A new expandable edge computing system combines server-class processors, multi-GPU scalability and high-speed connectivity to accelerate AI training, inference and real-time industrial analytics outside traditional data centres.
Edge AI deployments are becoming more compute-intensive, and new industrial platforms by ASRock Industrial are emerging to handle the growing demand for real-time AI workloads. A newly introduced expandable edge AIoT system aims to deliver data-centre-level performance for industrial environments, supporting advanced AI training, inference and analytics directly at the edge.
The platform is designed for high-performance AI workloads, including generative AI models, visual computing, predictive maintenance and industrial automation. Built around server-grade processors and a high-bandwidth chipset, the system supports large memory capacities and multiple GPU accelerators to handle demanding compute tasks. For AI developers and industrial operators, the ability to deploy powerful compute closer to data sources is becoming critical. Edge systems capable of hosting multiple GPUs enable parallel inference and faster model execution, reducing latency compared with cloud-based processing.
The new platform supports up to four discrete GPUs, enabling workloads such as large language model (LLM) processing, real-time analytics and machine vision. The architecture also supports large memory configurations, up to 2TB of DDR5 ECC RDIMM, to ensure stable processing for memory-intensive AI applications.
The key features are:
- Server-class processors with an enterprise chipset for edge AI workloads
- Support for up to four discrete GPUs for AI training and inference
- Up to 2TB DDR5 ECC RDIMM memory for data-intensive processing
- PCIe Gen5 expansion slots with high-speed networking, including optional 10GbE
- Industrial design with 1600W PSU, TPM 2.0 security and remote management
Designed for industrial reliability, the system integrates a high-capacity power supply, hardware-level security features and remote manageability. These capabilities make it suitable for distributed deployments where performance, uptime and secure remote operations are essential.The platform also offers extensive expansion and connectivity options. Multiple PCIe Gen5 slots allow additional accelerators and high-speed add-on cards, while networking options including 10GbE ensure rapid data transfer between edge systems and central infrastructure.
In addition to the flagship configuration designed for quad-GPU workloads, a more compact version of the platform is available for applications requiring fewer accelerators and lower power consumption. This configuration targets machine vision, inspection systems and industrial analytics where deployment flexibility is a priority.Together, the platforms provide scalable compute options for edge AI deployments, enabling enterprises to match system performance with workload demands while maintaining reliability in industrial environments.
