Lately, LightWheel accomplished $145 million in A++ and A+++ funding rounds. This spherical launched a number of industrial companions and monetary establishments, together with New Hope Group, Dingbang Funding (the household workplace of the chairman of San’an Optoelectronics), AUX Group, and Dingshi Asset Administration as strategic traders, alongside monetary traders resembling Jianto Huake, Guofang Innovation, Daohe Lengthy-Time period Funding, and Qingxin Capital.
Following this financing, LightWheel has develop into the world’s first unicorn firm within the embodied knowledge sector. The newly raised capital will primarily be used to additional develop its bodily simulation engine, improve large-scale mannequin analysis techniques, and strengthen international supply and localized deployment capabilities, reinforcing LightWheel’s management in bodily AI knowledge and simulation infrastructure.
The Bodily AI Period: New Infrastructure Is Taking Form
Over the previous decade, computing energy has outlined the core infrastructure of AI. GPU clusters supported the rise of deep studying, and huge fashions quickly emerged throughout the digital world.
At the moment, nonetheless, AI is present process a deeper transformation: transferring from the digital world into the bodily world.
As robots start to function in actual environments, the important thing bottleneck is now not {hardware} alone, however the potential to scale embodied knowledge. The trade is coming into what many describe as the primary 12 months of embodied intelligence knowledge, with demand for such knowledge rising exponentially.
At this stage, a brand new kind of infrastructure is rising—not simply computing energy, however knowledge and simulation infrastructure.
LightWheel positions itself as a builder of this new era of infrastructure.
Three-Layer Structure: A Scalable Engine for Embodied Knowledge and Simulation
By way of product structure, LightWheel has established a scalable embodied knowledge and simulation engine. Constructed round a **three-layer construction—World, Habits, and Eval—**the system covers the complete pipeline from bodily simulation and large-scale knowledge era to mannequin analysis.
World Layer
The World layer is predicated on LightWheel’s totally self-developed structure that integrates fixing, measurement, and era.
Its proprietary solver helps high-precision real-time simulations throughout a number of bodily techniques, together with inflexible our bodies, delicate our bodies, and fluids. Mixed with its bodily measurement manufacturing facility, a virtual-to-real benchmarking methodology, and large-scale non-rigid asset era capabilities, LightWheel has created a closed-loop system that connects real-world physics with scalable knowledge era.
Habits Layer
The Habits layer powers what LightWheel describes because the world’s largest non-embodied knowledge engine, enabling large-scale embodied knowledge manufacturing by means of two main sources:
- Simulation-generated knowledge
- Human video datasets, resembling EgoSuite
LightWheel’s knowledge options have already been delivered at scale globally, offering high-frequency, scalable knowledge gas for embodied basis fashions.
Eval Layer
On the analysis degree, LightWheel launched RoboFinals, the trade’s first high-difficulty, industrial-grade simulation analysis platform designed for cutting-edge giant fashions.
This platform helps set up standardized benchmarks for embodied intelligence, accelerating the event of a scalable analysis ecosystem throughout the trade.
Collectively, these three layers kind a self-reinforcing knowledge flywheel:
- Actual-world bodily measurements and EgoSuite human video knowledge feed simulations to take care of bodily realism
- Correct physics simulations then drive large-scale evaluations by means of RoboFinals
- Insights from evaluations information the following part of information assortment
With each cycle, simulation turns into extra correct, knowledge turns into simpler, and analysis turns into extra insightful. By means of this flywheel, LightWheel positive factors management over core embodied knowledge sources, constructing a whole knowledge and simulation infrastructure spanning coaching to analysis.
Industrial Breakthrough: Three International Supply Leaderships
On the commercialization aspect, LightWheel has already achieved international management in three key supply areas:
- Simulation-generated knowledge
- Simulation-based analysis
- Human video datasets
LightWheel is presently the one firm on the earth able to delivering all three capabilities at scale.
By way of income development, the corporate recorded 10× income development in 2025, and its projected income for Q1 2026 alone is predicted to exceed its complete income for 2025.
Its companions embrace main organizations resembling NVIDIA, Google, Determine AI, 1X Applied sciences, ByteDance, Alibaba, Zhiyuan Robotics, Galaxy Normal Robotics, Toyota, Bosch, BYD, and Geely.
All high 5 international world-model analysis groups have already partnered with LightWheel. In the meantime, over 80% of the simulation belongings and artificial knowledge utilized by main embodied AI groups worldwide come from LightWheel.
Past commercialization, LightWheel can be serving to construct trade infrastructure:
- Co-developed the Isaac Lab-Area benchmark framework with NVIDIA
- Partnered with World Labs to construct a normal pipeline from setting era to large-scale analysis
- Collaborated with Tongyi Qwen to create an industrial analysis loop based mostly on RoboFinals-100
- Its LeIsaac simulation workflow has been included within the official documentation of Hugging Face, turning into a extensively adopted simulation customary amongst international builders
Constructing Knowledge and Simulation Infrastructure for Bodily AI
Each paradigm shift in AI has been pushed by infrastructure breakthroughs. GPUs and CUDA outlined the computing infrastructure of the deep studying period.
Within the period of bodily AI, a brand new infrastructure is rising—knowledge and simulation infrastructure for the bodily world.
LightWheel founder Dr. Chen Xie defined:
If GPUs and CUDA outlined the computing infrastructure of the large-model period, then LightWheel goals to outline the info and simulation infrastructure of the bodily AI period.
The period of large-scale embodied knowledge enlargement has arrived, and LightWheel is constructing the long-term infrastructure that helps world modeling, knowledge era, and clever evolution.
Supply: GeekPark
Elevate your perspective with NextTech Information, the place innovation meets perception.
Uncover the newest breakthroughs, get unique updates, and join with a world community of future-focused thinkers.
Unlock tomorrow’s traits right this moment: learn extra, subscribe to our publication, and develop into a part of the NextTech neighborhood at NextTech-news.com
