2026-07-29
SHANGHAI, July 21, 2026 — From July 17 to 20, the 2026 World Artificial Intelligence Conference (WAIC) was held in Shanghai. At the Zhangjiang Science Hall, ALINX (ALINX Electronics Technology Co., Ltd.), a leading provider of FPGA-based hardware solutions, showcased its latest advancements in Edge AI, high-speed video processing, and automotive vision technology. From foundational compute platforms to field-deployed terminal solutions, ALINX demonstrated the critical role of FPGA hardware in empowering next-generation AI applications.

Bridging Foundational Compute with Edge AI Applications
As Large Language Models (LLMs) and physical AI transition from the cloud to the edge, edge computing architectures must handle real-time data ingestion, local pre-processing, and instant AI inference—all under strict low-latency, low-power, and long-term reliability constraints. In emerging fields like humanoid robotics, industrial vision, and autonomous driving, continuous streams from multiple cameras, LiDARs, and sensors create unprecedented challenges for underlying compute infrastructure.
To solve these edge bottlenecks, ALINX highlighted its advanced FPGA+GPU heterogeneous compute platforms at WAIC 2026:
HEA13 : A high-performance heterogeneous platform designed for next-generation Edge AI, built to work seamlessly with the NVIDIA Jetson Thor platform to support ultra-high-bandwidth video handling and complex AI workloads.
AXVU13F : Powered by the AMD Virtex UltraScale+ FPGA, offering top-tier hardware acceleration for high-speed data processing, advanced machine vision, and scientific research.
Z19-M : A versatile embedded platform optimized for embedded vision, intelligent control, and edge compute nodes across diverse performance requirements.
.jpg)
High-Speed Video Ingestion via NVIDIA Holoscan Ecosystem
At the booth, ALINX demonstrated a high-speed video capture and streaming solution developed in collaboration with AMD FPGAs and the NVIDIA Holoscan ecosystem.
By leveraging FPGA logic to perform real-time video ingestion, data framing, and protocol conversion, the solution offloads front-end data pipelines before streaming them directly to GPUs for AI inference. This end-to-end processing pipeline provides scalable hardware foundation for latency-critical applications, including medical imaging, surgical robotics, and automated optical inspection (AOI).
.jpg)