Skip to main content

Superb AI

Category: Computer Vision

South Korean all-in-one Vision AI / MLOps platform company providing end-to-end data labeling, model training, and video analytics for industrial applications, and winner of the CVPR 2026 Foundational Few-Shot Object Detection Challenge. Superb AI was founded in 2018. The company is led by Hyun Kim (Hyunsoo Kim). Based in Seoul, South Korea (US HQ: San Mateo/Foster City, CA). Team size: 101-500. Total funding raised: $43M. Latest round: Pre-IPO. Key investors include Y Combinator, Korea Development Bank, Hanwha Asset Management, POSCO Investment, Doosan Investment, Hyundai Motor Group, Samsung Next, Kakao Investment, Halla Group, Atinum Investment, KT, KT&G.

Founded
2018
Headquarters
Seoul, South Korea (US HQ: San Mateo/Foster City, CA)
Team size
101-500
Total funding
$43M

Value proposition

An integrated vision AI stack (Superb Platform, ZERO vision foundation model, Superb Video Analytics) that automates the full AI development lifecycle — from data collection/labeling to model training and deployment — enabling enterprises to build and deploy vision AI with minimal data and no coding expertise.

Products and solutions

Superb Platform (end-to-end MLOps from data construction to model training), ZERO (Korea's first industrial vision foundation model, lightweight zero-shot), Superb AI Video Analytics (real-time anomaly detection + natural-language video search across CCTV feeds), synthetic data generation.

Unique value

Korea's first industrial vision foundation model (ZERO) plus an end-to-end MLOps platform that cuts labeling effort by 10x and achieves 98% defect detection accuracy, with 2-second search latency across 10M video frames.

Target customer

Large enterprises in mobility/autonomous driving, security & defense, logistics, and manufacturing; Fortune Global 500 companies.

Industries served

Mobility/automotive, Security & defense, Logistics, Manufacturing, Smart cities, Public institutions, Steel, Energy.

Technology advantage

ZERO vision foundation model (lightweight zero-shot, few-shot object detection) that won CVPR 2026 challenge; multimodal AI + Visual Language Models (VLMs) for video analytics; data-centric AI approach; NVIDIA's only Physical AI partner in Korea; AWS Rising Star Partner of the Year.

How they differentiate

Unlike pure data-labeling competitors (Scale AI, Labelbox), Superb AI owns the full stack — data labeling, its own industrial vision foundation model (ZERO), and multimodal video analytics — enabling 10x labeling reduction and few-shot/zero-shot deployment with minimal data. Strong strategic-investor-customer synergy (investors like Hyundai, Doosan, POSCO become customers).

Main competitors

Scale AI, Labelbox, SuperAnnotate, Encord, V7

Key partnerships

NVIDIA (only Physical AI partner in Korea), AWS (Rising Star Partner of the Year), LG AI Research-led consortium (Ministry of Science and ICT national AI foundation model project / K-AI Companies), Hanwha Systems (AI Challenge winner 2 consecutive years).

Notable customers

Samsung Electronics, LG Electronics, Hyundai Motor, Toyota Motor, Nippon Steel, KT, Kakao, NCSoft, POSCO, Doosan, Hanwha Systems, Samsung SDS

Major milestones

2018 founded, YC W19 batch, 2022 Series B $16M, 2024 Series C $10.2M, 2025 launched ZERO vision foundation model and Superb AI Video Analytics, 2025 named AWS Rising Star Partner of the Year, 2025 pre-IPO funding KRW 14B, 2026 won CVPR 2026 Foundational Few-Shot Object Detection Challenge (first Korean company), 2026 passed first hurdle for Kosdaq listing with planned 2026 IPO.

Growth metrics

100+ global enterprise deployments; 98% defect detection accuracy; 2s search latency across 10M frames; 10x labeling reduction with ZERO; 92.7% abnormal-behavior detection rate; 13.2% drop in product defect rate.

Market positioning

Korea's national-representative AI company (K-AI), NVIDIA's only Physical AI partner in Korea, and AWS Rising Star Partner of the Year. Positioned as a full-stack industrial vision AI platform differentiated from pure data-labeling competitors by owning the foundation model (ZERO) and video analytics layer.

Geographic focus

South Korea (home), United States, Japan (subsidiary with Toyota and Nippon Steel customers); expanding globally

Patents and IP

Proprietary ZERO vision foundation model; CVPR 2026 Foundational Few-Shot Object Detection Challenge winner (first Korean company to top the challenge).

About Hyun Kim (Hyunsoo Kim)

Ex-SK Telecom / SK T-Brain Machine Learning Research Engineer (2016-2018); Robotics Team Lead @ Amazon Robotics; Duke University B.S./Ph.D. in Robotics & AI (dropout). Co-founded Superb AI in 2018; YC W19 alum; Forbes Under 30 Asia honoree.

Latest news about Superb AI

More Computer Vision companies

Official website: