Hardware·
NVIDIA DGX Spark Developer Kit Brings a Four-Node Supercomputer to the Desk
NVIDIA's new DGX Spark Developer Kit packs a four-node Grace Blackwell cluster onto a desk, giving researchers local multi-node HPC and AI capacity without queue delays.

NVIDIA has introduced the DGX Spark HPC Developer Kit at SC26, a compact four-node cluster that fits on a standard desk.
Each node is powered by the GB10 Grace Blackwell Superchip and linked directly to its two neighbours via ConnectX-7 networking, delivering 512 GB of aggregate memory and 20 TB of NVMe storage without the need for an external switch.
The upcoming NVIDIA HPC SDK playbook will bring GPU-accelerated C, C++ and Fortran compilers, libraries and MPI tools to the cluster, matching the software environment of large-scale supercomputers.
Early adopters are already running multi-node simulations such as LAMMPS with Kokkos, NAMD for molecular dynamics, and quantum-chemistry tensor-network workloads up to 70 qubits on a single node.
Why It Matters for GPU/AI Infrastructure
By placing a true multi-node HPC platform at the researcher's desk, the kit removes queue delays, keeps proprietary data on-premises, and lets teams prototype and debug before committing expensive supercomputer time.
- aigpu
- ai gpu
- ai gpu cloud
- aigpu dubai
- dgx spark
- nvidia
- supercomputing
- developer kit
By AiGpu Editorial · Editorial rewrite based on public reporting (NVIDIA Blog)
← All articles