MYiR launches fanless edge PC with optional 20 TOPS AI module
MYiR has announced the MAC-B5760, a fanless edge industrial PC powered by the Rockchip RK3576 SoC. The device is designed for edge computing and video analytics applications, featuring hardware-accelerated video processing and an optional AI accelerator module for LLM and VLM tasks.
Key Takeaways
- Rockchip RK3576 SoC includes a built-in 6 TOPS NPU and 8K at 30fps video decoding capabilities.
- Optional RK1828 M.2 module provides an additional 20 TOPS to support LLM and VLM models with up to 7 billion parameters.
- Standard industrial connectivity features dual Gigabit Ethernet, four USB 3.0 ports, and 4G/5G cellular expansion via mini PCIe.
- Base model with 8GB RAM starts at $299; the AI-accelerated variant is priced at $599.
Why It Matters
This launch signals the maturation of edge-based AI for localized video processing without cloud reliance. By offering a modular 20 TOPS upgrade specifically for larger vision and language models, MYiR is targeting the shift from basic object detection to sophisticated semantic video analysis in industrial settings. As cloud egress costs and latency remain bottlenecks for real-time security and monitoring, hardware vendors are increasingly integrating dedicated AI coprocessors into fanless form factors. For the ecosystem, this establishes a competitive baseline for high-performance, low-power inference at the edge. Strategic observers should monitor the adoption rate of these modular 7B-parameter models in private industrial deployments compared to cloud-based solutions.
Additional Context
The launch of the MAC-B5760 follows a broader industry trend toward "Physical AI," where real-time inference is prioritized for robotics and industrial vision. Per Rockchip’s 2025 technology updates, the RK1828 co-processor was specifically designed to overcome memory bandwidth limitations that typically hinder large model deployment on edge devices. By using 3D stacking packaging technology, Rockchip claims these modules can provide cloud-level response speeds while maintaining the low power profiles required for fanless hardware. Recent market analysis from IoT Analytics in May 2026 suggests that 2026 represents an inflection point where OEMs are moving from pilot programs to full portfolio refreshes of edge AI-enabled devices. Competitive pressure in this segment has intensified throughout mid-2026. In May 2026, Seeed Studio released its reComputer RK series, also based on the RK3576 and RK3588, featuring a one-command deployment workflow for NPU-accelerated models. Unlike the fanless MYiR design, Seeed's comparable units often utilize active cooling to manage thermal loads for high-duty-cycle AI tasks. Additionally, research from Grand View Research in early 2026 projected the global edge AI market would reach $118 billion by 2033, driven by a 33% compound annual growth rate in AI-powered video analytics. Industrial buyers are increasingly weighing the fixed costs of high-performance local hardware, like the $599 MAC-B5760 AI variant, against the recurring operational expenses and potential privacy risks of cloud-based inference services.
Read full article at cnx-software.com
Get this in your inbox → Subscribe
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source