DEEPX

🚀 Next Stop: electronica in Munich! Following our recent showcase—where we demonstrated the DX-M1 running on embedded platforms with Avnet Silica, Kontron Europe, Renesas Electronics, AMD, and ASRock Inc., alongside live vision-language model (VLM) demos with Neurocle, Ultralytics, and Raspberry Pi—DEEPX is heading to electronica this November! From edge vision AI to on-device VLMs and small language models (SLMs), DX-M1 runs them all efficiently in mass production, driving DEEPX’s expansion as a Physical AI infrastructure leader. At electronica, we will be presenting live demonstrations of our NPU solutions, purpose-built to deliver high-performance, power-efficient intelligence directly to edge devices. 📍 Find us at Hall C5, Booth C5-231! We would love to connect in person to explore collaboration opportunities, share technical insights, and discuss how DEEPX can empower your next-generation AI projects. #DEEPX #Electronica #Germany

😍 Wrapping up the 2026 ITS World Congress in Gangneung, Korea. Under the theme “Beyond Mobility, Connected World,” the world’s largest event dedicated to Intelligent Transport Systems brought together representatives from industry, academia, and government. At the congress, DEEPX shared how ultra-low-power AI semiconductors can transform traffic cameras from devices that simply record incidents into systems that help detect risks and support real-time responses. Operating at approximately 1–5W, depending on the configuration, DEEPX NPUs enable cameras and roadside devices to analyze vehicles, pedestrians, and unexpected events locally, reducing data transmission requirements and processing delays. Applications presented included traffic analysis at smart intersections, pedestrian risk detection in school zones, and accident and wrong-way driving detection on highways. 💥 Highlights from the week: → Commercialization in Korea: Ongoing projects with major transportation and mobility companies across smart tolling, railway safety, public transportation, and parking systems. → New partnerships: Signed MOUs with BYDA, a Korean radar technology company specializing in intelligent transportation systems, to jointly develop AI perception and object detection platforms for autonomous robots and vehicles; and SotaTek, a Vietnamese IT company, to develop and commercialize NPU-based AI solutions, with a focus on the Vietnamese market. → Global engagement: Discussed opportunities to apply ultra-low-power AI semiconductors to Vietnam’s transportation infrastructure with Le Hong Vinh, Vietnam’s Deputy Minister of Construction. “As AI technology advances, intelligent transportation systems are evolving beyond collecting and providing traffic information toward real-time analysis and proactive response. DEEPX’s Korean-developed AI semiconductor technology provides an important foundation for this transformation. ITS Korea will actively support the global expansion of Korean innovation through industry collaboration, field demonstrations, and commercialization.” -Cheong-Hoe Heo, Chairman of ITS Korea- Thank you to ITS Korea, our partners, and everyone who connected with us in Gangneung. #DEEPX #ITSWorldCongress2026 #PhysicalAI

🤖 DEEPX and Doosan Robotics are co-developing an AI welding solution for nuclear power plant manufacturing, to address the shortage of skilled welders on the factory floor. ⚡ The solution combines vision AI and collaborative robots. DEEPX will develop an on-device edge AI device and SDK based on our NPU, build a 3-channel vision data pipeline, develop thermal imaging AI models, and support environmental durability testing and reliability certification for welding environments. Doosan Robotics leads the collaborative robot welding cell, with SAIGE and Changwon National University contributing 3D vision, weld path correction, and digital twin technology. Step-by-step field validation is planned at Doosan Enerbility's nuclear component manufacturing site, with quality verification targeted for 2029 and commercialization from 2030. DEEPX is building the same foundation for robotics beyond this project. DX-Newton is our ROS 2-based robotics AI SDK, designed so developers can move existing ROS 2 robot applications onto DEEPX NPUs with minimal code changes, keeping the same topics, messages, and node structure. 📍 Want to learn more about DX-Newton? Check the comments for the product page and video. #DEEPX #PhysicalAI

🤝 This morning, DEEPX took part in the AI Semiconductor Companies Roundtable hosted by Korea's Ministry of Science and ICT, together with AMD CEO Lisa Su. Lisa Su said AMD's principle is to put an open ecosystem and collaboration at the center, and that AMD will expand AI semiconductor cooperation in Korea through three mega projects: semiconductors, Physical AI, and AI data centers. Vice Minister Ryu Je-myung of the Ministry of Science and ICT noted that as AI shifts from training to inference, heterogeneous AI computing is becoming more important. ⚡ DEEPX CEO Lokwon Kim presented our vision of Physical AI as AI infrastructure: ultra-low-power AI that runs on everyday devices. DX-M1 is already in mass production and has been integrated across AMD and Xilinx-based platforms. DEEPX will continue to support the industries at the core of the AI era. #DEEPX #PhysicalAI

🤖 Physical AI doesn’t have its CUDA yet. In an interview with EE Times | Electronic Engineering Times, our CEO Lokwon Kim explains why edge AI may not be decided by TOPS alone. What matters just as much is how easily developers can take a model from prototype to mass production. 🏅 Our approach: → Open ecosystems: DEEPX NPUs work with tools developers already use, including Ultralytics YOLO, PaddlePaddle, and Raspberry Pi. We are also working toward support for NVIDIA Isaac applications. → From PoC to production: a common hardware foundation that carries developers’ work from the first prototype to volume deployment. → Federated AI: with Amazon Web Services (AWS) IoT Greengrass, models trained in the cloud are deployed, run, and managed as a fleet on DEEPX NPUs. Thank you to Stephen Las Marias, MBA and EE Times for the thoughtful conversation. #DEEPX #PhysicalAI #EdgeAI

[DEEPX X BYDA Co,Ltd: Unlocking the Future of Advanced Traffic Safety Solutions 🚦🤝] We are excited to share highlights from our interview with Park Se-kyung, Executive Director of Product Development & R&D at BYDA! 🎥 BYDA is actively integrating DEEPX’s on-device AI processors into its vehicle speed reduction guidance systems and radar sensor solutions. In the interview, Executive Director Park highlighted the key strengths that DEEPX brings to edge applications: ⚡ Ultra-Fast Real-Time Processing: High-speed computations rapidly convert sensor-level data into real-time traffic and safety insights. 🔍 Edge-Optimized Miniaturization: Compared to conventional GPUs, DEEPX’s compact size and efficient computational load make it possible to build smaller, faster, and highly competitive edge solutions. Starting with vehicle speed reduction guidance, BYDA plans to expand the adoption of DEEPX processors across smart intersections and radar safety systems to accelerate global market expansion. 📢 [Meet Us at the ITS World Congress Gangneung!] You can also see the details of our collaboration with DEEPX at the BYDA booth during the ITS World Congress in Gangneung from October 19th to October 23rd. #DEEPX #BYDA #ITSWorldCongress