Leidi Network, March 11

Guanglun Intelligent has completed a 1 billion yuan A++ and A+++ round of financing. This round of financing has introduced multiple industry scenario parties and financial institutions, including New Hope Group, Dingbang Investment (the family office of the chairman of SANAN), AUX, Dingstone Asset Management, and other industrial investors, as well as Jiantou Huake, Guofang Innovation, Daohe Long-term Investment, and Qingxin Capital among financial investment institutions.

Guanglun Intelligent stated that after this round of financing, it has become the world's first unicorn enterprise in the field of embodied data. The funds will be primarily invested in the continuous research and development of physical simulation engines, the upgrading of large-scale model evaluation systems, and the construction of global delivery and local deployment capabilities.

Guanglun Intelligent noted that over the past decade, computing power has defined the infrastructure of AI. GPU clusters have supported deep learning, and large models have rapidly emerged in the digital world. Today, AI is undergoing a more profound leap: from the digital world to the physical world.

As robots begin to enter real environments, the real bottleneck is no longer just hardware, but the capability for data scaling. The era of embodied intelligent data has arrived, and the demand for embodied intelligent data is growing exponentially. At this stage, a new infrastructure is forming: no longer just computing power, but data and simulation infrastructure. Guanglun Intelligent is the builder of this new infrastructure.

Three-layer architecture: Scalable engine for embodied data and simulation

In terms of products, Guanglun has successfully established a scalable data and simulation engine for embodiment. With a three-layer architecture centered on World, Behavior, and Eval, it covers the complete link from physical real simulation, large-scale data production to model capability evaluation.

? World layer: Guanglun Intelligent has developed a full-stack self-research technology architecture centered on solving, measuring, and generating. The self-developed solver supports high-precision real-time solving of multiple physical fields such as rigid bodies, soft bodies, and fluids, combined with the innovative physical measurement factory and the virtual-real benchmarking methodology, as well as large-scale non-rigid asset production capabilities, forming a complete self-research closed loop from physical reality to data scaling.

? Behavior layer: Guanglun Intelligent has built the world's largest non-embodied data engine, covering two major paths of simulation synthetic data and human video data (EgoSuite), achieving large-scale production of embodied data. Guanglun Intelligent's data solutions have been delivered at scale globally, continuously leading the industry in production capacity, providing high-frequency scalable data fuel for embodied foundational models.

? Eval layer: Guanglun Intelligent has launched the industry's first simulation evaluation platform, RoboFinals, which is sufficiently challenging, meets industrial standards, and supports cutting-edge large models. It has established industry standards for embodied intelligent evaluation, promoting the large-scale implementation of the industry evaluation ecosystem Guanglun Intelligent claims that three layers constitute a self-reinforcing data flywheel: physical measurements from the real world and continuous input of EgoSuite human video data ensure physical authenticity; physically accurate simulations then drive RoboFinals to complete large-scale evaluations, with insights from these evaluations feeding back into the next steps of data collection—each rotation of the flywheel results in more precise simulations, more effective data, and deeper evaluations. Guanglun thus masters the initiative in data, creating a complete data and simulation infrastructure for embodied intelligence from training to evaluation.

Guanglun Intelligent points out that it expects a tenfold increase in annual revenue by 2025, with Q1 2026 projected to exceed the total revenue of 2025.

Partners include Nvidia, Google, Figure AI, 1X Technologies, ByteDance, Alibaba, Zhiyuan Robotics, Galaxy Universal Robotics, Toyota, Bosch, BYD, Geely, and other leading teams in large models, robotics, and industries. The top five world model teams globally have all collaborated with Guanglun, and over 80% of the simulation assets and synthetic data from major international embodied intelligence teams come from Guanglun Intelligent.

Dr. Xie Chen, founder of Guanglun Intelligent, stated: "If GPUs and CUDA defined the computational infrastructure of the large model era, then Guanglun Intelligent is defining the data and simulation infrastructure of the physical AI era."

———————————————