3D AI startup Endotlight (CEO Jinyoung Park) announced on the 28th that it was selected as one of the five finalist companies in the 'NVIDIA Inception Startup Grand Challenge 2025' hosted by NVIDIA.
Approximately 80 teams participated in this competition, and 15 companies that passed the preliminary round presented their technologies during a live pitching session held on October 21st. Endotlight was selected as one of the five finalists, receiving high marks for technological innovation, business feasibility, team capabilities, market competitiveness, and presentation quality.
In this pitch, EndoLight presented a technical solution to the problem of securing 3D data, a key challenge in the field of "physical AI and robotics," highlighted by NVIDIA CEO Jensen Huang in his GTC 2025 keynote. Currently, simulation data is primarily used for physical AI training, but existing 3D data is mesh-based and unstructured, making it difficult to learn in a way that resembles real environments.
In this regard, Endotrite introduced a method for generating structured and articulated 3D data through its self-developed "Trinix" platform. Data generated by Trinix can simulate real-world environments, making it ideal for robot behavioral learning. Furthermore, the company explained that its AI-powered 3D CAD generation technology, based on text and images, can reduce production time and costs by up to 80%.
CEO Jinyoung Park said, “This selection as a finalist is significant in that it recognizes Ndotlight’s technological competitiveness from the global company NVIDIA,” and added, “We will strengthen our role as a 3D data supply partner in the physical AI field in the future.”
- See more related articles
You must be logged in to post a comment.