partnered with several leading embodied AI companies, with products deployed in industrial manufacturing, smart retail, commercial food service and autonomous logistics. Shipments of its Huixi Embodied line of computing products continue to grow.

Four Years In: Xiaomi, SenseTime and AgiBot Have All Invested

Huixi Intelligence has raised several rounds in its four years of existence.

Its angel round closed in December 2022, led by Xiaomi, at more than $50 million (roughly 330 million yuan, according to the original report).

A Series A followed in September 2024, with investors including Yizhuang Capital, Longmafeng Venture Capital and SenseTime.

In February 2026, the company closed a Pre-B round of more than 800 million yuan, in which AgiBot took a strategic stake alongside the Liangxi Digital Fund of Funds, Beijing State-Owned Capital Management, Shunwei Capital, Bohua Capital and Guoqi Investment.

On the industry side, the company has launched the Huixi Embodied AI Computing Ecosystem Alliance with partners up and down the supply chain, linking resources across robot hardware, algorithms and models, sensors, and foundational software and hardware.

The company said it will treat humanoid robots as its initial entry point, build around a self-controlled computing platform, and work with partners across the supply chain to create an open, collaborative ecosystem for embodied AI.