Which companies make rack-scale systems for AI data centers?

This catalog lists 23 companies on Rack-scale systems, 22 of them with listed equity: AMD, ASRock Rack, ASUS, Cisco, Compal, Dell, Eviden/Atos, Foxconn (Hon Hai), Fujitsu, Gigabyte, and 13 more.

How reliable are these rack-scale systems supplier placements?

The 23 placements on this part break down as 3 disclosed, 20 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.

4.01

Who makes rack-scale systems for AI data centers

Rack-scale systems (NVL72-class, MI400 Helios-class)

A rack-scale system is the fully integrated cabinet of compute trays, switch trays, power shelves, cooling manifolds and busbar that ships as one wired unit rather than as discrete servers.

The rack-scale system is the top-level assembly layer: it takes the GPU baseboards (4.03), switch trays (4.04), power shelves (4.12) and liquid-cooling hardware from layer 5 and integrates them into a single cabinet wired for one scale-up fabric domain (layer 3). Nvidia and AMD publish the reference architecture and mechanical/electrical spec; the ODMs and OEMs listed against this part build to that spec, badge the result, and compete mainly on integration quality, lead time and support rather than on the underlying design. AI training and inference clusters push this part into existence at all: a conventional enterprise server rack has no equivalent, because the point of a rack-scale system is to turn dozens of GPUs into one coherent domain, something only large models require. For investors, the interesting split is between the reference-design owners (Nvidia, AMD), who capture the architecture value, and the ODM/OEM tier, which competes on manufacturing scale and qualification speed with each new rack generation. Each new generation resets the qualification race, so watch which ODMs get design wins early and which fall behind on ramp.

AI delta: AI training and inference workloads are the reason rack-scale systems exist at all — no other server class assembles dozens of GPUs into one wired, liquid-cooled domain.

Appears at:Large colo / hyperscale building AI campus GW-class campus· cooling:liquidhybrid — enterprise deployments typically use standalone or multi-GPU servers rather than a full rack-scale system; the fully wired cabinet becomes the norm once GPU counts justify NVLink or Infinity Fabric domains at hyperscale and above.

Companies on this part (23)

Show:

Company	Role	Ownership	Ticker	Price	1d	Confidence	Note
AMD	designer	public	AMD	517 USD	+2.2%	inferred	AMD design arm; manufacturing → Sanmina SANM; 2025
ASRock Rack	oem	public	3515.TW	224 TWD	-0.9%	inferred
ASUS	oem	public	2357.TW	946 TWD	-3.7%	inferred
Cisco	oem	public	CSCO	107 USD	+1.4%	inferred
Compal	odm	public	2324.TW	37.30 TWD	-0.9%	inferred
Dell	oem	public	DELL	456 USD	+1.7%	disclosed	· source
Eviden/Atos	oem	public	ATO.PA	23.28 EUR	-3.0%	inferred
Foxconn (Hon Hai)	odm	public	2317.TW	245 TWD	-1.6%	inferred	Ingrasys
Fujitsu	oem	public	6702.T	4477 JPY	-2.0%	inferred
Gigabyte	oem	public	2376.TW	349 TWD	-2.4%	inferred
H3C · unit of Unisplendour	oem	subsidiary	000938.SZ*			inferred
HPE	oem	public	HPE	56.69 USD	+3.4%	inferred
Inspur	oem	public	000977.SZ	58.50 CNY	+3.3%	inferred
Inventec	odm	public	2356.TW	60.50 TWD	-1.0%	disclosed	· source
Lenovo	oem	public	0992.HK	22.64 HKD	-2.2%	inferred
MiTAC (Tyan)	odm	public	3706.TW	80.50 TWD	-1.1%	inferred
NEC	oem	public	6701.T	5390 JPY	-2.8%	inferred
Nvidia	designer	public	NVDA	211 USD	-0.7%	disclosed	· source
Pegatron	odm	public	4938.TW	104 TWD	-1.0%	inferred
Quanta Computer	odm	public	2382.TW	322 TWD	-1.8%	inferred
Supermicro	oem	public	SMCI	32.94 USD	-1.7%	inferred
Wiwynn · unit of Wistron	odm	public	6669.TW	1840 TWD	-2.6%	inferred
xFusion	oem	private				inferred

Who competes here

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

Seen in real machines

rack-scale systems is named in 18 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.

Nvidia GB300 NVL72 · rack teardown
- Rack architecture and reference design — Nvidia
- Rack integration, ODM and OEM badge — Quanta Computer, Wiwynn, Inventec, Pegatron, Supermicro, Dell, HPE, Lenovo, Cisco, Gigabyte, ASUS
AMD Helios (MI455X) · rack teardown
- Rack architecture and reference design (built on Meta's Open Rack Wide standard, submitted to OCP) — AMD
- Rack integration, ODM and OEM systems — HPE, Celestica, Dell, Lenovo, Supermicro, Sanmina, Wiwynn
Google TPU7x (Ironwood) · rack teardown
- Pod architecture and reference design (custom ASIC, not a merchant GPU rack) — Alphabet (Google)
- Board assembly and rack integration — Celestica
El Capitan · supercomputer
- System design and cabinets (HPE Cray EX, 87 liquid-cooled cabinets, built with AMD and Red Hat) — HPE
Frontier · supercomputer
- System design and cabinets (HPE Cray EX, liquid-cooled, up to 64 compute blades in 8 chassis per cabinet) — HPE
Aurora · supercomputer
- HPE Cray EX liquid-cooled cabinet (166 cabinets; 8 compute chassis and 8 switch chassis each, no fans) — HPE, Intel
JUPITER · supercomputer
- System design, integration and rack platform (BullSequana XH3000, delivered by the ParTec-Eviden consortium) — Eviden/Atos, ParTec
Alps · supercomputer
- System design and cabinets (HPE Cray EX, liquid-cooled; 24 GH200 cabinets, 8 chassis each, no cabinet fans, cooled by direct liquid cooling and a coolant distribution unit) — HPE
LUMI · supercomputer
- System design and cabinets (HPE Cray EX liquid-cooled cabinets; eight compute chassis of up to eight blades each) — HPE
Leonardo · supercomputer
- System design, integration and rack platform (BullSequana XH2000 cabinets, 155 racks of 2 tons each; 138 compute racks) — Eviden/Atos
Isambard-AI · supercomputer
- System design, integration and cabinets (HPE Cray EX4000 for Phase 2, HPE Cray EX2500 for Phase 1, direct liquid cooled) — HPE
Venado · supercomputer
- System design and cabinets (HPE Cray EX, direct liquid cooled; designed and manufactured at HPE's Chippewa Falls, Wisconsin, facility and delivered to Los Alamos in March 2024) — HPE
Perlmutter · supercomputer
- System design and cabinets (HPE Cray EX liquid-cooled cabinets, eight compute chassis each; GPU cabinets and CPU-only cabinets) — HPE
Eos · supercomputer
- System design and integration (NVIDIA DGX SuperPOD built from 576 DGX H100 systems, designed and assembled by NVIDIA for its own use) — Nvidia
Fugaku · supercomputer
- System architecture, rack design and integration (jointly developed by RIKEN and Fujitsu; 432 racks, 384 nodes per full rack) — Fujitsu
MareNostrum 5 · supercomputer
- System design, contract and rack platforms (Bull SAS / Eviden BullSequana XH3000 for the accelerated partition, Lenovo ThinkSystem for the general-purpose partition) — Eviden/Atos, Lenovo
Horizon · supercomputer
- System design and racks (Dell Integrated Rack Scalable Systems with direct-liquid-cooled PowerEdge servers; 128 cabinets, GPU racks of 144 GPUs at about 215 kW) — Dell
Doudna (NERSC-10) · supercomputer
- System design, integration and rack platform (Dell Integrated Rack Scalable Systems with PowerEdge servers) — Dell
- Early-access system Cech (72 Grace CPUs, 144 Blackwell GPUs, Quantum-X800 InfiniBand; Dell IR7000-series direct-liquid-cooled rack with an in-row coolant distribution unit and rear-door heat exchanger) — Dell, Nvidia, Rittal

Nearby parts

Next in layer: EMS, system integration & rack-level test
Layer: Layer 4 — Servers, racks & integration

From the blog

Nine supercomputers, six countries, one foundry 2026-10-05
Part of the day: Rack-scale systems 2026-09-17
Eighteen companies cross four layers of this stack. 984 never leave one. 2026-09-03
Every company inside a GB300 NVL72 rack, by exchange 2026-09-02

Watch

NVIDIA GB200 NVL72 | ASUS AI POD

ASUS — vendor walkthrough of an NVL72-class rack-scale system build

Click to load from YouTube · open on YouTube ↗

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

Latest feed

MSI XpertStation WS300 Review NVIDIA DGX Station GB300 — ServeTheHome, 2026-10-23
The article reviews the MSI XpertStation WS300, an AI solution built upon the NVIDIA DGX Station platform.
ZTE unveils data center module designed for AI workloads — Data Center Dynamics, 2026-10-23
ZTE has unveiled a new data center module engineered for AI workloads, capable of providing rack power between 20kW and 60kW.
Lenovo mounts AI comeback with Nvidia Vera Rubin NVL72 deliveries and US plant expansion — DigiTimes, 2026-10-23
Jensen Huang warns Europe risks falling further behind US, China in AI — DigiTimes, 2026-10-23
Nvidia's CEO warned that Europe faces risks of falling behind the US and China in AI advancement, calling for faster adoption and eased regulations.
IBM infrastructure revenue bounces back after Q2 drop — Data Center Dynamics, 2026-10-22
IBM's CEO highlighted the positive momentum expected from sovereign AI initiatives boosting infrastructure revenue after a dip in the second quarter.
Dell unveils Nvidia Vera CPU offerings, launches its first Groq 3 LPU rack-scale inference solution — Data Center Dynamics, 2026-10-22
Dell has unveiled its offerings featuring Nvidia Vera CPUs and its inaugural Groq 3 LPU rack-scale inference solution, all designed for agentic AI applications.
[AI Infrastructure Insight] How AI infrastructure will be designed for the future — newsroom:sk-hynix, 2026-10-22
The design of future AI infrastructure will require integrating advanced compute, cooling, and power elements from various parts of the supply chain.
[AI Infrastructure Insight] How AI infrastructure will be designed for the future — newsroom:sk-hynix, 2026-10-22
The article discusses future data center designs for AI infrastructure, touching on cooling, power, and overall planning.
[AI Infrastructure Insight] How AI infrastructure will be designed for the future — newsroom:sk-hynix, 2026-10-22
The article discusses anticipated design principles for future AI infrastructure, focusing on elements like advanced compute components and specialized cooling/interconnect features.
General Compute signs on to purchase Positron AI hardware — Data Center Dynamics, 2026-10-22
General Compute has agreed to purchase Positron AI hardware, starting with Atlas systems and planning to adopt Titan later.

Questions about rack-scale systems

What are rack-scale systems?: A rack-scale system is the fully integrated cabinet of compute trays, switch trays, power shelves, cooling manifolds and busbar that ships as one wired unit rather than as discrete servers.
Which companies make rack-scale systems for AI data centers?: This catalog lists 23 companies on Rack-scale systems, 22 of them with listed equity: AMD, ASRock Rack, ASUS, Cisco, Compal, Dell, Eviden/Atos, Foxconn (Hon Hai), Fujitsu, Gigabyte, and 13 more.
How does AI change demand for rack-scale systems?: AI training and inference workloads are the reason rack-scale systems exist at all — no other server class assembles dozens of GPUs into one wired, liquid-cooled domain.
How reliable are these rack-scale systems supplier placements?: The 23 placements on this part break down as 3 disclosed, 20 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