DGX Spark
What it is
DGX Spark is NVIDIA's desktop-class AI workstation built around the GB10 Grace Blackwell superchip. It brings DGX-class capabilities — large unified memory, high-bandwidth interconnect, and full software stack compatibility — to a single developer's desk.
For teams that need to fine-tune or run large models locally without cloud dependencies, or for solo founders building personal AI products, DGX Spark fills the gap between Jetson edge devices and full DGX-class infrastructure. Information on availability, pricing, and configurations is updating; request a quote and we'll keep you posted.
Specifications
Compute
| Architecture | Grace Blackwell GB10 |
|---|---|
| AI Performance | TBD |
| Form Factor | Desktop tower |
Memory
| Unified Memory | 128 GB |
|---|---|
| Type | LPDDR5X |
| Bandwidth | TBD |
Connectivity
| Networking | TBD |
|---|---|
| Storage | NVMe SSD |
| Display | Multiple outputs |
Software
| OS | Linux (NVIDIA-tuned) |
|---|---|
| Stack | Full CUDA, NeMo, NIM |
| Models | Run open-source LLMs, VLMs locally |
Ask anything
Trained on the datasheet, JetPack release notes, and Jetson AI Lab tutorials. For pricing, quantity, and availability, request a quote.
Try one of these, or ask your own question.
Developer resources
Need pricing, availability, or volume terms?
We talk to you about what you actually need, when you need it, and any backup alternatives that meet your timeline.