1,024 GPU AI cluster network: DGX H100
Four scalable units: 32 leaf and 16 spine, the second published figure the engine reproduces. Still two tiers.
- GPUs1,024
- Nodes128
- Compute switches48
- Fabric tiers2
- Total racks34
- Total draw1,359 kW
The build
| Platform | NVIDIA DGX H100, 8 × H100 SXM5 per node |
|---|---|
| Compute fabric | infiniband, 8 rails at 400G |
| Leaf | 32 × QM9700 (Quantum-2) |
| Spine | 16 × QM9700 (Quantum-2) |
| Core | not required at this scale |
| Racks | 32 compute + 2 network, 4 nodes per rack, limited by fixed |
| Power | 1,306 kW compute + 53 kW network = 1,359 kW |
Fabrics
Compute
1,024 endpoints across 8 rails, 2 tiers, realised oversubscription 1:1 (non-blocking).
| Tier | Switch | Count | Down/sw | Up/sw | Ports used |
|---|---|---|---|---|---|
| leaf | NVIDIA QM9700 (Quantum-2) | 32 | 32 | 32 | 2,048 / 2,048 |
| spine | NVIDIA QM9700 (Quantum-2) | 16 | 64 | 0 | 1,024 / 1,024 |
Storage
288 endpoints across 1 rail, 2 tiers, realised oversubscription 1.91:1.
| Tier | Switch | Count | Down/sw | Up/sw | Ports used |
|---|---|---|---|---|---|
| leaf | NVIDIA QM9700 (Quantum-2) | 7 | 42 | 22 | 442 / 448 |
| spine | NVIDIA QM9790 (Quantum-2, externally managed) | 3 | 64 | 0 | 154 / 192 |
In-band management
256 endpoints across 1 rail, 2 tiers, realised oversubscription 3.67:1.
| Tier | Switch | Count | Down/sw | Up/sw | Ports used |
|---|---|---|---|---|---|
| leaf | Cisco Nexus 93600CD-GX | 12 | 22 | 6 | 328 / 336 |
| spine | Cisco Nexus 93600CD-GX | 3 | 28 | 0 | 72 / 84 |
Out-of-band management
275 endpoints across 1 rail, 2 tiers, realised oversubscription 1:1 (non-blocking).
| Tier | Switch | Count | Down/sw | Up/sw | Ports used |
|---|---|---|---|---|---|
| leaf | Cisco Nexus 9348GC-FX3 | 6 | 48 | 4 | 275 / 288 |
| spine | Cisco Nexus 93180YC-FX3 | 2 | 48 | 0 | 24 / 96 |
What the engine flagged
UNEVEN_STRIPING — 22 uplinks per leaf do not divide evenly across 3 spines. Link load will be slightly uneven; consider adjusting the switch count or oversubscription.
Bill of materials
Every line carries the calculation that produced its quantity. Prices are deliberately absent: the catalog ships without list prices, because inventing them would be worse than leaving them out.
| Item | Qty | Why this quantity |
|---|---|---|
| Servers | ||
| DGX H100 (8 x H100 SXM5) | 128 | 1,024 GPUs requested / 8 GPUs per node = 128 nodes (1,024 GPUs). |
| Network adapters | ||
NVIDIA ConnectX-7 400G (NDR)MCX75310AAS-NEAT |
1,024 | 128 nodes x 8 compute NICs per node (one per rail). |
| Compute fabric switches | ||
NVIDIA QM9700 (Quantum-2)MQM9700-NS2F |
55 | 55 across Compute fabric (rail-optimized) leaf, Compute fabric (rail-optimized) spine, Storage fabric leaf. |
NVIDIA QM9790 (Quantum-2, externally managed)MQM9790-NS2F |
3 | 3 across Storage fabric spine. |
| Management switches | ||
Cisco Nexus 93180YC-FX3N9K-C93180YC-FX3 |
2 | 2 across Out-of-band management spine. |
Cisco Nexus 9348GC-FX3N9K-C9348GC-FX3 |
6 | 6 across Out-of-band management leaf. |
Cisco Nexus 93600CD-GXN9K-C93600CD-GX |
15 | 15 across In-band management leaf, In-band management spine. |
| Optics — transceivers | ||
25G SFP28 SRSFP-25G-SR-S |
50 | 24 x 25G OOB switch to OOB aggregation at 30 m. 30 m run needs discrete optics over structured fiber. 25G SFP28 SR at OOB switch uplink and 25G SFP28 SR at OOB aggregation, over LC duplex OM4 multi-mode patch. |
800G OSFP 2xSR4 twin-port (NDR)MMA4Z00-NS |
1,214 | 1024 x 400G Leaf to Spine at 30 m. 30 m run needs discrete optics over structured fiber. 800G OSFP 2xSR4 twin-port (NDR) at Compute leaf and 800G OSFP 2xSR4 twin-port (NDR) at Compute spine, over MPO-12 APC OM4 multi-mode trunk. The Compute leaf module is twin-port, so it terminates 2 links and only 0.5 module is needed per link. |
| Cables and assemblies | ||
| 100G QSFP28 passive DAC | 264 | 256 x 100G In-band NIC to In-band switch at 2 m. 2 m run is within the 2 m passive copper limit. Optics are integrated into the assembly, so no separate transceivers are needed. |
800G OSFP to 2x400G OSFP splitter DACMCP7Y00-N001 |
676 | 1024 x 400G Compute NIC to Leaf at 2 m. 2 m run is within the 2 m passive copper limit. Optics are integrated into the assembly, so no separate transceivers are needed; one cage breaks out to 2 links. |
| Cat6a patch lead | 284 | 275 x 1G BMC / device management port to OOB switch at 2 m. 1G over structured copper, 2 m within the 100 m limit for Cat6a patch lead. |
| Structured fibre | ||
| LC duplex OM4 multi-mode patch | 25 | 24 x 25G OOB switch to OOB aggregation at 30 m. 30 m run needs discrete optics over structured fiber. 25G SFP28 SR at OOB switch uplink and 25G SFP28 SR at OOB aggregation, over LC duplex OM4 multi-mode patch. |
| MPO-12 APC OM4 multi-mode trunk | 1,214 | 1024 x 400G Leaf to Spine at 30 m. 30 m run needs discrete optics over structured fiber. 800G OSFP 2xSR4 twin-port (NDR) at Compute leaf and 800G OSFP 2xSR4 twin-port (NDR) at Compute spine, over MPO-12 APC OM4 multi-mode trunk. The Compute leaf module is twin-port, so it terminates 2 links and only 0.5 module is needed per link. |
| Racks | ||
| Rack | 34 | 32 compute racks at 4 nodes each (fixed-limited) plus 2 network racks. |
| Power | ||
| Rack PDU | 68 | 34 racks x 2 PDUs per rack. |