US Hyperscaler Datacenter Equipment Suppliers

The supply chain behind new US hyperscaler campuses (Microsoft Fairwater, Meta Prometheus, OpenAI Stargate, Amazon Ridgeland, etc.) splits cleanly into four layers:


Power Management & Distribution
Power Management & Distribution	ETN	Eaton Corp	$172.8B	Switchgear, PDUs, UPS, busway, critical power infrastructure	Broad hyperscaler supply — Microsoft, Amazon, Google, Meta
Power Management & Distribution	VRT	Vertiv Holdings	$104.0B	Power distribution units, UPS, DC power systems, busway	Named supplier to major US hyperscalers; AI datacenter ramp key growth driver
Power Management & Distribution	VICR	Vicor Corp	$9.6B	High-density power modules, 48V bus converters for AI racks	GPU rack power delivery — Nvidia-driven demand
Power Management & Distribution	ENS	EnerSys	$6.8B	UPS batteries, VRLA / lithium backup power	Supplies backup power to top US hyperscalers
Thermal / Cooling Equipment
Thermal / Cooling Equipment	VRT	Vertiv Holdings	$104.0B	Precision air cooling (CRAC/CRAH), liquid cooling, CDUs	Air-cooled chiller leader by product count in AI DC specs
Thermal / Cooling Equipment	MOD	Modine Manufacturing	$10.1B	Precision air conditioning, liquid-cooled heat exchangers	Hyperscale and AI datacenter cooling systems
Thermal / Cooling Equipment	AAON	AAON Inc.	$7.3B	Rooftop and precision cooling units for datacenters	DC-specific HVAC product line; growing hyperscaler wins
Thermal / Cooling Equipment	SPXC	SPX Technologies	$10.6B	SPX Cooling Technologies — wet cooling towers (largest modeled footprint)	Dominant cooling tower supplier in AI DC equipment specs
Alternative / On-site Power Generation
Alternative / On-site Power Generation	BE	Bloom Energy	$62.0B	Solid-oxide fuel cells for baseload on-site power	OpenAI Stargate New Mexico (Oracle campus); Microsoft partnership
Alternative / On-site Power Generation	SEI	Solaris Energy Infrastructure	$3.9B	Mobile gas turbines leased to hyperscale campuses	Bridging grid capacity gaps at new campuses
Electrical Contracting / Integration
Electrical Contracting / Integration	EME	EMCOR Group	$35.7B	Electrical & mechanical construction, datacenter buildout	Largest US electrical contractor; major DC campus backlog

Key dynamics by layer:

Power Management — Eaton (ETN) and Vertiv (VRT) dominate. Eaton covers the full switchgear-to-PDU stack; Vertiv is the dominant precision power + cooling vendor appearing in the most AI datacenter equipment specs. Vicor (VICR) is the niche play: its 48V bus converters are embedded inside AI racks at the GPU level, a position that scales with Nvidia GPU density.

Cooling — The shift to liquid cooling is the structural story. Vertiv and Modine (MOD) both supply coolant distribution units (CDUs) for direct-to-chip liquid cooling — the required architecture for GPU clusters drawing 100kW+ per rack. AAON supplies the rooftop air-side economization layer. SPX Cooling Technologies (SPXC subsidiary) is the single largest supplier of cooling towers by product model count in actual AI campus specs, handling the wet-side heat rejection for chillers and CDUs.

On-site Generation — Bloom Energy (BE) is the most interesting: fuel cells are already contracted for the OpenAI Stargate New Mexico campus (Oracle-owned) and a Microsoft partnership. Grid interconnect queues now run 5-7 years; on-site fuel cells let campuses reach 400-600MW without waiting. Solaris Energy (SEI) fills a shorter-term gap by leasing mobile gas turbines to cover construction-phase power needs.

Construction Integration — EMCOR (EME) is the single largest electrical and mechanical contractor for US datacenter buildouts. It converts all the above equipment into operational campuses.

Real-world project energy exposure from facility data: Georgia Power (Microsoft Atlanta, 636 MW), American Electric Power (Meta Prometheus + OpenAI Stargate Abilene, 1,052 MW combined), Memphis Light/Gas/Water (xAI Colossus, 340 MW), Dominion Energy (QTS Richmond campus, 562 MW combined). These are utilities, not investable pure-plays, but they confirm where the load growth is concentrating.

Sources: AI datacenter facility and equipment specifications database; chiller and cooling tower manufacturer specifications database; analyst consensus and market data.