Skip to main content

OnNeuroAI

Category: AI Chips / Semiconductors

A next-generation neuromorphic semiconductor company specializing in ultra-low-power 'Physical AI' chips that enable hyper-personalized, real-time learning on-device without cloud dependency. OnNeuroAI was founded in 2024. The company is led by Hoi-Jun Yoo. Based in Daejeon, South Korea. Team size: 10-20. Total funding raised: N/A. Latest round: N/A.

Founded
2024
Headquarters
Daejeon, South Korea
Team size
10-20
Total funding
N/A

Value proposition

Enables complex Large Language Models (LLMs) to run on edge devices with milliwatt-scale power consumption, offering true 'Physical AI' that adapts to individual user emotions, speech, and habits locally.

Products and solutions

SoulMate: A 9.8mW hyper-personalized mobile intelligence System-on-Chip (SoC) for on-device AI., C-Transformer (Complementary Transformer): A neuromorphic chip designed to accelerate LLMs with extreme energy efficiency (demonstrated GPT-2.0 at 400-470mW)., Physical AI Engine: A proprietary technology stack for real-time local learning of user-specific behavioral and environmental context., Neuromorphic Acceleration IP: Low-power architecture for integration into AIoT and wearable sensors.

Unique value

Shifts the AI paradigm from 'Cloud AI' to 'Physical AI,' where the device learns and adapts to the user's physical presence and habits in real-time through low-power neuromorphic hardware.

Target customer

Mobile and wearable device manufacturers, AIoT service providers, healthcare/biomedical device companies, and robotics firms.

Industries served

Consumer Electronics (Smartphones & Wearables), AIoT (Artificial Intelligence of Things), Digital Healthcare & Biomedical Systems, Robotics & Autonomous Systems

Technology advantage

Utilizes a 'Complementary Transformer' architecture and mixed-signal processing to reduce power consumption to the milliwatt level, making it feasible to host personalized LLMs on tiny battery-powered devices.

How they differentiate

N/A

Key partnerships

KAIST (School of Electrical Engineering / ViP Lab spin-off), Ministry of Science and ICT (MSIT), South Korea (Technology demonstration partner), ISSCC (International Solid-State Circuits Conference) - frequent technology highlight presenter

Growth metrics

N/A

Market positioning

N/A

Geographic focus

N/A

Patents and IP

Holds multiple registered patents in neuromorphic semiconductor technology and ultra-low-power transformer acceleration (referenced by the Korean Intellectual Property Office).

About Hoi-Jun Yoo

Hoi-Jun Yoo is a world-renowned authority in AI semiconductors and a Chair Professor at the School of Electrical Engineering, KAIST. He has over 20 years of experience in AI semiconductor research, specializing in low-power neuromorphic chips and Processing-In-Memory (PIM) technology. He is a Fellow of the IEEE and previously served as the President of the Semiconductor Engineering Society of Korea. Prior to OnNeuroAI, he founded UX Factory and led numerous breakthroughs in AI chip design presented at the International Solid-State Circuits Conference (ISSCC).

Official website: