The Definitive Guide toAI Data Centers
Ask the GuideAccount
ComparedInfiniBand vs Ethernet (RoCE / scheduled fabrics)

InfiniBand vs Ethernet (RoCE / scheduled fabrics)

The back-end fabric question is no longer 'can Ethernet do it' — xAI's Colossus runs ~200k GPUs on scheduled Ethernet at ~95% effective throughput. It is now a systems decision about ecosystem, operations, and how much you pay for determinism.

AxisInfiniBandEthernet (RoCE / scheduled fabrics)
Effective throughputthe historical benchmark for collectives; credit-based, lossless by constructionscheduled/telemetry fabrics reach ~95% effective (Spectrum-X at Colossus); plain RoCE needs careful tuning
Ecosystemsingle-vendor (NVIDIA/Mellanox lineage) — tight, closed, provenmulti-vendor: Broadcom Tomahawk/Jericho, UEC consortium, every switch OS
Cost per portpremium; captive optics/cablesmerchant-silicon economics; 102.4 Tbps ASICs, 512×200G radix
Radix / scaleproven at large scale; managed subnethighest-radix ASICs are Ethernet-first; flatter fabrics, fewer tiers
Operationsseparate skill set + tooling from the front-end networkone operational model for the whole estate; your NetOps already speaks it
Lock-in exposurehigh — fabric, NICs, optics, and GPUs from one vendorlower — but scheduled-fabric features can re-introduce vendor coupling
Where it winsmaximum-determinism training pods inside one vendor stackcost-at-scale, multi-vendor strategy, inference fleets, anyone already Ethernet-native

How the decision falls

Training pods bought as a single NVIDIA system still default to InfiniBand or NVLink-adjacent fabrics; everything else is converging on Ethernet — and the UEC roadmap plus merchant-silicon economics keep pushing the crossover point down. Decide on the silicon layer first (SerDes, buffers, NIC offloads); the protocol fight is downstream of it.

Full derivations, worked examples, and the numbers behind this matrix: Network silicon: the gating layer (Ch 8.3) · Scale-out back-end fabric design (Ch 8.4) · AI traffic characterization (Ch 8.1)