X Sq. Robotic Know-how Co. right now mentioned it has closed 4 consecutive financing rounds, culminating in a Collection C. These rounds deliver the embodied AI and basis mannequin developer’s valuation to greater than $2.8 billion.
The Shenzhen, China-based firm mentioned it is going to use the funding to additional put money into foundational analysis and core applied sciences. X Sq. Robotic mentioned it plans to advance towards general-purpose embodied AI.
“Since Day 1, X Sq. Robotic has targeted on in-house growth of basis fashions, pursuing a difficult however obligatory path,” acknowledged Wang Qian, founder and CEO of X Sq. Robotic. “In the present day, our investments in embodied AI fashions; a scalable, model-driven, high-quality information pipeline system; and real-world deployment are starting to ship clear outcomes.”
Based in 2023, X Sq. Robotic develops “end-to-end” embodied AI techniques. Slightly than depend on conventional rule-based automation, the corporate mentioned its strategy permits robots to adapt to altering environments and generalize throughout a variety of duties.
The financing brings collectively strategic and monetary buyers, together with main expertise firms, industrial companions, and enterprise capital companies, the corporate mentioned. IDG participated within the Collection C spherical, whereas HongShan and Xiaomi have backed X Sq. in a number of earlier rounds.
X Sq. Robotic focuses on end-to-end autonomy
X Sq. Robotic mentioned it’s constructing a full-stack embodied AI system. Its system combines basis fashions, robotics {hardware}, a proprietary data-pipeline system, and real-world deployments. At its core is a general-purpose embodied AI mannequin designed to allow robots to understand, motive, and act in complicated bodily environments.
In April 2026, the corporate launched WALL-B, a basis mannequin constructed on its World Unified Mannequin structure. Not like modular vision-language-action (VLA) approaches, WALL-B trains notion, language, motion, and bodily prediction inside a unified community. This permits stronger multimodal understanding, spatial reasoning, and continuous studying from real-world interactions, in line with X Sq..
The corporate has additionally open-sourced WALL-OSS-0.5 and WALL-WM, extending its unified strategy to robotic manipulation and world modeling. WALL-OSS-0.5 achieved over 80% autonomous completion on 4 of 17 real-robot duties with out post-training, X Sq. mentioned.
WALL-WM introduces event-level prediction by aligning language, imaginative and prescient, and motion information round significant occasions. This permits stronger cross-modal studying and physical-world prediction throughout reasoning, manipulation and generalization duties.
To speed up mannequin growth, X Sq. Robotic has constructed a scalable, model-driven information pipeline spanning automated information assortment, cleansing, annotation, high quality management, and augmentation. When mixed with real-world deployment, the corporate claimed that its system permits fast mannequin iteration whereas creating high-quality datasets for complicated, long-tail eventualities.
X Sq. takes on the problem of house robots
X Sq. Robotic is deploying its mannequin and {hardware} stack throughout family, industrial, and logistics eventualities. Amongst these, family settings characterize one of the complicated and essential testbeds for adoption and deployment.
On this space, the corporate has partnered with 58.com to launch an AI-powered cleansing service in Shenzhen and Beijing, the place robots work alongside individuals in actual residential environments. Since Might, X Sq. Robotic has additionally launched the “X Household Member Program,” the place robots reside with customers’ households for as much as one month as family companions, responding to a broader vary of on a regular basis wants.
Collectively, these initiatives deliver embodied robots past staged demos and into actual houses and on a regular basis life, mentioned X Sq. Robotic. The corporate asserted that these deployments create a steady suggestions loop wherein operational information improves mannequin efficiency, serving to speed up progress towards general-purpose embodied intelligence.
“As AI strikes past digital experiences into the bodily world, progress will depend upon shut integration between fashions, information, and robotics,” Wang mentioned. “We’re constructing that basis so embodied AI can turn into a part of on a regular basis life.”

