- The embodied AI startup plans to raise 1.5 billion yuan in its first independent funding round and complete a second round by end-2026
- Robbyant is betting that the next robotics race will be won by AI models that can power multiple machines, not by hardware alone
Ant Group’s embodied intelligence unit Robbyant (蚂蚁灵波) has launched its first independent fundraising round as the company seeks to accelerate development of a general-purpose AI “brain” for robots, marking a new push by China’s technology giants into physical AI.
Robbyant said on August 3 that it plans to raise 1.5 billion yuan ($222 million) in the round and complete a second financing round by the end of 2026.
If completed on schedule, the fundraising pace would rank among the fastest in the rapidly expanding embodied AI sector.
The company said it will continue focusing on foundational technologies for general-purpose robotic intelligence, increasing investment in embodied AI-native technology routes and accelerating commercial deployment.
LingBot model series
Unlike traditional robotics companies that develop complete machines, Robbyant has positioned itself as an intelligence-layer provider, following a “build the brain, not the body” strategy.
The company aims to create a universal AI system that can be adapted across different robot platforms.
Robbyant has released more than 10 models covering key capabilities from perception to action, including the embodied foundation model LingBot-VLA, embodied action model LingBot-VA, spatial vision model LingBot-Vision and world model LingBot-World.
Its LingBot-VLA 2.0 model, designed around a “one brain, multiple robots” architecture, has been pretrained on more than 20 robot configurations from 17 manufacturers, including Unitree (宇树科技), AgiBot (智元机器人), Galbot (银河通用) and Leju Robotics (乐聚机器人).
The company said the model demonstrates the ability to transfer intelligence across different robot forms, a key challenge for the industry.
Robbyant’s open-source models have also gained traction among developers. As of July, its GitHub projects had accumulated nearly 30,000 stars, about 2.7 times the level of its closest competitor, according to the company.
From hardware to intelligence
The fundraising comes as embodied AI enters a new phase of competition. After years of focus on robot hardware, the industry is increasingly shifting toward the development of more capable “brains” that allow machines to understand environments, plan tasks and generalize across scenarios.
Real-world data, model scalability and deployment efficiency are emerging as the key competitive advantages.
Robbyant is leveraging Ant Group’s accumulated capabilities in large-scale AI inference, consumer applications and data infrastructure to pursue a software-first approach.
The company believes that a universal intelligence layer can eventually support a wide range of robot platforms, from humanoids to industrial machines.
This strategy contrasts with many robotics startups that focus on building proprietary hardware as their core moat.
A platform play
Robbyant’s independent fundraising marks a broader shift as major technology companies move deeper into embodied intelligence.
Ant Group has previously invested strategically in 12 robotics companies, and the decision to bring its internal embodied AI team into the external capital market signals ambitions to build an ecosystem around a general-purpose robotic brain.
The move also reflects a global trend among technology giants. As AI moves beyond screens into the physical world, companies with strengths in computing, data and AI infrastructure are seeking to replicate their digital-era advantages in robotics.
For global investors, Robbyant offers a glimpse into a different path for physical AI development: rather than competing on mechanical engineering alone, companies may increasingly fight for control of the intelligence layer that enables robots to operate across different environments.
The next stage of the robotics race may not be determined by who builds the most advanced machine, but by who develops the most adaptable brain.
