Imagine bringing a full datacenter AI right to your desk—portable enough to carry in a briefcase, yet powerful enough to fine-tune and infer 200 billion parameter models.
Built around NVIDIA’s GB10 Grace Blackwell Superchip, NVIDIA DGX™ Spark delivers extraordinary performance in a compact form. Highlights include:
Powered by NVIDIA Grace Blackwell architecture
Supports models up to 200B parameters
Scale up to 405B parameter clusters with 2 units
150×150×50 mm, 1.2 kg, ~170 W
Preloaded with NVIDIA’s DGX OS and AI software stack—including PyTorch, RAPIDS, NGC containers, plus optimization libraries—it’s ready to accelerate prototyping, finetuning, inference, data science, and edge AI workflows.
View Datasheet| Specification | Value |
|---|---|
| Architecture | NVIDIA Grace Blackwell |
| GPU | Blackwell Architecture |
| CPU | 20 core Arm, 10 Cortex-X925 + 10 Cortex-A725 |
| CUDA Cores | Blackwell Generation |
| Tensor Cores | 5th Generation |
| RT Cores | 4th Generation |
| Tensor Performance | 1,000 AI TOPS |
| System Memory | 128GB LPDDR5x, unified system memory |
| Memory Interface | 256-bit |
| Memory Bandwidth | 273 GB/s |
| OS | DGX OS |
| Storage | 4 TB NVME.M2 with self-encryption |
| USB | 4x USB Type C |
| Ethernet | 10 GbE 1x RJ-45 connector |
| NIC | ConnectX-7 Smart NIC |
| Wi-Fi | Wi-Fi 7 |
| Bluetooth | BT 5.3 |
| Audio Output | HDMI® multichannel audio output |
| Power Consumption | TBD |
| Display Connectors | 1x HDMI® 2.1a |
| NVENC | NVDEC | 1x | 1x |
| System Dimensions | 150 mm L × 150 mm W × 50.5 mm H |
| System Weight | 1.2 kg |
| Warranty | 1-year limited warranty |
Source: pny.com
We help determine if DGX Spark fits your goals. We align recommendations with your AI strategy—whether you're exploring on-prem LLMs, building portable GenAI workflows, or accelerating model development on the desktop.
Thinkmate provides proactive, high-touch support designed to ensure continuity and top performance throughout the system lifecycle.
