[June 5, 2025] K-ASIC Showcases Three Korean AI Semiconductor Innovators at AI & Big Data Expo 2025

Santa Clara, CA | June 4–5, 2025 – The Korea AI & System IC Innovation Center (K-ASIC) successfully hosted the Korea Pavilion at the AI & Big Data Expo North America 2025, held at the Santa Clara Convention Center. Co-organized with KOTRA, the Korea Pavilion spotlighted Korea’s emerging AI semiconductor startups and served as a strategic platform for global business matchmaking and technology promotion.

This year’s pavilion featured three K-ASIC-affiliated companies: AiM Future, Dnotitia, and Mobilint, each presenting disruptive technologies in edge AI, generative intelligence, and on-device computing.

AiM Future: Pioneering NPU IP for Generative AI

Founded in 2020 as a spin-off from LG Electronics’ North America Research Lab, AiM Future develops and licenses neural processing unit (NPU) IP optimized for AI acceleration. Its solutions are already deployed in LG robot vacuum cleaners and 4K TVs for object recognition and image enhancement.

At the expo, AiM Future unveiled its next-generation GAIA (Generative AI Architecture) series—designed to succeed its NMP 300, 500, and 700 lines. GAIA offers major improvements in performance, power, and area (PPA) efficiency, while supporting a range of AI frameworks via the company’s SDK platform, NeuroMosaic Studio. The SDK allows customers to rapidly prototype and deploy AI models across various devices, from smart home to industrial systems.

Dnotitia: Redefining Local LLMs and Vector AI Infrastructure

Dnotitia introduced two flagship solutions to the North American market for the first time:

• Mnemos – An on-device LLM system that integrates Dnotitia’s proprietary foundation model (DNA) with its lightweight inference engine (Aether). Mnemos enables real-time text and code generation in offline environments, making it ideal for high-privacy sectors such as defense, finance, and healthcare. Its 14B-parameter model is optimized for constrained GPU environments through quantization and compression. A live demo of a no-code Local Coding Assistant built on Mnemos drew significant interest.

• Seahorse – A cloud-based vector database SaaS platform that transforms multimodal data (text, image, video) into high-dimensional vectors for contextual search and RAGOps (Retrieval-Augmented Generation Operations). Seahorse features a GUI-driven MCP (Model Context Protocol) system that simplifies deployment without installation, enabling enterprise teams to build RAG systems in minutes.

  
  

By launching Seahorse as Korea’s first SaaS-based vector DB platform, Dnotitia is enabling enterprises to integrate meaning-based search and generative AI workflows more efficiently.

Mobilint: Launching MLA100 MXM for High-Performance Edge AI

Mobilint exhibited its latest innovation, MLA100 MXM, an edge AI accelerator module powered by its proprietary ARIES chip. With 80 TOPS of compute at just 25W, the module delivers efficient, scalable performance for industrial applications.

Key specs:

• 8-core NPU architecture for parallel model execution

• Compact design (82x70mm, 110g) ideal for constrained environments

• Supports robotics, autonomous systems, and smart manufacturing use cases

• Offers a compelling alternative to GPUs with significantly lower power consumption and cost

  
  

Mobilint’s solution directly addresses the growing demand for lightweight, on-device AI compute across real-world deployment environments.

Global Reception & Strategic Outlook

Throughout the two-day event, all three companies engaged with potential customers, investors, and strategic partners, showcasing Korea’s rising influence in the global AI semiconductor landscape. By combining deep technical innovation with tailored global outreach, K-ASIC and its partners continue to demonstrate the competitiveness and scalability of Korea’s fabless ecosystem.