• NVIDIA GB10 Grace Blackwell Superchip | 1 petaFLOP FP4 — 20-core ARM CPU (10× Cortex-X925 + 10× Cortex-A725) with Blackwell GPU featuring 5th-gen Tensor Cores for server-class AI in a desktop footprint
• 128GB LPDDR5X Unified Memory (8×16GB, 256-bit) — High-bandwidth coherent memory shared between CPU and GPU for running large AI models locally
• 4TB PCIe Gen5 NVMe SSD with Self-Encryption — Ultra-fast Gen5 storage with hardware-level data security for sensitive AI datasets and model checkpoints
• NVIDIA ConnectX-7 SmartNIC — High-performance networking for enterprise integration and dual-unit AI scaling up to 405B parameter models
• Wi-Fi 7 + Bluetooth 5.1 — Latest wireless standards for ultra-fast connectivity and peripheral support
• 4× USB 3.2 Type-C + HDMI 2.1b — Comprehensive high-speed connectivity for displays, peripherals, and data transfer
• NVIDIA DGX Base OS + AI Software Stack — Pre-configured with PyTorch, Jupyter, Ollama, and NVIDIA AI tools ready out of the box
• Ultra-Compact 150×150mm | Under 1.5kg — True desktop footprint for office, lab, and edge AI deployment

The Acer Veriton GN100 AI Mini Workstation VGN100-UD13 is a compact AI supercomputer powered by the NVIDIA GB10 Grace Blackwell Superchip — delivering 1 petaFLOP of FP4 AI performance in a 150×150mm desktop chassis weighing under 1.5kg. The GB10 combines 20 ARM cores with Blackwell GPU featuring 5th-generation Tensor Cores and 4th-generation RT cores, backed by 128GB of LPDDR5X unified memory on a 256-bit bus for seamless large language model inference, fine-tuning, and multi-modal AI workloads. The 4TB PCIe Gen5 self-encrypting NVMe SSD provides ultra-fast, secure local storage for sensitive AI datasets and model checkpoints.

The NVIDIA ConnectX-7 SmartNIC enables two Veriton GN100 units to be linked together for multi-node AI scaling — handling models up to 405 billion parameters locally. NVIDIA DGX Base OS arrives pre-configured with the full AI software stack including PyTorch, Jupyter, Ollama, and NVIDIA developer tools for immediate model development upon first boot. Wi-Fi 7, 4× USB 3.2 Type-C, HDMI 2.1b, and a Kensington lock slot complete a feature set that makes this the most connectivity-rich mini AI workstation in its class.

Ideal For

• Local LLM Inference & Fine-Tuning — Run and fine-tune large language models locally without cloud API dependency
• AI Research & Development — Full NVIDIA CUDA/TensorRT ecosystem for model development, testing, and deployment
• Multi-Node AI Scaling — Link 2 units via ConnectX-7 for 405B parameter model support
• Data Science & Deep Learning — 128GB LPDDR5X unified memory for large dataset processing and model training
• Privacy-First Enterprise AI — Self-encrypting SSD and local inference keep sensitive data on-premises

Variant Options

• 2TB SSD — MPN: VGN100-UD11
• 4TB SSD — MPN: VGN100-UD13 | Acer Part Number: DT.R6LAA.003

Specifications

• MPN — VGN100-UD13
• Acer Part Number — DT.R6LAA.003
• Platform — Acer Veriton GN100 AI Mini Workstation
• SoC — NVIDIA GB10 Grace Blackwell Superchip
• CPU — 20-core ARM (10× Cortex-X925 + 10× Cortex-A725)
• GPU — NVIDIA Blackwell (5th-gen Tensor Cores, 4th-gen RT Cores)
• AI Performance — Up to 1 petaFLOP (FP4)
• Memory — 128GB LPDDR5X, 256-bit bus (8×16GB)
• Storage — 4TB PCIe Gen5 NVMe SSD (self-encrypting)
• Networking — NVIDIA ConnectX-7 SmartNIC + LAN
• Wireless — Wi-Fi 7, Bluetooth 5.1+
• USB — 4× USB 3.2 Type-C
• Display — HDMI 2.1b
• OS — NVIDIA DGX Base OS
• Dimensions — 150 × 150mm
• Weight — Under 1.5kg
• Security — Kensington lock slot, self-encrypting SSD