Exhibitor login
Nvidia 16 November 2026

NVIDIA DGX Spark Developer Kit: Supercomputing at Your Desk

NVIDIA DGX Spark Developer Kit: Supercomputing at Your Desk

During SC26, an annual conference on supercomputing taking place this week in Chicago, NVIDIA introduced the DGX Spark HPC Developer Kit. This kit is designed for researchers, academics, and developers and consists of a cluster of four NVIDIA DGX Spark systems that are directly connected to each other. This design allows researchers to run multi-node scientific workloads and AI models locally, without having to wait for time on a shared GPU cluster.

The DGX Spark Developer Kit transforms the way researchers work. The kit can scale from one to four systems, with direct connections via NVIDIA ConnectX-7 networking technology, without the need for a network switch to be purchased or set up. With four nodes, it offers a total of 512GB of memory and 20TB of NVMe storage, enabling researchers to leverage more capacity for distributing models and workloads across the cluster.

With the power this kit offers, researchers can now independently develop, debug, and securely store their results without being part of long wait times for server capacity. Additionally, a playbook for the NVIDIA HPC SDK will soon be available, allowing for GPU-optimized development tasks. This enables teams to build and test scientific applications on the four nodes, as long as they work towards scaling to larger systems.

The kit supports various applications, from scientific simulations to AI models. Successful workloads have already been demonstrated, such as for material simulation and molecular dynamics. Researchers can assist professors with agent-assisted workflows and train AI models that replicate simulations, significantly improving the efficiency of their work. The DGX Spark Developer Kit is now on display at SC26 and available for purchase through various OEM partners like Dell and HP. The NVIDIA Marketplace can also be consulted for this.

Read the full article from Nvidia.