130 parts · 1,202 companies
Prices as of2026-10-25
Search
3.02

Who makes scale-out switch ASICs for AI data centers

The switch ASICs that route traffic between racks and pods, distinct from the scale-up fabric inside a single rack.

Scale-out switch ASICs sit between the rack-level scale-up fabric (3.01) and the switch systems and white-box ODMs (3.03) that build them into deployable hardware. They carry traffic between racks, pods, and clusters — the layer that turns individual GPU-dense racks into a single, larger training or inference domain. AI clusters push both the port count and per-port bandwidth of these ASICs higher as accelerator counts grow, since network bandwidth between racks has to keep pace with the compute inside them or become the bottleneck. An investor should watch supplier concentration at the ASIC level itself, since a small number of designs feed a much larger number of switch-system builders, and watch how Ethernet-based and InfiniBand-based approaches split share as hyperscalers weigh openness against performance.

AI delta: AI clusters need far more inter-rack bandwidth than general-purpose data centers, pushing scale-out switch ASICs toward higher port counts and per-port speeds sooner.

Appears at:Large colo / hyperscale buildingAI campusGW-class campus· cooling:any — Scales with cluster size: enterprise and small deployments may use merchant switch systems without needing the highest-radix ASICs that only very large clusters require.

Companies on this part (5)

CompanyRoleOwnershipTickerPrice1dConfidenceNote
BroadcommanufacturerpublicAVGO322 USD-0.3%disclosedTomahawk/Jericho · source
CiscomanufacturerpublicCSCO107 USD+1.4%inferredSilicon One
MarvellmanufacturerpublicMRVL229 USD+3.7%inferredTeralynx
NvidiamanufacturerpublicNVDA211 USD-0.7%disclosedSpectrum (Ethernet) and Quantum (InfiniBand) · source
Huaweimanufacturerprivateinferred

Who competes here

The same table read the other way: every company above competes for scale-out switch ASICs sockets. Confidence tags matter — an inferred placement is a lead, not a confirmed rivalry.

Seen in real machines

scale-out switch ASICs is named in 15 of the rack teardowns and supercomputer breakdowns on this site. Each line below is a piece of that machine. The companies shown are the ones named on that line, not necessarily the maker of this part.

Nearby parts

From the blog

Watch

Broadcom Tomahawk 6 Scaling AI Networks with the World’s First 102.4 Tbps Ethernet Switch
Tech Field Day — covers Broadcom's flagship scale-out switch ASIC for AI clusters
Click to load from YouTube · open on YouTube ↗
Broadcom Tomahawk 5: World’s Fastest Switch Chip at 51.2Tbps
Tech Field Day — prior-generation Tomahawk switch ASIC launch coverage
Click to load from YouTube · open on YouTube ↗

Independent videos — no affiliation; nothing loads from YouTube until you click.

Latest feed

Questions about scale-out switch ASICs

What are scale-out switch ASICs?
The switch ASICs that route traffic between racks and pods, distinct from the scale-up fabric inside a single rack.
Which companies make scale-out switch ASICs for AI data centers?
This catalog lists 5 companies on Scale-out switch ASICs, 4 of them with listed equity: Broadcom, Cisco, Marvell, Nvidia, Huawei.
How does AI change demand for scale-out switch ASICs?
AI clusters need far more inter-rack bandwidth than general-purpose data centers, pushing scale-out switch ASICs toward higher port counts and per-port speeds sooner.
How reliable are these scale-out switch ASICs supplier placements?
The 5 placements on this part break down as 2 disclosed, 3 inferred. "disclosed" means the company or a filing says so; "reported" comes from trade press or a reference-design list; "inferred" is a taxonomy lead with no citation yet.

Sources: placements without a source URL come from the taxonomy seed and are tagged accordingly. How this data is built.· Markdown · JSON