Vertiv 360AI High Density Reference Design #014: 2x 1.2 MW NVIDIA Blackwell DGX SuperPODs, 2N
Vertiv
A 3 MW-class hall feeding two GB200-generation SuperPODs from fully 2N power, the counterpoint to the 4-to-make-3 designs in the series.
On this page
Overview
This 2024 reference design is a 3 MW-class AI hall with two GB200-generation SuperPODs. Total IT load is 2,352 kW (1,176 kW per pod) against a solution capacity of 3 MW. It holds 16 compute racks at 132 kW and 20 support racks at 12 kW, 36 racks in all. Per pod, 898 kW (76%) goes to liquid and 278 kW (24%) to air, with the liquid side carried by 1,350 kW CDUs.
The defining difference is the power architecture. Unlike the 4-to-make-3 designs in the same series, two 2,500 kVA UPS systems run 2N, each able to carry both pods, and static transfer switches (STS) feed two 600 kVA row PDUs per pod. The 5,000 kVA installed compares with 2,352 kW of IT load. Cooling UPS is 240 kVA 2N and perimeter CRAH units are N+1.
Distribution uses busbar instead of tap-off busway: two 1,000 A busbars per pod, 185 A tap-off boxes and two 185 A rack PDUs per rack. That current scale shows the cabling and protection burden of 130 kW-class racks. The record gives no further calculations; vendor values should be taken as indicative. Two 185 A feeds per rack split a 132 kW rack's current across two paths, which is where the rack-level redundancy comes from.
Key points
- Total IT 2,352 kW (1,176 kW per pod); solution capacity 3 MW.
- 16 racks at 132 kW + 20 racks at 12 kW (36 racks).
- Per pod 898 kW liquid (76%) / 278 kW air (24%).
- IT UPS: 2 x 2,500 kVA in 2N, each supplying both pods.
- Cooling UPS 240 kVA 2N; perimeter CRAH N+1.
- Rack PDUs 185 A (2 per rack); busbar 1,000 A (2 per pod); tap-off 185 A (2 per rack).
- Row PDUs 600 kVA, 2 per pod, fed via STS.
Using it in design
- Provide a "2N AI hall" preset to compare with the 4N/3 preset: 2N makes installed UPS 200% of load against 133% for 4N/3; show the capex and efficiency deltas.
- A 132 kW rack class needs 2 x 185 A (415 V, 3-phase) feeds; validate rack PDU and tap-off sizing with P = sqrt(3) x V x I x PF.
- Compare the fault domain of the STS-fed 2N chain with 4N/3 at the same IT load.
Monthly newsletter · free
Keep your reading list current
Once a month: new library sources, changes to Turkish regulations and upcoming data centre events.
SoSo CAD
Review your power design in SoSo CAD
Inspect supply paths, equipment loads and UPS failure scenarios in one model.
Request a demo
Who should read it
Designers weighing capacity and cost of 2N power against four-to-make-three.
Related standards
- Uptime Tier Standard: Topology
- TIA-942
- EN 50600-2-2
- IEC 61439-6 (busbar trunking)
Publication details
- Authors
- Vertiv
- Publisher
- Vertiv
- Year
- 2024
- Document number
- AI Reference Design #014
- Document language
- English
- Source type
- Manufacturer
- Access
- Free to read
- Licence
- Copyright retained; citation only — the publisher’s terms apply to the original document
Citation
Vertiv (2024). Vertiv 360AI High Density Reference Design #014: 2x 1.2 MW NVIDIA Blackwell DGX SuperPODs, 2N. Vertiv. AI Reference Design #014. https://www.vertiv.com/49c616/globalassets/documents/ai-hub-reference-design/vertiv-360ai-reference-design-014.pdf
Events on this subject
Browse the events calendarAll content and rights belong to the respective authors and publishers. SoSo CAD provides only the citation, its own summary and a link to the source, and is not responsible for the sources' content.