Shenzhen-based AI2 Robotics and X Square Robot Technology Co. Ltd. have both achieved valuations exceeding 50 billion RMB, or approximately $2.8 billion, following recent funding rounds. These capital injections underscore a growing investor appetite for embodied artificial intelligence, a field focused on integrating advanced machine learning models with physical hardware to navigate and manipulate the real world.
AI2 Robotics secured $735 million in new capital through a series of four funding rounds that concluded with a Series C. The company specializes in developing both the physical robotic platforms and the underlying vision-language-action models that provide spatial reasoning for its machines. Its flagship AlphaBot series utilizes a wheeled humanoid design, which the company argues provides superior stability and safety compared to traditional bipedal systems. The AlphaBot features over 34 degrees of freedom and a specialized waist-leg lifting mechanism, allowing it to perform high-precision tasks while maintaining a lower center of gravity. By avoiding the regulatory and mechanical complexities associated with bipedal locomotion, AI2 Robotics aims to accelerate the deployment of its machines across industrial, biotech, and retail environments.
The AlphaBot hardware relies on a sophisticated sensor suite that integrates high-resolution cameras and microphones to feed environmental data into the company’s proprietary Alpha Brain foundation model. This sensory input allows the robot to interpret complex, goal-based commands and manipulate objects without requiring manual teleoperation. The mechanical design incorporates a waist-leg lifting mechanism that provides an arm span of approximately 2.3 feet, facilitating reach and dexterity in constrained spaces. By prioritizing a wheeled base, the engineering team ensures the platform remains stable during high-speed movements, effectively mitigating the risk of tipping that often plagues bipedal humanoid designs in dynamic, real-world settings.
X Square Robot Technology Co. Ltd. has focused its development on the proprietary Wall AI model family and the Quanta humanoid series. The company distinguishes itself through the development of specialized data-capture tools, including exoskeletons and teleoperation interfaces designed to translate human movement into machine-readable data. This pipeline allows the firm to train its robots on complex environmental interactions, improving their performance in both household and industrial logistics settings. In September 2025, the company released Wall-OSS, an open-source version of its model family intended to foster community-driven development for various robotic form factors. The firm targets sectors currently facing labor shortages, aiming to automate repetitive physical tasks through its heavy-duty industrial variants and domestic service robots.
The Quanta robot series utilizes an advanced control architecture that processes sensory data through a specialized interface designed to understand human movement. By deploying exoskeletons during the data-capture phase, X Square engineers can map human kinematic patterns directly onto the robot’s actuators, ensuring more natural and efficient movement. These robots are equipped with high-torque motors capable of handling heavy lifting and precision assembly, making them suitable for both factory floor operations and complex household chores. The integration of this data pipeline allows the Quanta series to adapt to varied environments, providing a scalable solution for tasks that require both physical strength and environmental awareness.
The rise of these firms highlights a shift toward general-purpose robotics that prioritize functional stability over purely human-like movement. By utilizing wheeled bases, these companies address the immediate need for industrial-grade reliability in unpredictable environments. This engineering choice reflects a pragmatic approach to the current limitations of bipedal balance and the stringent safety standards required for machines operating in public or shared workspaces. The integration of vision-language-action models allows these robots to interpret complex, goal-based commands without the need for constant manual oversight.
Investors are betting that the convergence of foundation models and physical actuators will reduce cycle times in manufacturing or decrease labor costs in logistics. The ability to capture human movement data via exoskeletons provides a critical feedback loop for training, enabling robots to replicate nuanced physical actions. As these companies scale, the focus remains on refining the sensory-motor integration that allows machines to comprehend their surroundings in real time. The standardization of these robotic platforms could eventually lower the barrier to entry for widespread automation in labor-intensive industries.
The success of these funding rounds suggests that the market for embodied intelligence is maturing beyond experimental prototypes. Future development will likely hinge on the ability of these firms to demonstrate consistent performance in high-stakes industrial environments. Watchers of the sector will monitor the commercial adoption rates of the AlphaBot 2 and the Quanta series as they move from controlled testing into broader market applications. The ongoing competition between proprietary model development and open-source initiatives like Wall-OSS will also shape the trajectory of the market for industrial automation hardware.