DEEPX
Category: AI Chips / Semiconductors
A fabless AI semiconductor company specializing in high-performance, low-power AI chips and computing solutions for edge devices. DEEPX was founded in 2018. The company is led by Lokwon Kim. Based in Seoul, South Korea. Team size: 100+. Total funding raised: $103M. Latest round: Series C. Key investors include Skylake Equity Partners, BNW Investments, AJU IB Investment, Timefolio Asset Management.
AMW Analysis
DeepX is a Seoul-based fabless AI semiconductor company, founded in 2018, developing high-performance, low-power AI chips for edge devices, with a value proposition centered on enabling complex on-device AI processing for battery-powered and edge hardware at competitive cost. The company has raised $103M in total funding, most recently through a Series C round backed by investors including Skylake Equity Partners, BNW Investments, AJU IB Investment, and Timefolio Asset Management.
Recent news flow shows the company advancing on both product and commercial fronts. In January 2026, DeepX partnered with Advantech to launch the EAI-1961 series, an Edge AI acceleration module built around its DX-M1 NPU, delivering 25 TOPS of performance and targeting industrial environments such as robotics and smart vision. By April 2026, the company had closed a ₩110 billion ($79M) Series C round, secured a 40,000-chip order with Baidu, expanded to 27 orders across 8 countries, and established a three-year production partnership with Hyundai Motor's Robotics Lab. The company is also finalizing a domestic IPO, indicating a shift toward public-market positioning alongside continued commercial expansion in the edge-AI chip segment.
AMW analysis, generated from 2 tracked news signals.
- Founded
- 2018
- Headquarters
- Seoul, South Korea
- Team size
- 100+
- Total funding
- $103M
Value proposition
Providing ultra-low power, high-efficiency on-device AI processing capabilities, enabling complex AI operations on battery-powered and edge devices at a competitive cost.
Products and solutions
DX-V1: An AI NPU for vision applications, capable of processing video from over 16 channels in real-time., DX-M1: A versatile and scalable AI accelerator module for a wide range of edge AI applications., DX-H1: A high-performance AI booster chip designed for AI servers and computing boxes., DX-M2: Next-generation chip in development, designed to bring generative AI capabilities to portable, battery-powered devices., DX-GEN2: A comprehensive software development kit (SDK) for easy and fast development on the DEEPX hardware platform.
Unique value
DEEPX's core innovation is its proprietary Neural Processing Unit (NPU) technology, which is engineered for ultra-low power consumption without compromising high AI performance. The company offers a comprehensive 'All-in-4' solution, providing four distinct classes of AI chips (DX-V1, DX-M1, DX-H1, DX-M2) that cater to the entire spectrum of edge AI applications, from simple sensor-based devices to complex AI servers.
Target customer
Developers and manufacturers of edge AI devices such as smart cameras, robotics, AI-powered servers, and smart factory equipment.
Industries served
Smart Cities, Surveillance and Security, Smart Factories and Industrial IoT, Robotics, Drones, AI Servers and Edge Computing
Technology advantage
The key technological advantage lies in their advanced quantization techniques and hardware design, which dramatically reduce power consumption and manufacturing costs. This allows for the deployment of sophisticated AI models on resource-constrained edge devices. Their business advantage is providing a scalable, efficient, and cost-effective hardware and software platform that accelerates time-to-market for their customers.
How they differentiate
DeepX specializes in ultra-low-power, high-performance Neural Processing Units (NPUs) for on-device AI, particularly for edge computing, robotics, and computer vision applications. Their key differentiator lies in their proprietary quantization technology, which optimizes AI models to run efficiently on resource-constrained devices without significant loss of accuracy. This allows them to deliver superior power efficiency and performance for complex AI workloads at the edge.
Main competitors
Hailo, Kneron, SiMa.ai
Key partnerships
Samsung Foundry (for chip manufacturing), Baidu PaddlePaddle (for AI ecosystem development in drones, robotics, and OCR), DeGirum (for software and AI application integration), BIOSTAR (for developing Edge AI platforms and industrial PCs), Network Optics (for video management systems integration), Dell, HP, AWS, AAEON Technology Inc. (three-year Global Mass Production Cooperation MOU signed at COMPUTEX 2026), Hyundai Motor Group Robotics LAB (strategic collaboration for Physical AI compute platform for robotics)
Notable customers
Hyundai Kia Motors Robotics Lab, POSCO DX, Jahwa Electronics, LGU+, Inventec
Major milestones
Successfully secured $80.5 million in a Series C funding round in May 2024., Announced IPO preparation in April 2026, targeting domestic KOSPI listing with potential US listing, at ~$700M valuation., Named a 'Challenger' in the AI chip market by CB Insights., Won two CES 2026 Innovation Awards (Double Honoree) for embedded tech and hardware design., Signed three-year Global Mass Production Cooperation MOU with AAEON Technology Inc. at COMPUTEX 2026., Partnered with Hyundai Motor Group Robotics LAB to develop next-generation Physical AI compute platform for robotics., Developed comprehensive AI chip lineup (DX-V1, DX-M1, DX-H1, DX-M2) serving various edge AI applications.
Growth metrics
DeepX has received technology validation requests for its DX-M1 prototype from over 300 corporations worldwide. The company's valuation surged more than eightfold from its Series B to its Series C funding round, reaching $529 million.
Market positioning
DeepX is positioned as a leading challenger in the edge AI semiconductor market. The company focuses on providing comprehensive solutions, including four specialized AI chips (DX-L1, DX-L2, DX-M1, DX-H1) and a software development kit (DX-Studio), targeting a wide range of applications from smart cameras and home appliances to industrial IoT and robotics. They are recognized for their technological prowess, surpassing many global competitors in power and computational efficiency.
Geographic focus
Headquartered in South Korea, DeepX has a strong global focus. While building a solid foundation in its domestic market, the company is actively expanding its presence in the Greater China market and has received validation requests from over 300 companies worldwide, indicating a broad international interest. Their collaboration with global partners and participation in international trade shows like CES and MWC underscore their global ambitions.
Patents and IP
DEEPX holds a significant intellectual property portfolio with over 400 patents filed and more than 130 granted globally, primarily in the US, China, and Korea. These patents cover their core NPU architecture, AI processing methods, and quantization technologies.
About Lokwon Kim
Prior to founding DEEPX, Lokwon Kim was a senior research scientist at Apple, where he led application processor design for devices like the iPhone and Apple Watch. He also held senior design roles at Cisco Systems, IBM Thomas J. Watson Research Center, Broadcom, and SK hynix.
Latest news about DEEPX
- South Korea's DeepX secures $30M Series D first close for AI edge chips
- DeepX, the Korean on‑device AI chip maker, is finalising a domestic IPO after wrapping a Series C round that raised ₩110 billion ($79 M) and secured a 40,000‑chip order with Baidu. The firm now has 27
- Advantech and DEEPX have launched the EAI-1961 series, an energy-efficient Edge AI acceleration module featuring the DX-M1 NPU. This M.2 module delivers 25 TOPS of performance with 4GB of LPDDR5 memor
More AI Chips / Semiconductors companies
Official website: https://deepx.ai/
