AI is transforming the world, but powering and cooling data centres sustainably is one of the defining infrastructure challenges of this decade

By Dani van Elk

AI is transforming the world, but powering and cooling data centres sustainably is one of the defining infrastructure challenges of this decade.

Last week I spent two days in Geraldton at GreenTech Hub's "Green Compute at the Edge" Ecosystem Workshop on data centres. The Mid-West brings together a rare combination: strong complementary renewable energy, low-cost land, and an emerging subsea cable landing point in Geraldton.

As AI demand grows, so will the infrastructure behind it. Data centres are high-power loads, but their needs vary. Some require low latency and high responsiveness, others allow far more flexibility in where and how they run. That variability is the opportunity: an intelligent layer that decides whether a workload can run elsewhere, or wait for lower-carbon, lower-cost power, without compromising latency, accuracy, reliability or service quality.

It points to a bigger shift. The next wave of data centres won't be standalone facilities but integrated utility ecosystems, where power, water and digital infrastructure are planned together. Done right, they can be a growth sector for Australia, and especially Western Australia. And "right" comes down to one question: how do you define the lowest-cost, most reliable, hourly carbon-free energy mix within a network that also captures heat reuse, cooling and shared infrastructure?

So why not close the loop, using AI-based power system modelling to match data-centre demand with renewables, storage and firming hour by hour across the network? The same technology driving demand for these facilities can help us plan, optimise and power them more intelligently, without getting too high on our own supply.

More to come on the work we're doing at Bivios to solve exactly this challenge.