Part 1: Designing next-generation data centres - Principals for efficiency and resilience

Overview

This piece is the first of two Beca articles setting out design principles for data centres in Australia and New Zealand, dated 2026. It covers hall layout and containment, rack density ranges, modular plantrooms and MEP skids, and phased expansion with space reserved for liquid-cooling retrofit. It stresses right-sizing space early while planning incremental growth. The piece rests on a consultancy's regional experience and presents its principles as design guidance rather than a numerical calculation method.

Typical rack densities are quoted as 20-50 kW, with AI and HPC reaching about 100 kW per rack. Redundancy guidance says N+1 is cost-effective, 2N suits very high uptime and 2N+1 suits systems that must avoid any downtime. The article summarises redundancy levels from N+1 to 2N+1 and Uptime Tiers. Seismic design to AS/NZS 1170 is part of resilience in the region.

Modular plantrooms and MEP skids allow off-site build, and space should be left for added UPS modules, cooling modules, switchboards and future liquid cooling. Different spatial densities should be planned for AI clusters and general compute, with maintenance clearances and future row expansion reserved. The piece is a freely accessible consultancy guide; only the record's summary and key facts were used here and the full text was not read.

Key points

1. Typical rack densities are quoted as 20-50 kW, with AI and HPC reaching about 100 kW per rack.
2. Redundancy guidance: N+1 cost-effective, 2N for very high uptime, 2N+1 for systems that must avoid any downtime.
3. Seismic design to AS/NZS 1170 is part of resilience in the region.
4. Modular plantrooms and MEP skids allow off-site build; leave space for added UPS modules, cooling modules, switchboards and future liquid cooling.
5. Plan different spatial densities for AI clusters and general compute, and reserve maintenance clearances and future row expansion.

Using it in design

• Allow mixed-density zones within one hall, each with its own density assumptions.
• Define reserve-space objects for future UPS and cooling modules and liquid-cooling piping.
• Include the AS/NZS 1170 seismic zone as a site input for Australia and New Zealand projects.
• Choose the redundancy level (N+1, 2N, 2N+1) by linking it to the client's uptime requirement.

Who should read it

Designers and investors planning data centre layout, redundancy and phased expansion in Australia and New Zealand.

Related standards

• AS/NZS 1170
• Uptime Institute Tier Standard: Topology

Publication details

Authors

Akshat Malhotra; Alex Grant; Stuart Ritchie

Publisher

Beca

Year

2026

Document language

English

Source type

Expert / community

Access

Free to read

Licence

Copyright retained; citation only — the publisher’s terms apply to the original document

Citation