ASRock Industrial NUC(S) Ultra 300 BOX Series Is Powered by Intel Core Ultra 5 325 or Ultra 7 358H “Panther Lake” SoC

Why It Matters

The series brings high‑performance AI compute to a sub‑100 mm footprint, enabling edge devices to run sophisticated inference locally and reduce latency and bandwidth costs.

Key Takeaways

• Panther Lake SoC delivers up to 180 TOPS AI performance
• Supports DDR5 up to 7200 MHz and PCIe Gen5 SSDs
• Two models: NUC BOX (49 mm) and NUCS BOX (38 mm)
• Up to 128 GB memory, 2.5 GbE, Wi‑Fi 7 connectivity
• No Linux support; Windows LTSC versions targeted

Pulse Analysis

The industrial edge market has been hungry for compact platforms that can handle demanding AI workloads without relying on cloud resources. ASRock’s Ultra 300 BOX series answers that need by integrating Intel’s Panther Lake processors, which combine high‑core‑count CPUs, Intel Arc graphics and a dedicated AI accelerator into a single silicon package. With up to 180 TOPS of AI throughput, these mini PCs can run deep‑learning inference for video analytics, predictive maintenance, and robotics directly at the source, cutting latency and data‑transfer costs.

Beyond raw performance, the Ultra 300 BOX series offers future‑proof connectivity and storage. DDR5‑7200 MHz memory and PCIe Gen5 NVMe slots ensure bandwidth‑intensive models can ingest and process high‑resolution sensor streams, while 2.5 GbE Ethernet and Wi‑Fi 7 provide fast, reliable networking for edge deployments. The inclusion of USB4/Thunderbolt 4 and multiple HDMI 2.1/DisplayPort 2.1 outputs supports multi‑display and peripheral configurations common in industrial control rooms and digital signage. Compared with the earlier Ultra 200 series, the new models add higher memory speeds, a more powerful Arc B390 GPU, and a slimmer chassis option, expanding placement possibilities in tight enclosures.

From a market perspective, ASRock’s pricing‑on‑request strategy signals a focus on OEMs and system integrators who require volume discounts and customized support. The lack of official Linux drivers may limit adoption among open‑source‑centric developers, but Windows LTSC compatibility aligns with many enterprise edge solutions that prioritize long‑term stability. As AI inference moves further to the edge, platforms like the Ultra 300 BOX series will likely become building blocks for smart factories, autonomous vehicles, and remote monitoring stations, driving demand for compact, high‑performance compute modules.