Australia’s largest home-grown data centre operator says the artificial intelligence boom has accelerated the land and energy challenges posed by the last decade’s transition to cloud computing.
NextDC runs a network of 15 data centres in Australia, with four more in planning or development, part of a business worth $10.3 billion. Chief customer and commercial officer David Dzienciol says the uptake of artificial intelligence is driving rapid expansion, but with unstable energy prices amid global conflict and the transition to renewable energy, will “no doubt” place pressure on energy grids.
David Dzienciol, COO of NextDC, says artificial intelligence is just “the next big wave” for the data industry.
“We will need, collectively, to be looking at all alternatives to ensure that we can cope with demand,” he says. The company’s CEO, Craig Scroggie, has previously said nuclear energy should be considered to support greater computing power.
Australia’s data centres use around 5 per cent of the nation’s power, or around 1050 megawatts in 2024 – similar to South Australia’s entire consumption – rising to anywhere from 8 to 15 per cent by 2030, according to research by Morgan Stanley. The Australian Energy Council says a large data centre can consume the same energy as 50,000 homes, or a small city.
The industry also faces high industrial property prices in the major centres, with NextDC saying in 2021 that it would pay $124 million for a site now in planning in Sydney’s Horsley Park. Dzienciol says the company is seeing “workloads start to move out more to the edge”: it has a site in Port Hedland and another in development in Newman in northern WA, and has just opened its first data centre in Darwin.
NextDC was founded in May 2010, in a decade which saw data move out of server rooms in offices to remotely accessed “cloud” storage with data centre operators.
It announced this month it had been certified by US tech giant Nvidia to support Australian organisations to use the company’s platforms, which make use of GPU (“graphics processing unit”) chips that have driven investors wild, earning Nvidia a prime spot in the “Magnificent Seven” tech stocks.
Dzienciol said the brute strength of GPU computing, where hundreds or thousands of less powerful units run the same processes at high speed and volume, better supports artificial intelligence than traditional CPU (“central processing unit”) computing, whose more versatile, brain-like systems can switch between processes, but at lower volumes.
Source Agencies
