Skip to content

NewsGPU

NVIDIA's DGX Spark HPC Developer Kit Links Four Sparks Into a 512GB Desk-Side Cluster

NVIDIA's DGX Spark HPC Developer Kit, announced at SC26 on 16 November 2026, links four DGX Spark systems in a ring over ConnectX-7 with no network switch, for 512GB of combined memory and 20TB of NVMe storage. NVIDIA says it can spread models of up to 700B parameters across the nodes with NVFP4. No UK kit price yet; Scan sells two-Spark bundles at about £12,845.

By , Editorial team

3 min read

NVIDIA's DGX Spark HPC Developer Kit Links Four Sparks Into a 512GB Desk-Side Cluster

NVIDIA has used the SC26 supercomputing conference in Chicago to announce the DGX Spark HPC Developer Kit, a four-node cluster of its small DGX Spark AI computers aimed at researchers, academics and developers. NVIDIA announced it on its blog on 16 November 2026.

Forge PC hasn't tested DGX Spark or the new kit. This report is based on NVIDIA's announcement and DGX Spark product page, Notebookcheck's coverage and Scan's UK listings, all checked on 17 November 2026.

What NVIDIA announced

  • Four DGX Spark systems in a ring. NVIDIA says each node connects directly to its two neighbours over ConnectX-7 networking, "with no network switch to buy or set up".

  • 512GB of combined memory and 20TB of NVMe storage across the four nodes. Each DGX Spark has up to 128GB of unified memory, so this is four separate pools linked by networking, not one big memory pool.

  • Models of up to 700 billion parameters. NVIDIA says that with its NVFP4 format, researchers can spread models of that size across the four systems.

  • An HPC software playbook "coming soon". NVIDIA says it will bring its HPC SDK (GPU-accelerated C, C++ and Fortran compilers, libraries and MPI tools) to the cluster.

  • Scientific workloads. NVIDIA says LAMMPS (materials simulation) and NAMD (molecular dynamics) have been demonstrated across four systems, and that the wider DGX Spark community has validated GROMACS, OpenFOAM and WRF.

NVIDIA says the kit is on display at SC26 and available to buy, with DGX Spark products and cables sold through NVIDIA Marketplace and partners including Acer, ASUS, Dell, GIGABYTE, HP, Lenovo and MSI. Notebookcheck notes that NVIDIA hasn't given a price for the four-node kit.

Why the ring matters

NVIDIA's own DGX Spark product page says two or three systems can be linked directly with QSFP cables, but that clustering four systems needs an external network switch. The HPC Developer Kit gets round that by wiring the four nodes in a ring, each linked straight to its two neighbours. For a small lab, not having to buy and configure a 200GbE-class switch is a real saving in cost and setup time.

What it costs in the UK

There's no UK price for the four-node kit yet. Scan already sells DGX Spark bundles: on 17 November 2026 it listed a two-system bundle with one QSFP cable at £12,845 and a three-system bundle with three cables at £19,338 (approximate prices, checked 17 November 2026). Wait for an actual UK listing before budgeting for the four-node kit.

Should home users care?

Not directly. This is a research and development tool, built so teams can test multi-node software at a desk before using time on a large shared cluster. For running local AI models at home, a single machine is still the sensible starting point. Our DGX Spark vs Ryzen AI Max+ 395 comparison, best PC for local AI guide and NVIDIA RTX Spark desktops in the UK report cover options that make more sense for one person.

Our verdict

NVIDIA's DGX Spark HPC Developer Kit links four DGX Spark systems in a ring over ConnectX-7, without a network switch, for 512GB of combined memory and 20TB of storage, and NVIDIA says it can spread models of up to 700 billion parameters across them using NVFP4. It's aimed at researchers testing multi-node scientific and AI software, not home users. There's no UK price for the kit yet; Scan sells two- and three-system DGX Spark bundles from about £12,845. Forge PC hasn't tested it.

The bottom line

NVIDIA's DGX Spark HPC Developer Kit links four DGX Spark systems in a ring over ConnectX-7, without a network switch, for 512GB of combined memory and 20TB of storage, and NVIDIA says it can spread models of up to 700 billion parameters across them using NVFP4. It's aimed at researchers testing multi-node scientific and AI software, not home users. There's no UK price for the kit yet; Scan sells two- and three-system DGX Spark bundles from about £12,845. Forge PC hasn't tested it.

Sources

  1. 01NVIDIA Blog, NVIDIA DGX Spark Developer Kit Turns a Desk Into a Supercomputing Lab (16 November 2026)
  2. 02NVIDIA, DGX Spark product page
  3. 03Notebookcheck, Nvidia unveils DGX Spark HPC Dev Kit with 512 GB of memory
  4. 04Scan, NVIDIA DGX Spark search results (UK prices, 17 November 2026)

About the author

Forge PC Staff

Editorial team

The shared byline for articles researched, written and edited by the Forge PC editorial team. How we source, check and correct our work is set out in our editorial policy.

More from Forge PC Staff

Read next

How to Enable AMD Fluid Motion Frames (AFMF 2.1) on a Radeon Graphics Card

How-To

How to Enable AMD Fluid Motion Frames (AFMF 2.1) on a Radeon Graphics Card

To enable AMD Fluid Motion Frames, install Adrenalin 25.3.1 or newer, then select the HYPR-RX profile (RX 7000 and newer) or turn on AMD Fluid Motion Frames under Gaming > Graphics, globally or per game. AFMF 2.1 works on RX 6000 cards and newer in DirectX 11, DirectX 12, Vulkan and OpenGL games, with VSync off; RX 6000 cards need exclusive full screen.

Forge PC Staff2 min

AMD's FSR 4.1.2 Unreal Engine 5 Plugin Brings ML Upscaling to RX 6000 Cards

News

AMD's FSR 4.1.2 Unreal Engine 5 Plugin Brings ML Upscaling to RX 6000 Cards

AMD's FSR Unreal Engine 5 plugin now uses FSR Upscaling 4.1.2, bringing ML-based upscaling to Radeon RX 6000 cards and select Ryzen integrated graphics as well as RX 7000 and RX 9000, in UE 5.3 to 5.8 games that adopt it. It's DirectX 12 only for now and adds Frame Generation 4.0.2 with driver-assisted frame pacing on RX 9000 cards.

Forge PC Staff2 min

Best Free Benchmark Software for PC in 2026: CPU, GPU, SSD and AI Tests

Best Picks

Best Free Benchmark Software for PC in 2026: CPU, GPU, SSD and AI Tests

The best free PC benchmarks in 2026: Cinebench 2026 for CPUs, Unigine Superposition Basic or 3DMark's free Steel Nomad test for graphics cards, CrystalDiskMark for SSDs, Geekbench 7 for cross-platform scores, Geekbench AI for NPUs, Blender Benchmark for creators and OCCT for stability. Run each test twice and download only from official sites.

Forge PC Staff5 min

RTX PRO 6000 vs RTX 5090 for Local AI: Is 96GB of VRAM Worth It?

Versus

RTX PRO 6000 vs RTX 5090 for Local AI: Is 96GB of VRAM Worth It?

The RTX PRO 6000 Blackwell has 96GB of ECC GDDR7 to the RTX 5090's 32GB, so it can run 70B-class models such as Llama 3.3 70B (43GB) and gpt-oss 120B (65GB) on one card, and adds MIG for sharing. The 5090 suits models such as gpt-oss 20B and doubles as a gaming card. UK: 5090 from about £7,600; PRO 6000 is quote-only at Scan.

Forge PC Staff4 min