The investment reality: unprecedented capital intensity

Access to capital is critical but does not on its own solve the delivery challenge.

AI-oriented data centres are much more capital intensive than conventional facilities. A single 100MW facility designed for AI compute workloads now costs A$1.5–2.0 billion to construct.¹ Securing grid connections, substations and transmission infrastructure also requires substantial upfront capital, often years before capacity is fully leased.

Investors increasingly view data centres as infrastructure-like assets, supported by long-dated customer contracts of 10–15 years (or longer). But the upfront capital intensity, combined with long construction timelines and limited ability to manage back-to-back delivery risk, creates a risk profile that is meaningfully higher than for traditional infrastructure, even where revenue appears stable.

Unlike legacy PPP or utility models, data centre developers:

• can't fully transfer construction and delivery risk to contractors;
• carry considerable operational and timing risk themselves; and
• have limited negotiating leverage with hyperscale customers.

This means that even where demand is clear and capital is available, the gap between committed investment and delivered capacity continues to widen.