Skip to content

Research insight · Infrastructure

· 2 MIN

AI-ready is a capacity question, not a label.

A hall is AI-ready when its power, cooling and commissioning evidence prove the density it is allocated at — not when a brochure says so.

The density gap.

Uptime Institute's 2026 global survey puts the most common rack density at about 11 kW, up from 9 kW a year earlier. Rack-scale AI systems sit in a different class: NVIDIA's reference architecture for its GB300 NVL72 rack calls for up to about 142 kW per rack.

Uptime's guidance is that air cooling becomes impractical above roughly 50 kW per rack. AI readiness is therefore not an incremental upgrade of an existing hall; it is a step change in what the power and cooling chain must prove.

The demand behind it.

The IEA's Energy and AI report projects global data-centre electricity use to more than double to around 945 TWh by 2030, with AI-optimised facilities the main driver.

Capacity allocated against that demand will be tested by real load, not design intent.

What “ready” has to prove.

Four questions decide it: what the electrical chain can carry at the rack, measured not nameplate; what the cooling chain can reject with redundancy intact (for example a CDU still at N+1 under full load); what commissioning has verified versus what is still assumed; and what is already committed to customers.

Where any answer rests on design documents alone, the readiness claim is a forecast.

THE CLAIMTHE POSITION

From label to position.

The engine treats “AI-ready” as a position, not a tag: evidence from systems, meters, commissioning records and contracts is reconciled into one trusted position, the gap to the claimed density is quantified as a P10–P90 range, and agents prepare the options to close it.

People decide which to fund, and the capacity is counted only when measured.

In short

The useful question is not “is this hall AI-ready?” but “how many verified kilowatts per rack can we release here today, and what would it take to prove more?”

Sources

  1. Uptime Institute, Global Data Center Survey 2026 (as reported by Data Center Knowledge). https://www.datacenterknowledge.com/ai-data-centers/ai-rack-density-s-real-limits-power-cooling-failure-risk
  2. IEA, Energy and AI (2025), executive summary. https://www.iea.org/reports/energy-and-ai/executive-summary

All insights

Research

The research behind the insight.

turns research, industry expertise and verified outcomes into methods the engine can run.

Explore Research