Which companies make switch systems for AI data centers?

This catalog lists 13 companies on Switch systems & white-box ODMs, 11 of them with listed equity: Accton (Edgecore), Alpha Networks, Arista, Celestica, Cisco, Dell, Delta Electronics, HPE, Nvidia, Quanta Computer, and 3 more.

How reliable are these switch systems & white-box ODMs supplier placements?

The 13 placements on this part break down as 2 disclosed, 11 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.

3.03

Who makes switch systems for AI data centers

The physical switches — built by branded vendors and white-box ODMs — that house scale-out switch ASICs and connect racks into a cluster fabric.

Switch systems take the ASICs from 3.02 and package them into deployable network hardware, connecting to the transceivers and cabling in 3.04–3.11 on one side and to server NICs on the other. The market splits between branded systems from established networking vendors and white-box designs built by ODMs to hyperscaler specifications, a split that mirrors how hyperscalers buy compute servers. AI buildouts favor white-box and merchant-silicon designs where hyperscalers want to control their own network software stack, while branded systems remain common in enterprise and smaller AI deployments that value a single vendor's support. An investor should watch how this branded-versus-white-box mix shifts with buyer type, since it changes who captures margin on a switch system versus who merely captures the ASIC or optics content inside it.

AI delta: Hyperscale AI clusters increasingly specify white-box switch systems built to their own designs around merchant silicon, shifting share away from fully branded switch platforms.

Appears at:Enterprise / small colo Large colo / hyperscale building AI campus GW-class campus· cooling:any — Present at every scale; the mix shifts from branded systems at enterprise scale toward white-box, merchant-silicon designs as deployments approach hyperscale and campus scale.

Companies on this part (13)

Show:

Company	Role	Ownership	Ticker	Price	1d	Confidence	Note
Accton (Edgecore)	odm	public	2345.TW	1775 TWD	-2.2%	inferred
Alpha Networks	odm	public	3380.TW	32.40 TWD	-0.5%	inferred
Arista	manufacturer	public	ANET	172 USD	+1.3%	inferred
Celestica	odm	public	CLS	260 USD	-2.7%	disclosed	· source
Cisco	manufacturer	public	CSCO	107 USD	+1.4%	inferred
Dell	manufacturer	public	DELL	456 USD	+1.7%	inferred
Delta Electronics	odm	public	2308.TW	1560 TWD	-3.4%	inferred
HPE	manufacturer	public	HPE	56.69 USD	+3.4%	inferred
Nvidia	manufacturer	public	NVDA	211 USD	-0.7%	inferred
Quanta Computer	odm	public	2382.TW	322 TWD	-1.8%	inferred
Wistron NeWeb	odm	public	6285.TW	230 TWD	-6.5%	inferred
Huawei	manufacturer	private				inferred
Ufispace	odm	private				disclosed	· source

Who competes here

The same table read the other way: every company above competes for switch systems & white-box ODMs sockets. Confidence tags matter — an inferred placement is a lead, not a confirmed rivalry.

Seen in real machines

switch systems & white-box ODMs 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.

El Capitan · supercomputer
- Slingshot-11 fabric (HPE Rosetta switches, dragonfly topology, 64 ports at 200 Gb/s) — HPE
Frontier · supercomputer
- Slingshot switches (64-port Rosetta ASIC on TSMC 16nm; 74 groups of 32 switches in a dragonfly) — HPE, TSMC
Aurora · supercomputer
- Slingshot Rosetta switch ASIC (64 ports, 685 mm2, TSMC 16 nm FinFET, about 160 W typical) — HPE, TSMC
JUPITER · supercomputer
- InfiniBand fabric (Quantum NDR switches, Dragonfly+ with 27 groups, about 50,000 links and 867 switches) — Nvidia
Alps · supercomputer
- Slingshot-11 fabric (64-port Rosetta switch ASIC on TSMC 16 nm; 200 Gb/s links; also carries Capstor, Iopsstor and Ritom storage traffic) — HPE, TSMC
LUMI · supercomputer
- Slingshot switches and dragonfly fabric (LUMI-G: 24 groups of 32 switches; LUMI-C: 8 groups of 16 switches; copper inside groups, optical between) — HPE, TSMC
Leonardo · supercomputer
- InfiniBand fabric (QM8700 HDR switches, 823 in total, Dragonfly+ across 23 cells with an all-to-all link between every cell pair) — Nvidia
Isambard-AI · supercomputer
- Slingshot 11 switch blades and fabric (high-performance networking switch blades inside the HPE Cray EX cabinets) — HPE
Venado · supercomputer
- Slingshot 11 fabric (64-port Rosetta switch ASIC on TSMC 16 nm; 200 Gb/s per port per trade press) — HPE, TSMC
Perlmutter · supercomputer
- Slingshot switches and dragonfly fabric (switch blades at the rear of each cabinet; 16-switch groups; copper inside groups, optical between) — HPE, TSMC
- External network gateway (two Arista 7804 routers bridging the Slingshot fabric to NERSC's Ethernet networks and archive) — Arista
Eagle (Microsoft Azure) · supercomputer
- InfiniBand fabric (NVIDIA Quantum-2 NDR switching in a non-blocking fat-tree; 3.2 Tb/s per VM) — Nvidia
Eos · supercomputer
- InfiniBand fabric (Quantum-2 NDR switches, 448 in a 3-tier, rail-optimized, non-blocking fat tree per NVIDIA's statement) — Nvidia
MareNostrum 5 · supercomputer
- InfiniBand NDR200 fabric (324 QM9790 switches in a three-layer fat tree; 3 GPP islands, 7 ACC islands, 1 storage island) — Nvidia
Horizon · supercomputer
- InfiniBand fabric (NVIDIA XDR switches in two non-blocking fat trees joined by a third level; Quantum-X800) — Nvidia
Doudna (NERSC-10) · supercomputer
- InfiniBand network (NVIDIA Quantum-X800, with quality-of-service for data streamed in over ESnet) — Nvidia

Nearby parts

Previous in layer: Scale-out switch ASICs
Next in layer: Optical transceivers
Layer: Layer 3 — Networking & interconnect

From the blog

Part of the day: Switch systems & white-box ODMs 2026-09-06

Watch

Introduction to Broadcom's Switch Portfolio

Tech Field Day — overview of the merchant switch silicon that white-box ODM systems are built around

Click to load from YouTube · open on YouTube ↗

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

Latest feed

Inside the Secretive AWS Networking Lab — ServeTheHome, 2026-10-14
The report details a tour of a private AWS networking laboratory showcasing advanced innovations in management switches, high-capacity 102.4T switches, and optical technologies.
Cisco expands AI networking with new chips, SONiC, cross-site deployment — DigiTimes, 2026-10-14
Cisco is expanding its AI networking capabilities by deploying new switches powered by high-performance silicon and the SONiC operating system across various data center sites.
Google data center strategic sourcing exec heads to OpenAI — Data Center Dynamics, 2026-10-13
Google's strategic sourcing executive is joining OpenAI to focus on high-speed interconnects and rack power infrastructure.
Microchip Technology updates its networking and storage products to better support AI and HPC workloads — Data Center Dynamics, 2026-10-13
Microchip Technology has updated its networking and storage product portfolio to support the specific infrastructure requirements of AI and High-Performance Computing (HPC) workloads.
New USFP Pluggable Optics Form Factor for Higher Speeds from Cisco and NVIDIA — ServeTheHome, 2026-10-13
Cisco and NVIDIA have announced a new USFP pluggable optics form factor designed to support higher data speeds.
NVIDIA and Cisco Introduce New Pluggable Optics Form Factor to Scale AI Factories — HPCwire, 2026-10-12
NVIDIA and Cisco have launched a new pluggable optics form factor designed to enhance scalability for AI data centers.
Micas Networks Unveils 3 102.4T Open Ethernet Platforms for AI Factories — HPCwire, 2026-10-12
Micas Networks has launched three 102.4T Open Ethernet platforms designed for high-scale AI factory interconnects.
Sigurd, Winstek align optical testing push as AI demand shifts toward connectivity — DigiTimes, 2026-10-12
Sigurd and Winstek are collaborating to enhance optical testing capabilities to meet the rising demand for high-speed AI connectivity.
CyberTAN enters 1.6T AI switch race with Marvell-backed platform — DigiTimes, 2026-10-12
CyberTAN has launched a new platform designed for 1.6T AI switching, supported by Marvell Technology.
Growing pains: how distributed AI training changes the network between datacenters — The Register, 2026-10-09
The article discusses how distributed AI training requirements are necessitating more robust and low-latency network architectures between data centers.

Questions about switch systems & white-box ODMs

What are switch systems & white-box ODMs?: The physical switches — built by branded vendors and white-box ODMs — that house scale-out switch ASICs and connect racks into a cluster fabric.
Which companies make switch systems for AI data centers?: This catalog lists 13 companies on Switch systems & white-box ODMs, 11 of them with listed equity: Accton (Edgecore), Alpha Networks, Arista, Celestica, Cisco, Dell, Delta Electronics, HPE, Nvidia, Quanta Computer, and 3 more.
How does AI change demand for switch systems & white-box ODMs?: Hyperscale AI clusters increasingly specify white-box switch systems built to their own designs around merchant silicon, shifting share away from fully branded switch platforms.
How reliable are these switch systems & white-box ODMs supplier placements?: The 13 placements on this part break down as 2 disclosed, 11 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