지역별 뉴스를 확인하세요.

HyperVisual AI Secures Investment from K-Ground Ventures Innovation IP Fund for On-Device AI Vision Intelligence

HyperVisual AI, a startup developing on-device vision intelligence solutions that apply AI technology directly on edge devices without relying on cloud connectivity, announced on June 2nd that it has secured investment from K-Ground Ventures, a leading VC in deep tech commercialization in Korea and the first operator of the Scale-up TIPS program.

[K-Ground Ventures CEO Nam-Hoon Cho (left) and HyperVisual AI CEO Sam-Yun Jung (right) Photo credit: HyperVisual AI]

Founded in 2023, HyperVisual AI is a semiconductor-based AI startup that is developing a General-Purpose Neural Processing Unit (GPNPU)—an AI processor operable on-device—as well as proprietary AI models and frameworks. Their GPNPU architecture, a hybrid structure combining general-purpose GPU computing (GPGPU) and Neural Processing Units (NPU) into a single core, enables efficient low-power AI computation. This technology is considered highly optimized for on-device environments. The company was recently selected for the Ministry of SMEs and Startups' “1000+ Super Gap Startups” program, drawing significant interest from fabless firms, VCs, and investors.

CEO Samyun Jung of HyperVisual AI stated that “We are pleased to have our patent value highly recognized and to have secured investment from the K-Ground Innovation IP Fund, which specializes in deep tech IP investments.
HyperVisual AI is actively conducting joint development projects with client companies in the on-device AI industry, where convergent technologies are the key to competitiveness. With our outstanding R&D capabilities and Korea’s deep tech infrastructure, we aim to grow into a global hybrid vision intelligence total solution provider through collaboration with clients and investors.”

Nam-Hoon Cho, CEO of K-Ground Ventures and a leading expert in deep tech technology commercialization, said, “We have long recognized the GPNPU hybrid architecture technology of HyperVisual AI as an innovative IP and technology that can lead the future on-device market. Based on mentorship and business support, we decided to proceed with the investment.

The target markets for on-device applications—mobile, consumer electronics, mobility, and robotics—are areas where Korean companies are particularly strong. We believe that competitive Korean firms can lead the global market through convergent technology. Bold government policy support is needed to establish a new ecosystem that will lead the global on-device AI market, which is expected to become a next-generation growth engine.”

K-Ground Ventures also announced that, through this investment, it plans to build an on-device semiconductor alliance in collaboration with organizations like the Fabless Industry Association, and actively work to discover, invest in, and foster innovative companies.

According to a Deloitte report, the on-device AI market is expected to grow rapidly at an annual rate of 27.95%, reaching USD 16.7 billion by 2031.

Attachment 1: Introduction to HyperVisual AI Technology
The GPNPU architecture currently under development is designed so that the GPU and NPU share memory, with each handling parallel and matrix operations, respectively. This approach resolves bottlenecks, memory access delays, and complex pipeline issues common in traditional heterogeneous computing architectures (CPU, DSP, NPU), significantly improving power efficiency.

Moreover, GPNPU is designed to support not only Vision Transformers but also multimodal and cross-modal AI architectures on edge devices, showing strong potential for future scalability.

Based on this technological foundation, the company plans to complete GPNPU IP design by 2025 and begin commercializing its semiconductor IP in 2026. The ultimate goal is to secure technological leadership in the on-device AI market with a total vision intelligence solution that integrates their proprietary processor and models.