DGX Spark / GB10 Cluster Cable

DGX Spark and GB10 Cluster Cable: 0.5m QSFP112 400G DAC

To cluster two NVIDIA DGX Spark units, or any GB10 Grace Blackwell workstation, you need one 0.5m QSFP112 400G passive DAC cable plugged into the QSFP112 port on each machine. It is the single cable NVIDIA calls out in its Spark Stacking documentation (Amphenol NJAAKK0006 / Luxshare LMTQF022-SD-R spec), and because every GB10 system shares the same ConnectX-7 NIC, this one cable fits the DGX Spark Founders Edition and every OEM build alike. Petronella Technology Group, Inc. keeps it in stock at $159 with free US shipping, while distributors quote $179 to $229 and are usually backordered.

In stock - ships within about a week | NVIDIA-approved spec | Free US shipping | Questions? Call 919-348-4912

The one cable every GB10 workstation needs

NVIDIA's GB10 Grace Blackwell superchip put a petaFLOP-class AI machine on the desk, and the way you scale past a single 128 GB node is to cable two or three of them together over their built-in QSFP112 ports. That interconnect is one specific, hard-to-source part, and it is the component distributors rarely keep on the shelf. Whether your box says NVIDIA, Dell, ASUS, MSI, Lenovo, HP, Acer, or Gigabyte on the front, the cable is identical.

Cable0.5m QSFP112 400G passive direct-attach copper (DAC)
Approved specAmphenol NJAAKK0006 / Luxshare LMTQF022-SD-R (NVIDIA Spark Stacking docs)
FitsBoth QSFP112 ports on the ConnectX-7 NIC in every GB10 system
Link rate200 Gb/s class per link (cable is 400G-rated for headroom)
For two unitsOne cable
For a three-node ringThree cables (switchless, each box uses both ports)
Price$159, free US shipping (in stock)

Clustering DGX Spark and GB10 units: cables, ports, and when you need a switch

Every GB10 system has exactly two QSFP112 ports, so the topology is the same no matter which vendor built it. Two units link directly with one cable. Three units form a switchless ring using three cables, following NVIDIA's "Connect Three DGX Spark in a Ring Topology" playbook, where each box uses both of its ports for full pairwise connectivity. Three nodes is the switchless ceiling: with only two ports per unit, a fourth node cannot be cabled directly to all the others, so four or more units need a 200G-class QSFP switch.

It helps to be clear about what clustering buys you. Linking nodes pools their memory so you can load a model too large for a single 128 GB machine, and it lets a workload span multiple GPUs. It is a capacity play, not a raw speed boost, because each link runs at roughly 25 GB/s. For memory-bound inference and fine-tuning of larger models that is exactly the win you want; for a job that already fits in one node, a second box adds headroom rather than throughput.

GB10 cluster cable compatibility by system

The same 0.5m QSFP112 400G DAC clusters any of these GB10 Grace Blackwell workstations, in matched pairs or mixed-vendor. Pick your machine for the exact fit notes and to order.

WorkstationInterconnectCluster cable page
NVIDIA DGX Spark (Founders Edition)2 x QSFP112, ConnectX-7You are here
Dell Pro Max with GB102 x QSFP112, ConnectX-7Dell Pro Max GB10 cable →
ASUS Ascent GX102 x QSFP112, ConnectX-7ASUS Ascent GX10 cable →
MSI EdgeXpert2 x QSFP112, ConnectX-7MSI EdgeXpert cable →
Lenovo ThinkStation PGX2 x 200Gb QSFP, ConnectX-7Lenovo ThinkStation PGX cable →
HP ZGX Nano G1n2 x QSFP112, ConnectX-7HP ZGX Nano cable →
Acer Veriton GN1002 x QSFP112, ConnectX-7Acer Veriton GN100 cable →
Gigabyte AI TOP ATOM2 x QSFP112, ConnectX-7Gigabyte AI TOP ATOM cable →

Why buy the cluster cable from Petronella

The DGX Spark and its OEM siblings shipped faster than the accessory channel could keep up, so the one cable that turns two boxes into a cluster is routinely backordered at the big distributors, and quoted at $179 to $229 when it is in stock at all. Petronella Technology Group, Inc. carries the NVIDIA-approved 0.5m QSFP112 400G DAC in stock at $159 with free US shipping, ships within about a week, and can quote multipacks for three-node rings and small fleets. If you are standing up more than a handful of nodes, or want the 200G-class QSFP switch guidance for four or more units, call 919-348-4912 and we will spec it with you.

DGX Spark and GB10 cluster cable FAQ

What cable do I need to cluster two DGX Spark units?
One 0.5m QSFP112 400G passive DAC, one end in each unit's QSFP112 port. NVIDIA's Spark Stacking docs specify the Amphenol NJAAKK0006 / Luxshare LMTQF022-SD-R spec, and the same cable fits every GB10 system because they share the ConnectX-7 NIC. One cable links two units; a switchless three-node ring needs three.
Is the cable the same across every OEM GB10 workstation?
Yes. The NVIDIA DGX Spark Founders Edition and the Dell, ASUS, MSI, Lenovo, HP, Acer, and Gigabyte GB10 builds all use the same ConnectX-7 NIC and dual QSFP112 ports, so the 0.5m QSFP112 400G DAC is the correct part for all of them, including mixed-vendor clusters.
How many units can I cluster without a switch?
Up to three, switchless: two units link directly with one cable, and three units form a switchless ring with three cables, each box using both ports. Four or more units need a 200G-class QSFP switch, since each unit has only two ports.
How much does clustering actually speed things up?
Clustering pools memory so you can run a model too large for one 128 GB node. It is a capacity play rather than a raw speed boost, because each link runs around 25 GB/s. For memory-bound larger-model inference and fine-tuning that is the win; a job that already fits in one node gains headroom, not throughput.
Is it in stock and how fast does it ship?
Yes. Petronella Technology Group, Inc. stocks the 0.5m QSFP112 400G DAC at $159 with free US shipping and ships within about a week, while distributors quote $179 to $229 and are usually backordered. See full details and multipack bundles.