The super powered HyperCube AI 4125GS enables up to 10 PCI-E Gen 5.0 x16 devices with extreme scalability and performance. From this cutting-edge design, ensuring that the maximum PCI-E bandwidth is available utilized for direct inter GPU card communications with minimum latency which makes it a perfect solution for GPU accelerated applications and libraries like those used for deep learning.
■ 4U 10x PCI-E 5.0 x16 device GPU server
■ Dual AMD EPYCTM 9005/9004 series processor, TDP up to 400W
■ Up to 6TB 3DS ECC DDR5 6000MT/s RDIMM, 24x DIMM sockets
■ Supports up to 10 NVIDIA Double-Width GPUs
■ 24x Hot-swap 2.5″ drive bay (8x U.2 NVMe drive bay included)
■ Perfect for highly parallel applications such as deep learning and data analytics
The ASUS Ascent GX10, available through ClearCube, is a compact AI supercomputer built around the NVIDIA GB10 Grace Blackwell Superchip with 128GB of unified memory — enough to run large local language models that would normally require a full server rack, in a desktop-sized footprint.
Yes. The GX10 processes all AI workloads on-device, so prompts, searches, and generated content never leave the machine. This makes it well suited for users who need private or air-gapped AI processing, such as testing sensitive LLMs or keeping internal search queries off the cloud.
The GX10 is built on NVIDIA’s Blackwell GPU architecture, the same graphics generation used in gaming and creative workstations, so it can also handle GPU-accelerated rendering, graphics asset creation, and creative workloads alongside AI tasks.
Compact AI desktops like the GX10 are priced well below multi-GPU AI servers (such as ClearCube’s HyperCube AI 4125GS) while still delivering 1–2 petaFLOPs of AI performance — making them a budget-friendly entry point for local LLM development, fine-tuning, and testing.
Yes. The GX10 includes NVIDIA ConnectX-7 networking, which allows two units to be connected to handle larger models that exceed a single unit’s memory capacity.
The NUC 14 Pro AI is a lighter, NPU-based mini PC suited for everyday Copilot-style AI tasks and office use, while the GX10 is a dedicated AI supercomputer with a full NVIDIA Blackwell GPU and unified memory designed specifically for running and developing local LLMs.
Compact yet powerful, the ASUS Mini PC is designed to seamlessly run CoPilot AI, delivering cutting-edge performance and intelligence right at your fingertips—perfect for individuals looking to harness the power of AI on a personal level.
The fastest, most intelligent Windows PCs ever.
Dedicated AI chip
All-local AI processing with 45+ TOPS performance, compatible with a wide range of ISVs.
ClearCube, in coordination with Theatro, a Motorola Solutions Company, provided mini PCs to QuikTrip to run local installations of their new AI-enabled customer support system. The software QuikTrip chose was Theatro’s mobile communication platform for front-line workers. Theatro is powered by GenAI and delivers a hands-free, voice-activated interface to communicate with team members, stay up to date, and access information.
The ClearCube mini PCs that will be running the QuikTrip Theatro system are being deployed to over 1,000 QuikTrip locations. ClearCube computers were selected because of their outstanding and reliable performance running the Theatro system, as well as the design/form factor, configuration, custom accessories, and flexible distribution. ClearCube’s high quality, warranty, and 100% U.S. support team will ensure long-term performance, satisfaction, and success.
Holds up to 3 mini PCs
Mounts with 4 bolts to the front two posts of a standard rack
Front panel provides access to controls and front I/O ports while preventing warm air from escaping out the front of the rack
