Predictika validator · live

Cisco Nexus 9000 AI Cluster Order Check

An AI agent, or a person, prepares an order for an AI training cluster built on Cisco’s Nexus 9000 reference architecture: nodes with 8 GPUs, BlueField-3 NICs and Nexus 9000 east-west switches. Enter the order below, or load a sample, and the validator checks it against the published design: fabric size, optics, Spectrum-X software and licenses, storage links and any totals the order states. This validator contains no prices.

ReadyThe validator is called when you click Validate order.

1. Load a sample or enter the order

Samples (fictional customers):

Fields left blank are not sent. The validator only sees what is in these fields.

Cluster
Fabric
Spectrum-X software and licenses
Totals the order states

Sources and what is not modeled

Rules come from the public Cisco N9000 Enterprise Reference Architecture data sheet for AI clusters (updated April 25, 2026), read October 8, 2026. Reference sizes, optics, Spectrum-X requirements and the storage minimum are as the data sheet gives them; the port count is derived by arithmetic (8 400G ports per node against 128 per 800G leaf switch or 32 per N9332D-GX2B). The data sheet was read through a summarizing fetch in three passes, with the size tables cross-checked between passes but not against a raw copy. Not modeled: the north-south NIC count, the 16-node switch count, and whether NX-OS 10.6(1)F and DCN Advantage are needed with the N9164E-NS4-O. The F train of the NX-OS release is assumed, and a difference from the reference switch counts goes to approval rather than block. Cluster design, release numbers and licenses are supplied by the customer.