Datacentres are a ticking time bomb. We must make sure AI’s benefits outweigh the costs | Nicki Hutley
TL;DR
Nicki Hutley frames the datacentre boom as the collision of two major pressures: the climate crisis and AI. More than 10,000 datacentres are already active worldwide; McKinsey estimates the sector could grow 3.5 times and require about US$7tn in investment. Australia is a key case: 286 active or planned sites, plus interest from companies such as Anthropic. Hutley argues governments too often label datacentres as infrastructure without proving who benefits and how.
Nauti's Take
The strong argument is not that datacentres are bad. It is that they are not automatically good just because AI is attached to them.
Anyone directing billions into server halls, grid connections and water use should have to show the public upside. Productivity promises are not enough if households, local communities and the climate get the bill first.
Briefingshow
The piece moves the AI debate away from abstract model risk and toward physical costs: electricity, water, heat, grids and land. If governments treat datacentres as inevitable future infrastructure without rigorous cost-benefit tests, they may end up socialising the resource burden while private AI firms capture the upside.
Sources