← All perspectives

CAPITAL ALLOCATION

AI infrastructure: follow the power, then the capital.

An ambitious data-centre pipeline does not become operating capacity simply because funding is available. Power, equipment and delivery schedules have to arrive together.

A forecast, with a date attached

The IEA’s April 2026 report projects global data-centre electricity consumption rising from 485 TWh in 2025 to around 950 TWh in 2030. The 2030 figure is a projection, not an observed outcome. The report also identifies constraints across electricity infrastructure, equipment and chips.

Our interpretation: timing is part of the underwriting

A project can be exposed to a gap between when capital is committed and when revenue begins. Grid connection delays, equipment lead times and customer ramp-up can extend that gap. The financing structure needs to absorb it.

For an Indian business evaluating participation in the supply chain, a global demand forecast is only the starting point. The commercially relevant questions are local: who is the customer, what is contracted, when is cash collected, and what capacity must be funded first?

Three possible trajectories

Capacity catches up: improved equipment availability and timely connections support commissioning. The focus moves towards utilisation and cash generation.

Physical bottlenecks persist: announced projects take longer to complete. Working capital and funding tenor become more important, even where customer demand exists.

Demand or financing disappoints: slower adoption, efficiency gains or tighter capital availability can weaken project economics. A supplier’s exposure depends on contract protections, customer concentration and payment terms.

What to track

  • Commissioned capacity against announced capacity.
  • Power connection milestones and equipment delivery commitments.
  • Contracted customer demand, utilisation and cancellation provisions.
  • Cash conversion, customer concentration and funding headroom.

These are diligence questions, not a recommendation to buy a sector. A growing end market and an attractive business are separate conclusions; the second needs company-level evidence.

Keep exploring.