A surge in investment is fueling rapid advancements in robotics as X Square Robot secures $140 million in Series A++ funding. The Shenzhen-based company, which operates under the name Variable Robotics Technology Co., focuses on developing embodied AI models—robots that learn from real-world physical interactions, not just digital simulations. The latest funding signals growing interest from both technology giants and investors in the potential of robotics to tackle practical tasks, from food delivery to logistics. With an expanding product lineup, including the Quanta X1 and the semi-humanoid Quanta X2 robots, X Square aims to bring intelligent robotics beyond controlled lab environments into everyday scenarios. The deployment of AI-driven robots in unpredictable real-world settings brings both technical hurdles and new opportunities for industries worldwide.
Earlier reports on X Square Robot largely centered on its initial launches and promising demonstrations in controlled environments. While robust AI integration was evident from early product trials, the scale and focus on actual field deployment across various industries have grown noticeably. Previous investments were already substantial, but this latest round, led by players like ByteDance and HongShan, highlights intensified investor confidence as physical robot AI becomes increasingly validated outside of lab tests. This expansion into full-stack research and mass production readiness sets X Square apart from many startups still focused on prototypes or software-only models.
How Do X Square’s Models Learn from the Physical World?
X Square Robot’s FOUNDATION AI platform, referred to as the “robot brain,” is designed to enable machines to adapt to everyday environments through a combination of large-scale reinforcement learning and continuous data collection. The WALL-A system integrates visual, language, and action-based reasoning, allowing robots to interpret and respond to complex sensory data in real time.
“At X Square, we believe the key to enabling robots to truly master real-world tasks lies in the ‘robot brain’—a foundation model for the physical world that parallels virtual LLMs to shatter generalization bottlenecks,”
stated Wang Qian, founder and CEO.
What Sets the Quanta X1 and X2 Robots Apart?
Deployed in demanding scenarios, the Quanta X1 has demonstrated its capabilities by autonomously delivering food and managing logistics tasks like sorting irregular-shaped packages. It leverages causal inference to navigate obstacles such as visual obstructions or physical deformities in its environment, ensuring completion of assigned missions without human intervention. The Quanta X2, a semi-humanoid variant, is positioned for broader applications owing to its versatile manipulation abilities. The company emphasizes the importance of building a comprehensive ecosystem, stating,
“X Square Robot continues to iterate across our three core pillars: models, data pipelines, and hardware.”
Why Are Investors Backing X Square’s Approach?
Major investments from ByteDance, HongShan, Alibaba Group, and Meituan support X Square’s focus on tightly coupled AI models and custom hardware tailored for industrial-scale production. The company’s hardware architectures are driven by constantly evolving model and data requirements, not the other way around. This approach appeals to partners seeking scalable solutions that translate directly to immediate, high-value applications in manufacturing, logistics, and healthcare sectors.
X Square’s innovative use of advanced data capture tools—such as teleoperation systems, exoskeletons, and its Universal Manipulation Interface—contributes to a closed data loop, continually improving the robots’ performance. The company’s open-source WALL-OSS also attempts to foster a wider developer ecosystem, giving outside contributors a platform to collaborate on embodied intelligence.
As robotics moves further into real-world industries, practical data-driven models like those from X Square Robot could help bridge AI’s current limits in physical cognition and manipulation. Lessons from their approach show that success in robotics at scale depends on deep integration across software, hardware, and continuous feedback, instead of focusing exclusively on one domain. For those looking to deploy intelligent robots, attention to real-world reinforcement learning, robust multimodal sensor integration, and closed-loop development pipelines will be critical. Businesses should also weigh the growing trend of open-sourcing core technologies, which could accelerate collective progress but pose challenges in maintaining competitive advantage and data quality.
