H3C Launches MegaCube, a Compact NVIDIA GB10 Superchip Mini PC with 128 GB LPDDR5x Memory Priced Above $5000
The rapid adoption of NVIDIA’s GB10 Superchip continues across the industry, and the latest entrant is H3C with its newly unveiled LinSeer MegaCube, a compact yet highly capable AI-focused Mini PC. While GB10 powered systems have already begun appearing from OEMs such as Dell, Lenovo, HP, ASUS, MSI, Acer, and Gigabyte “with starting prices between $2500 and $3500” the MegaCube positions itself in the ultra premium tier, designed for users who require compact AI performance with support for extremely large language models.
The MegaCube stands out with its 150×150 mm chassis, a distinctive gold finish, and a configuration built around the NVIDIA GB10 Superchip, which integrates advanced compute and Blackwell GPU features into a unified platform. Inside, the system hosts 20 ARM cores divided between 10 Cortex X925 cores and 10 Cortex A725 cores, paired with a Blackwell based GPU featuring 6144 CUDA cores. Memory is equally impressive, offering 128 GB of unified LPDDR5x, which is a crucial factor for handling large scale AI workloads.
Storage configurations include options up to 4 TB of NVMe M.2, while the system maintains a peak TDP of approximately 100 W under full load. The unit ships with NVIDIA DGX OS, benefitting from recent OTA updates that add performance enhancements and stability improvements aligned with NVIDIA’s broader AI software ecosystem.
In terms of AI capabilities, H3C claims that a single MegaCube can run LLMs up to 200B parameters, while stacking four units enables workloads such as Qwen3 coder 480B (INT4), Llama 3.1 405B (FP8), and Qwen3 235B (FP8). H3C demonstrated such a four Cube stack at a recent exhibition in China, highlighting the system’s scalability for enterprise level inference and development.
The H3C LinSeer MegaCube is now listed on JD.com at 36,999 RMB, which converts to over $5000 USD. The premium pricing reflects not only the compact yet high performance capabilities of the GB10 Superchip platform, but also market conditions: AI hardware demand remains elevated, and LPDDR5x pricing continues to rise. Given that GB10 based Mini PCs rely on 128 GB LPDDR5x, further retail price increases across this category are expected.
As AI adoption accelerates across consumer, enterprise, and research sectors, systems like the H3C MegaCube illustrate the industry’s pivot toward powerful, small form factor AI nodes designed to run increasingly large models locally. This positions the MegaCube as a cutting edge solution for users willing to invest at the high end of the emerging AI PC segment.
Do you see compact AI PCs like the MegaCube gaining mainstream traction, or will their high pricing keep them targeted at niche professional users? Share your thoughts.
