Acer Veriton AI Mini Workstation Personal Computer GN100-UD11 Series

The Acer Veriton GN100 AI Mini Workstation puts server-grade AI compute on your desk instead of in a data centre. Built around the NVIDIA GB10 Grace Blackwell Superchip — the same silicon at the heart of NVIDIA’s own DGX Spark platform — this compact workstation delivers up to 1 petaFLOP of FP4 AI performance in a sealed chassis you can hold in one hand. For AI developers, researchers, and businesses that need to run and fine-tune large models locally without cloud API costs, egress fees, or data leaving the building, the GN100 is purpose-built for exactly that job.

Key Features

• NVIDIA GB10 Grace Blackwell Superchip — a unified 20-core Arm CPU (10 Cortex-X925 performance cores + 10 Cortex-A725 efficiency cores) fused with a Blackwell-generation GPU carrying 6,144 CUDA cores and fifth-generation Tensor Cores.
• Up to 1 PFLOPS of FP4 AI performance (with sparsity) — enough throughput to run and fine-tune models with 200B+ parameters locally.
• 128GB unified LPDDR5x memory shared between CPU and GPU at 273GB/s, eliminating the CPU-GPU data transfer bottleneck that limits typical PC-plus-GPU setups.
• 4TB self-encrypting NVMe SSD — fast, secure local storage for models, datasets, and containers, backed by an integrated Kensington lock and tamper-resistant 1.2kg sealed chassis.
• Dual 200Gbps NVIDIA ConnectX-7 ports — direct-attach a second GN100 for combined 405B-parameter inference, or link up to four units through a RoCE 200GbE switch for a small AI cluster.
• NVIDIA DGX OS pre-installed — a Linux-based platform tuned for the full NVIDIA AI stack: CUDA, PyTorch, NIM microservices, and the NeMo Framework, ready to run out of the box.
• Wi-Fi 7 wireless connectivity, four USB 3.2 Type-C ports, HDMI 2.1b output, and a dedicated Gigabit LAN port for standard networking alongside the high-speed ConnectX-7 links.

Bring AI Server Performance In-House

The case for on-premise AI compute has never been stronger, and the GN100 was engineered specifically to answer it. Businesses working with proprietary models, regulated data, or client-confidential information can now run inference and fine-tuning entirely on-device — nothing touches a third-party cloud, nothing is billed per token, and nothing depends on an external provider’s availability. For legal, healthcare, financial services, and R&D teams, that combination of privacy and cost control is often the deciding factor over a subscription-based cloud AI service.

Because the GB10’s unified memory architecture lets the CPU and GPU share the same 128GB pool directly, the GN100 avoids the classic bottleneck of consumer GPU setups, where model size is capped by discrete VRAM. Developers get room to load large open-weight models whole, iterate on fine-tuning jobs, and run NVIDIA NIM microservices locally, all pre-configured through DGX OS so there’s no lengthy driver and framework setup before you’re productive.

And when a single unit isn’t enough, the GN100 scales the way a proper AI workstation should: connect two units directly over ConnectX-7 for larger combined models, or build a four-node cluster over a RoCE switch for serious throughput — all while each unit still occupies less desk space than a small router. For teams that have outgrown a single high-end GPU workstation but aren’t ready to commission a server room, the GN100 is the practical middle step.

Specifications

• Processor: NVIDIA GB10 Grace Blackwell Superchip, 20-core Arm CPU (10x Cortex-X925 + 10x Cortex-A725)
• GPU: Blackwell-generation GPU, 6,144 CUDA cores, 5th-gen Tensor Cores
• AI performance: up to 1 PFLOPS FP4 (with sparsity)
• Memory: 128GB unified LPDDR5x, 273GB/s bandwidth (shared CPU/GPU)
• Storage: 4TB self-encrypting NVMe PCIe Gen5 SSD
• Networking: 2x NVIDIA ConnectX-7 (200Gbps each), Gigabit LAN, Wi-Fi 7
• Ports: 4x USB 3.2 Type-C, HDMI 2.1b
• Operating system: NVIDIA DGX OS (Linux), pre-loaded with CUDA, PyTorch, NIM microservices, NeMo Framework
• Security: self-encrypting SSD, integrated Kensington lock slot, tamper-resistant sealed enclosure
• Form factor: mini workstation, approximately 1.2kg
• Clustering: direct-connect two units for 405B-parameter class inference, or up to four units via a RoCE 200GbE switch