NVIDIA launches DGX spark - The world’s smallest AI supercomputer for developers

NVIDIA has officially begun shipping the DGX Spark™, the world’s smallest AI supercomputer, designed to bring cutting-edge AI performance to desktops and labs worldwide. The launch marks a milestone in making advanced AI development accessible to individual developers and research teams alike.

The compact DGX Spark delivers a full NVIDIA AI stack, combining GPUs, CPUs, networking, CUDA libraries, and AI software in a desktop form factor. Built on the NVIDIA Grace Blackwell architecture, it offers up to 1 petaflop of AI performance and 128GB of unified memory. This allows developers to run inference on AI models with up to 200 billion parameters and fine-tune models up to 70 billion parameters locally capabilities previously reserved for cloud or large-scale data centers.

NVIDIA founder and CEO Jensen Huang personally hand-delivered one of the first DGX Spark systems to Elon Musk at SpaceX in Starbase, Texas. Huang reflected on the origins of NVIDIA’s AI supercomputers: “In 2016, we built DGX-1 to give AI researchers their own supercomputer. I hand-delivered the first system to Elon at a small startup called OpenAI—and from it came ChatGPT, kickstarting the AI revolution. With DGX Spark, we return to that mission—placing an AI computer in the hands of every developer to ignite the next wave of breakthroughs.”

DGX Spark systems support agentic and physical AI development and come preinstalled with the NVIDIA AI software stack. Developers can immediately access models, libraries, and NVIDIA NIM™ microservices to create applications such as image-generation refinement with FLUX.1, vision search and summarization agents using NVIDIA Cosmos™ Reason, or AI chatbots optimized for DGX Spark.

Early adopters include leading technology companies and research organizations like Anaconda, Cadence, Docker, Google, Hugging Face, Meta, Microsoft, NYU Global Frontier Lab, and Roboflow. Kyunghyun Cho, professor at NYU Global Frontier Lab, praised the system, saying, “DGX Spark allows us to access peta-scale computing on our desktop. This new way to conduct AI research enables rapid prototyping of advanced algorithms, even for privacy- and security-sensitive applications like healthcare.”

The DGX Spark combines high-performance NVIDIA GB10 Grace Blackwell Superchips, NVIDIA ConnectX®-7 200 Gb/s networking, and NVIDIA NVLink™-C2C technology, providing five times the bandwidth of PCIe Gen 5. With these features, the desktop supercomputer transforms ordinary office or lab spaces into powerful AI development hubs.

Starting Oct. 15, DGX Spark systems are available for order on NVIDIA.com and through partners including Acer, ASUS, Dell Technologies, GIGABYTE, HP, Lenovo, MSI, and Micro Center in the U.S., as well as worldwide through NVIDIA’s channel partners.

As AI workloads continue to outgrow traditional PCs, workstations, and laptops, NVIDIA’s DGX Spark is poised to democratize access to high-performance computing, empowering developers to drive the next generation of AI innovation from their own desktops.

About NVIDIA: NVIDIA (NASDAQ: NVDA) is a global leader in AI and accelerated computing, delivering breakthrough technologies for developers, enterprises, and research organizations worldwide.