UK’s trusted IT infrastructure partner since 2003
sales@servnetuk.com
0800 987 4111
Servnet
ConfiguratorGet in Touch
Technical · AI Infrastructure · Cooling

AI server cooling 2026: air vs liquid vs immersion

Servnet Editorial · AI Infrastructure Practice9 min read

NVIDIA H100 SXM was the last GPU generation comfortably air-cooled at rack density. H200, B100, B200, and GB200 NVL72 push thermal density beyond what air can manage in standard UK colocation facilities. This is the practitioner's guide to the three cooling architectures + when each fits.

Air · Direct-liquid · Immersion — 2026
Air-cooledDirect liquidImmersionMax rack density25 kW80 kW200 kW+PUE1.4-1.71.05-1.151.02-1.10Facility retrofitNoneCDU + manifoldsTanks + dielectricOEM warranty riskNoneVendor-suppliedMostly Asetek / GRC partnersBest for< 700 W GPUH100 / H200 denseB200 / 1 kW silicon

Air cooling — where it still works

Supermicro SYS-821GE-TNHR (8U 8× H100/H200 SXM) — air-cooled at ~10.5 kW per server. Fits standard 42U rack with 6-8 servers and thermal headroom in most UK colos.

NVIDIA HGX H100 SXM5 (700W per GPU) — manageable in air-cooled 8U chassis.

Reality: most UK enterprise AI clusters in 2026 are still air-cooled. Sufficient for inference + small-scale training.

Direct-to-chip (D2C) liquid cooling — where it fits now

B200 SXM (1000W per GPU) — requires direct-to-chip liquid cooling at rack density. Supermicro SYS-421GE-TNHR2-LCC 4U is the typical platform.

NVIDIA DGX B200 — D2C cooling. Requires facility CDU + manifold infrastructure.

Per-rack density: 25-35 kW achievable. Many UK colocation facilities now support D2C — confirm before procurement.

Full immersion cooling — frontier scale

NVIDIA GB200 NVL72 rack-scale superchip — ~120 kW per rack. Requires purpose-designed immersion-cooled facility.

Not all UK colos support this density. Limited to specialist AI-focused facilities (some hyperscaler-adjacent).

Frontier model training only — most UK enterprises don't need this.

AI-cluster cooling stack — choose your layer
4GPU silicon (700-1000 W)H100 → H200 → B200 trend3Cold plate / CDUDirect-liquid loop2Rack manifold + RDHxHot-aisle containment1Facility chilled water15-25 °C supply

UK colo reality 2026

Standard 4-8 kW/rack — most legacy UK colo.

15-25 kW/rack — modern colo supports for AI workloads.

30-60 kW/rack — newer AI-focused colo + some Tier 1 international.

60-120 kW/rack — specialist immersion / liquid-cooled AI facilities. Limited UK options.

What Servnet does

Servnet pre-surveys UK colocation power + cooling before quoting AI infrastructure. We have established relationships with UK colos supporting 25-60+ kW/rack — see our colo procurement guide.

Key takeaways
  • Air cooling sufficient for H100 + most enterprise AI in 2026.
  • D2C liquid cooling required for B200 SXM + DGX B200 at density.
  • Full immersion required for GB200 NVL72 (~120 kW/rack).
  • Pre-survey UK colo power + cooling before AI procurement — many legacy facilities cap at 4-8 kW/rack.
  • Standard UK colos increasingly support 25-35 kW/rack for D2C; specialist facilities support 60-120 kW immersion.
Frequently asked

FAQs — AI server cooling 2026

Cooling

Can we run liquid-cooled servers in our existing colo?

Pre-survey first. Most UK colos built post-2020 support D2C liquid at 25-35 kW/rack. Legacy colos (pre-2015) typically cap at 4-8 kW/rack — requires facility upgrade or move to AI-focused colo.

Related

Got a question this article didn't answer?

One conversation with an engineer who's done this before. No sales script.

Talk to Servnet →