Taoyuan, Taiwan – March 12, 2026 – ADLINK Technology Inc., a global leader in edge AI computing, today announced its participation in NVIDIA GTC 2026 at booth 141 from March 16 to 19. This year, ADLINK will showcase how its Edge AI platforms enable scalable architectures across humanoid robotics, medical image AI analytics, and high-performance embedded graphics.
Powering Humanoid Robotics with Industrial-Grade Edge AI
ADLINK will showcase a humanoid robot “Moby”, developed by Noble Machines and powered by the ADLINK Edge AI Platform DLAP series, built on NVIDIA Jetson modules. Leveraging DLAP’s high-performance computing, low-latency perception, and industrial-grade reliability, the DLAP platform enables humanoid robots to operate efficiently in demanding industries such as manufacturing, logistics, semiconductors, mining, and oil & gas.
AI-Powered Medical Imaging with ADLINK DLAP-701 and NVIDIA Jetson Thor
At GTC, ADLINK will showcase its partner DeepRad.AI’s AI-integrated medical imaging solutions, demonstrating how edge AI can enable scalable and high-performance medical imaging-analytics for preventive care.
The solution runs on ADLINK’s DLAP-701, powered by NVIDIA Jetson Thor, and integrates Phison’s aiDAPTIV+ technology to enhance large-model inference efficiency through hardware-based storage acceleration. Together, these technologies deliver high-density, low-latency processing for complex DICOM image analysis at the edge, supporting early multi-risk LDCT screening and MRI dementia risk prediction, enabling more proactive care workflows. Through integration of ecosystem hardware and software technologies, ADLINK streamlines the deployment of medical AI at scale.
Beyond medical applications, DLAP-701, powered by NVIDIA Jetson T5000 and Jetson T4000 modules, delivers up to 2070 FP4 TOPS with 128GB memory, enabling high-density AI workloads and real-time reasoning at the edge. Its compact, high-performance architecture is also well-suited for humanoid robotics.
Twice the Edge AI Throughput with MXM Modules powered by NVIDIA Blackwell
ADLINK will display its latest MXM GPU module, EGX-MXM-BW500, based on NVIDIA Blackwell architecture, a new generation of GPU technology for ADLINK, delivering high-performance edge AI and power-efficient computing.
The series also includes the EGX-MXM-BW2000, which sets a new benchmark for compact performance, delivering up to 100 W within a standard MXM 3.1 Type A (82 mm x 70 mm) form factor. For high-end workloads, the EGX-MXM-BW5000 features 24 GB of GDDR7 memory, enabling local LLM execution at unprecedented speed, alongside MIG (Multi-Instance GPU) technology for precise resource partitioning. Engineered for flexibility, the entire Blackwell-based MXM series supports configurable power budgets ranging from 45 W to 150 W, enabling designers to balance peak compute performance with strict thermal and system constraints.
The NVIDIA Blackwell architecture also introduces FP4 precision, achieving 2x AI inference throughput while drastically reducing memory bandwidth demands. ADLINK bridges the gap between high-performance computing and embedded efficiency, delivering the ultimate compact, power-efficient solution for next-generation autonomous systems.
To learn more about ADLINK’s solutions based on NVIDIA architecture, visit the ADLINK website or join us at the upcoming NVIDIA GTC 2026 at booth 141 from March 16–19, 2026, in San Jose, California.