NVIDIA announced on 2025‑10‑14 that it will begin shipping its new DGX Spark supercomputer, a compact AI platform that delivers up to 1 petaflop of performance and 128 GB of unified memory in a desktop‑sized form factor. The system is designed to enable developers to run inference on models with up to 200 billion parameters and fine‑tune models up to 70 billion parameters locally, eliminating the need to move workloads to the cloud.
The DGX Spark is powered by NVIDIA’s GB10 Grace Blackwell Superchip, coupled with ConnectX‑7 200 Gb/s networking and NVLink‑C2C technology that provides five times the bandwidth of PCIe 5.0. With 128 GB of CPU‑GPU coherent memory, the platform supports a full NVIDIA AI software stack, including CUDA libraries, NVIDIA NIM micro‑services, and pre‑installed models such as FLUX.1 and Qwen3, allowing rapid prototyping of agentic and physical AI applications.
NVIDIA delivered one of the first units to Elon Musk at SpaceX’s Starbase on 2025‑10‑14, and the company has announced that the DGX Spark will be available worldwide starting 2025‑10‑15. Partners such as Anaconda, Cadence, ComfyUI, Docker, Google, Hugging Face, JetBrains, LM Studio, Meta, Microsoft, Ollama, and Roboflow are already testing and optimizing their tools for the new platform, signaling broad industry interest in the compact AI supercomputer.
The content on BeyondSPX is for informational purposes only and should not be construed as financial or investment advice. We are not financial advisors. Consult with a qualified professional before making any investment decisions. Any actions you take based on information from this site are solely at your own risk.