Independent Variable Robotics Raises $1 Billion from ByteDance, HSG, and Others to Build Foundation Models for the Physical World
The only embodied artificial intelligence company in China to have received investment from all three major tech giants — Meituan, Alibaba, and ByteDance
Unity Ventures portfolio company Independent Variable Robotics has recently completed a RMB 1 billion Series A++ funding round. This round was jointly backed by top-tier investors and diversified local government platforms including ByteDance, HSG, Beijing Information Industry Development Fund, Shenzhen Capital Group, Shenzhen Nanshan Strategic Emerging Industry Investment, and Xi Chuang Tou. Sources indicate this is also the first investment made by Shenzhen Capital Group's AI fund since its establishment.
Notably, aside from ByteDance, Independent Variable has previously secured investments from Meituan and Alibaba, making it the only Embodied Artificial Intelligence company in China to have received backing from all three major internet giants.
The coordinated bets from cross-sector investors highlight both the capital markets' collective consensus on the importance of embodied foundation models and their deep recognition of Independent Variable's technological leadership and growth potential.
Building Foundation Models for the Physical World, Enabling Robots to Actually Work
Over the past two years, Embodied Artificial Intelligence has continued to capture market attention. The "body" — robot locomotion and control capabilities — has made significant strides. The industry's competitive focus has shifted from "limbs" to the "brain." The key breakthrough lies in how to build an intelligent "brain" for robots that can understand the physical world, manipulate objects, and flexibly adapt to complex, ever-changing scenarios, enabling them to truly perform diverse real-world tasks.
Embodied intelligence foundation models are physical-world foundation models that operate independently of and in parallel to virtual-world foundation models such as large language models and multimodal models. The core of foundation models lies in breaking through generalization and versatility bottlenecks. The complexity of physical reality demands that robots possess the ability to process unstructured, dynamic, and stochastic tasks in real time. Independent Variable's embodied foundation model takes all robot sensory information (such as vision, touch, voice, etc.) as input and directly outputs robot actions, vision, and language.
Qian Wang, founder and CEO of Independent Variable Robotics, stated: "The next phase of competition in embodied intelligence is fundamentally about the competition between foundation models built on data closed loops and model evolution capabilities." Under this assessment, the world is accelerating investment across data, models, and compute to rapidly advance embodied intelligence.
- Deep Integration of VLA and World Models, Real-Robot Reinforcement Learning for Autonomous Evolution
Independent Variable's self-developed WALL-A model features a core architecture that pioneers the deep integration of VLA (Vision-Language-Action) and world models. As a native multimodal input-output architecture, WALL-A is the first to achieve embodied multimodal chain-of-thought. WALL-A leverages world model mechanisms for spatiotemporal state prediction, collaborates with visual causal reasoning to understand environmental feedback, and internalizes physical commonsense from data through learnable memory mechanisms.
This fusion mechanism significantly enhances robots' zero-shot generalization capabilities for mobile manipulation tasks in unstructured environments.
Meanwhile, through large-scale real-robot reinforcement learning, the foundation model further acquires high-quality learning experience through interaction with the real physical world, autonomously solving long-tail problems and enabling continuous capability evolution.
Independent Variable has built a closed technical loop of physical-world foundation model to real-robot autonomous evolution through a fully end-to-end technical approach.
- High-Quality Real-Robot Data, Building the Model Evolution Engine
Data is the core fuel for foundation model evolution. Since its founding, Independent Variable has made heavy investments, adhering to a closed-loop iteration of hardware-data-model.
As one of the earliest companies in China to scale real-robot data collection, Independent Variable has self-developed multiple data collection devices including master-slave teleoperation, exoskeletons, and body-free systems, achieving data validation and model breakthroughs across various collection devices.
The company has also built a model-driven data pipeline that continuously produces scaled, high-quality data through data generation, filtering, augmentation, and annotation.
Independent Variable insists on using the foundation model to provide feedback to every stage of data processing and hardware design, iterating higher-quality data and more efficient collection devices to further improve foundation model performance.
- Model Iteration Drives Capability Leaps, Autonomous Task Completion in the Real World
Continuous model evolution has enabled Independent Variable's robots to demonstrate exceptional adaptability in real-world scenarios.
As the world's first mobile manipulation example based on a physical-world foundation model to successfully bridge outdoor and indoor scenarios, the robot in food delivery and cardboard recycling tasks — facing strong wind interference or visual occlusion — relies on the foundation model's generalization capabilities and the world model's causal reasoning. It can not only mentally reconstruct the full appearance of occluded objects like humans, but also autonomously correct errors through reinforcement learning strategies when stuck, completing task closed loops without human intervention.
This adaptability is also evident in complex, challenging real-world logistics scenarios. Facing chaotically stacked packages, the robot leverages the foundation model's zero-shot generalization to identify irregular items and uses reinforcement learning to rapidly adapt to work rhythms.
Notably, the evolution of Independent Variable's foundation model has also unlocked the potential of high-degree-of-freedom dexterous hands, with robots autonomously mastering human-like skills such as in-hand reorientation — from using tools to dealing cards, actions requiring extremely precise fingertip force control — successfully conquering the final centimeter of embodied intelligence fine manipulation.
While continuously pushing technological boundaries and focusing on frontier exploration, in September 2025, Independent Variable also open-sourced its self-developed end-to-end embodied foundation model WALL-OSS, promoting open access and popularization of embodied intelligence technology.
From Full-Stack Self-Development to Multi-Industry Deployment, Unlocking the Key Path to Model-Driven Commercialization
Independent Variable adheres to full-stack self-development of software and hardware. Starting from model algorithms and data-driven requirements, it has deeply defined robot hardware architecture, designed and released two high-performance robot bodies — "Quantum One" and "Quantum Two" — and simultaneously achieved full self-development and deep algorithmic adaptation of core components including robotic arms, joint modules, power drivers, and main controllers, significantly reducing overall system costs and laying a solid foundation for scaled production and commercial deployment of embodied intelligence robots.
Currently, Independent Variable has gradually entered multiple high-value sectors including industrial manufacturing, logistics, and elder care. Cross-industry applications demonstrate that Independent Variable's robots are precisely addressing real market commercial demands with high generalization and low-cost deployment capabilities.
Qian Wang has stated on multiple occasions that in the frontier track of embodied intelligence, one should strive to be a leader, not a follower. Independent Variable continues to deepen its efforts in three core areas — model iteration, data pipelines, and robot hardware — and through solid technical accumulation and full-stack self-development capabilities, constantly breaks through existing capability boundaries.
Going forward, Independent Variable will continue to use leading model capabilities as a lever to unlock the deeper forces of embodied transformation, fully releasing the technological value of embodied intelligence in industrial applications, promoting scaled application of model-driven embodied intelligence across thousands of industries, and injecting new momentum into industrial upgrading and productivity leaps.
The Map of a New World Is Unfolding | Unity Ventures New Year
