N.Light to Pitch at NVIDIA GTC: '3D Data' Generation Tackles the Robot Learning Bottleneck

3D AI startup NdotLight, led by CEO Park Jin-young, announced on the 16th that it has been selected as one of the companies to present at the NVIDIA Inception Demo Pitch at NVIDIA GTC 2026, the world’s largest AI conference, taking place in San Jose, California, from the 16th to the 19th.
The NVIDIA Inception Demo Pitch is an event that selects promising startups leveraging NVIDIA technologies in innovative ways and provides them with an opportunity to showcase their technologies and business models to investors, customers, and partners from around the world. At GTC 2026, NdotLight will introduce its core technology, TRINIX, through a five-minute live pitch.
NdotLight’s TRINIX is a solution that generates Sim-Ready 3D data that can be immediately used for robotics and Physical AI. Conventional, manual methods of creating 3D assets have been identified as a bottleneck in robot training due to the significant time and costs involved.
TRINIX addresses this challenge by combining NdotLight’s proprietary 3D CAD engine with generative AI models. By inputting text or images, users can automatically generate high-quality 3D data containing physical properties and joint information. The solution can also instantly convert existing legacy CAD files into OpenUSD, a format optimized for simulation.
NdotLight has further enhanced its solution by actively leveraging NVIDIA’s core technology stack. It uses Omniverse to build and expand OpenUSD-based data, Isaac Sim to conduct robot simulation and large-scale training workflows, and Cosmos to generate large-scale simulation episodes and build Vision-Language-Action (VLA) datasets.
Through these technologies, NdotLight has achieved a more than 70% reduction in data generation time compared with conventional methods, while large-scale GPU-based data generation has increased the success rate of robot training by more than threefold. More recently, the company was also selected as a technology supplier for the National Physical AI Foundation Model Consortium, led by the South Korean government, further demonstrating its technological capabilities.
NdotLight’s team consists of experts with backgrounds at global technology companies including Samsung Electronics, NAVER, and AMD. Through its participation in GTC 2026, the company plans to accelerate its efforts to enter the global Physical AI market, which is estimated to be worth approximately $63 billion.
Park Jin-young, CEO of NdotLight, said, “We are delighted to have the opportunity to showcase NdotLight’s technology on the global stage at NVIDIA GTC. Through TRINIX, we will accelerate digital transformation (DX) in manufacturing and become a key partner in solving the shortage of 3D data faced by robotics companies around the world.”
Choi Ji-ho, Reporter
jhochoi51@irobotnews.com
Source: Robot News
Robot News


