N.Dlight (엔닷라이트), a South Korean physical AI data company, has raised 15 billion won (~$10.5 million...
The AMW Read
Incremental funding round for a known player in a expanding but early-stage segment; no new strategic pattern or debate resolution.
N.Dlight (엔닷라이트), a South Korean physical AI data company, has raised 15 billion won (~$10.5 million) in a Series B round led by KDB Industrial Bank. Existing investors Naver D2SF, IMM Investment, and Capstone Partners participated, joined by new investors NH Venture Investment, Ourcrowd, and Stonebridge. The company develops TRINIX, a 3D asset generation solution that creates physics-aware virtual environments for robot training, addressing the sim-to-real gap by automating the creation of data that includes mass, friction, joint structure, and collision parameters. N.Dlight is integrated with NVIDIA's Omniverse platform, was named a Physical AI Ecosystem Partner at GTC last May, and currently supplies data to Hyundai Motor Group, LG Electronics, and humanoid robotics firms including Holiday Robotics, Airobot, and RealWorld.
The company sits at the intersection of two structural bottlenecks: the data scarcity problem in embodied AI and the compute-intensive nature of high-quality simulation-to-real transfer. While the financing is modest by foundation-model standards, it signals growing enterprise appetite for domain-specific data infrastructure that bridges simulated training and real-world robotic deployment. N.Dlight's TRINIX engine addresses the incentive asymmetry where general-purpose 3D assets are insufficient for robot learning, since physical properties like friction and collision geometry are often missing from standard digital twins. The continued backing of Naver's D2SF accelerator — following a pre-Series A in 2021 and a Series A in 2022 — reinforces the thesis that dedicated physical AI data tooling is becoming a viable standalone category rather than a bolt-on feature of robotics platforms.
For robotics builders, the near-term implication is that the cost and cycle time of obtaining production-grade training environments may drop significantly as specialized data engines like TRINIX mature. For investors, this Series B signals that physical AI data infrastructure is attracting enterprise-led syndicates alongside sovereign-backed capital (KDB), suggesting the segment is gaining recognition as a necessary enabler for the broader robotic labor transition. The capital will fund a generative CAD engine, physical property databases, and deeper integration with the NVIDIA Isaac Sim and Omniverse ecosystem.
