The Double-Edged Sword: AI Boom and Datacentre Growth in Australia

The two great existential threats of our time – the climate crisis and AI – come hurtling together in the explosion of datacentres across Australia and around the world.

You can hardly avoid hearing about them these days, either with awed reverence of the promised benefits to humankind or with fear and anger given the implications for the climate, inflation, jobs and even housing affordability.

And that’s before we get to the implications of artificial intelligence itself – to me, both inspiring and terrifying – which is the primary driver of the datacentre boom.

Worldwide, there are more than 10,000 active datacentres, with this number expected to increase by 3.5 times at an estimated cost of US$7tn. For perspective, that’s a little over 5% of the entire world’s annual GDP – we are talking big bickies. The US hosts most of these centres but Australia is attracting activity, with 286 active or planned centres and global AI leaders including Anthropic looking to Australia as a potential training ground for its models.

The economic, environmental and social consequences of this datacentre investment boom are profound. Yet governments here and around the world are, by and large, taking a laissez-faire approach – perhaps from Fomo on the alleged benefits or from fear of upsetting the billionaire tech bros, or both.

Governments and proponents of datacentres often refer to them as “infrastructure”, which certainly sounds like something we need. But they are neither so-called “hard” infrastructure (think roads, telecommunications or power and water) nor “soft” infrastructure (healthcare or education). Unlike roads or education, it is unclear who is benefiting from all this investment (aside from the tech bros) or how. If we are going to call datacentres infrastructure, they should have to face examination as to whether their benefits outweigh the costs, just as any other projects would.

There’s no doubt that AI can benefit humankind – and I don’t mean getting help designing your travel itinerary or anti-tax meme. In Shanghai, it’s relieving congestion; around the world it’s improving diagnosis accuracy and speed for X-ray, CT, MRI and other imaging; and it is helping optimise energy grids to avoid blackouts. The potential economic and social benefits are enormous. But we cannot look at these benefits without assessing the costs.

And those potential costs are large. The Australian Prudential Regulation Authority has written to banks to warn of the accelerating cybersecurity risk posed by AI. Its recommendation, without any irony, is to use AI tools to help prevent the AI threat.

It is by now fairly well-known that datacentres use huge amounts of energy and water. Datacentres in Australia are expected to triple our consumption of both by 2030. At a time when our best answer to the climate crisis is to electrify as fast as possible using renewable energy and storage, allowing these energy vampires to strain grids could slow the transition to net zero emissions – and add to energy costs for everyday consumers.

While fossil fuels still power about half our energy demand, we will be adding huge amounts of polluting greenhouse gases to the atmosphere. Alarmingly, Queensland says it’s happy to keep using fossil fuels for datacentres, resisting the federal government’s “expectations”. Any cost-benefit analysis must include the collective impact of datacentre emissions.

Waste heat from datacentres is also a significant problem: intense energy going in turns into heat. While this might be useful in cold climates such as Finland, where the waste heat is used to heat homes, in most parts Australia we are already facing more days of extreme heat as the planet warms.

Of the potential to boost economic growth and employment, while the datacentre boom has lifted business investment off the floor over the last year, most of the equipment must be imported. This means that the direct effect of the investment on the size of our economic pie is close to zero. Beyond the construction phase, datacentres do not create many jobs – far less than other sectors such as manufacturing.

When Australian politicians or industry proponents talk about the benefits of datacentres, they are really talking about the possible benefits of the AI that they enable, and especially the productivity gains AI is expected to drive, whatever the size and timing of these might be. In a speech to the Australian Business Economists in February, the assistant minister for science, technology and the digital economy, Andrew Charlton, noted that Australia was now at a crossroad. From here, we could continue to be a “technology taker”, with some productivity benefits, or we could become “a world-class adopter and creator and exporter of AI technology”. Australia’s poor past record on commercialising our ideas and keeping the profits at home suggests the better option will also be much harder.

Charlton also said the government should ensure “that technology works for the Australian people, and not the other way around”. Looking at the datacentre and AI landscape and their associated costs, it has not succeeded.