Funding Flocks to 'Training Data', Not Robots… Ndotlight Secures 15 Billion Won
Date
Aug 11, 2026
Classification
- Startups
#
- Trends/Industries
Author
Ndotlight, a physical AI data company, announced on the 11th that it has secured a 15 billion won investment led by KDB Industrial Bank. Naver D2SF participated in the third consecutive investment round, following the 2021 Pre-Series A and 2022 Series A rounds. Venture capital is increasingly flowing into the field of simulation training data rather than robot bodies.
Venture capital is shifting from robot hardware to companies that create the data for robots to learn from.
Korea Development Bank Makes Third Consecutive Investment in Lead by Naver D2SF… NH and Stonebridge Newly Join
Ndotlight, a physical AI data company, announced on the 11th that it has secured 15 billion won in investment. This investment round was led by KDB Industrial Bank, with existing investors Naver D2SF, IMM Investment, and Capstone Partners participating as follow-up investors. NH Venture Investment, OurCrowd, and Stonebridge joined as new investors. This marks Naver D2SF's third consecutive investment, following its Pre-Series A in 2021 and Series A in 2022.
A shift in the investor's perspective can be observed. While Naver D2SF focused on 3D content creation technology during its initial investment in 2021, it has recently been expanding the scope of collaboration to center on data technology necessary for robot learning. At the time of the first investment, Ndotlight primarily developed content creation solutions utilizing its proprietary 3D engine, and subsequently moved into Naver 1784 to continue collaborations related to the 3D content creation environment.
Ndotlight provides high-precision 3D assets that can be immediately used in simulation and digital twin environments, based on its self-developed SimReady 3D asset creation solution, TRINIX.
The bottleneck is data, not chips or models… "There are no 3D assets with living physical properties"
This field is attracting attention due to the location of the technical bottleneck. The bottleneck in physical AI learning is the lack of usable 3D data. For robots to learn in a simulation environment, they require data with intact physical properties such as shape, mass, and coefficient of friction.
Trinix is a solution that automatically generates and converts text, images, and existing 3D CAD data into Sim Ready assets. Its distinguishing feature is that it goes beyond simple 3D model rendering to include information necessary for simulation, such as physical attributes and joint connection structures, within the 3D assets. Through this, it transforms manufacturing plants, robots, parts, and work environments into assets that can be utilized for digital twins and robot training. Trinix integrates via connectors with NVIDIA Omniverse's OpenUSD-based 3D workflows and NVIDIA Isaac Sim's robot simulation capabilities, enabling manufacturing companies to train, verify, and optimize robot operations prior to actual deployment.
Park Jin-young, CEO of Ndotlight, stated that the key to the era of physical AI lies in how efficiently high-quality data capable of immediate learning in simulation environments can be secured. The company plans to invest the funds in the advancement of its generative CAD engine and material property database, as well as in expanding integration with Isaac Sim and Omnibus workflows, while also pursuing global market expansion and talent acquisition.
Accumulation of deals on the same layer… Government establishes behavioral data training center
Investments at the same level have been appearing repeatedly recently. Sky Intelligence, a digital twin and industrial synthetic data company, secured Series A funding from DS Investment Partners. DS Investment Partners cited the company's differentiated technological competitiveness in the data sector—a key bottleneck in the era of physical AI—as the reason for the investment. ROAI, a spin-off from Hyundai Motor Company's Manufacturing Solutions Division, raised 13 billion won in Series A funding and is developing an autonomous manufacturing infrastructure that integrates everything from automation process design to on-site application.
Policies also target the same point. The government has identified 'behavioral data,' rather than chips or models, as the bottleneck for physical AI and decided to establish behavioral data training centers in five regions nationwide; additionally, on June 19, the Ministry of Science and ICT launched the 'Physical AI Alliance Phase 2' along with the 'K-Physical AI Full Stack Strategy.' The scope of investment, which had previously been concentrated on AI model development companies, appears to be expanding to technologies that implement AI in actual industrial settings, such as robots, sensors, simulations, and training data. However, Ndotlight's funding round stage, corporate valuation, and revenue size were not disclosed.
The center of gravity of physical AI investment is shifting from the robot bodies themselves to the data infrastructure that trains them. The trend of synthetic data and autonomous manufacturing infrastructure companies, including Ndotlight, securing funding in succession reveals where the market identifies the bottleneck. Given that the government's establishment of behavioral data infrastructure and the inflow of private capital are moving in the same direction, it is highly likely that the concentration of funds in this sector will continue for the time being.
#NdotLight #PhysicalAI #SymReady #TrainingData #NaverD2SF
[Advertisement]
Collaboration Proposals / Business Inquiries: hello@startlounge.kr
IR/Pitch deck manufacturing inquiries: hello@pitchmound.kr
IR/Pitch deck manufacturing inquiries: hello@pitchmound.kr
Everything about startups . Start Lounge - Launch Pad 「Lunch Pad」
ⓒ 2026. Start lounge. All rights reserved.
