NdotLight wins up to $2.24M government grant for robot-training 3D data
What's the deal? South Korea's NdotLight has been selected for the 2026 Scale-up TIPS R&D (Specialized) programme, securing up to US$2.52M in government research funding over three years. The grant comes from the Ministry of SMEs and StartupsDealroom has a profile for this one. Try Dealroom → and the Korea Technology Information Service for SMEsDealroom has a profile for this one. Try Dealroom →, which back technology companies first vetted and funded by private operators.
What does the company do? NdotLight builds data infrastructure for physical AI. Its platform, TRINIXDealroom has a profile for this one. Try Dealroom →, uses a proprietary 3D CAD engine to generate 3D models from text and images, automatically infer their physical properties, and export them as "SimReady" assets for robot training and evaluation.
What's the money for? The company will use the funding to verify the physical properties of TRINIX-generated assets against real-world measurements, then correct any differences. The aim is to narrow the "Sim-to-Real Gap" — the discrepancy robots face when moving from virtual training to real environments.
Why now? For a robot to replicate learned movements in the field, simulation data must reflect real physical traits such as weight and friction. NdotLight, which previously won selection for TIPS and the Baby Unicorn programme, is now adding verification to a platform that so far focused on generating data.
What's the endgame? NdotLight plans to expand abroad, strengthening its role as a data supplier within Nvidia's physical AI ecosystem and targeting manufacturing and robotics markets in North America and beyond. The company was named a key partner during Nvidia chief executive officer Jensen Huang's keynote in May, has won a CES innovation award, and sits in Nvidia's partner ecosystem.
Chief executive officer Park Jin-young said the selection would help "grow TRINIX into core data infrastructure for the global robotics and physical AI market" (translated from Korean).
The signal: At roughly $2.24 million, the grant ranks in the 95th percentile among all-time grant rounds for the sector in South Korea, a sign of how much weight Seoul is placing on physical AI. As robotics firms race to close the gap between simulation and reality, verified training data is becoming its own category of infrastructure.
Read more: venturesquare.net
Image credit: NdotLight
