The Australian Energy Market Operator Estimates Data centers will eat 6% of Australia's grid-powered electricity by 2030.
To put that in context: This is greater than that current share the Australian health and social care industry.
This reflects the rapid growth of Australia’s data center industry – the backbone of artificial intelligence (AI). This growth is fueled partly by billion-dollar investments from major technology corporations reminiscent of AWS, Microsoft, CDC and NextDC. Atlassian co-founder Scott Farquhar even has really useful Australia could grow to be the information center hub of Southeast Asia.
The federal government can also be fertilizing the information center industry. In August, for instance, Treasurer Jim Chalmers announced the event of “national interest principles for data centers” in consequence of the Round Table on Economic Reform.
However, the power-hungry nature of knowledge centers poses major problems for Australia's current energy network. However, there are three steps Australia can take to offer reliable and sustainable power to those facilities.
Increased volatility, increased risks
Unlike homes and most industries, data centers require a continuing power supply. This increases pressure on an energy grid designed for variable consumption.
As more people use AI for more complex tasks, data center workloads will increase. This results in an increased base load requirement. But it also results in unpredictable peaks and falls in demand that the network will not be designed to handle. This volatility poses real risks.
In 2024, there can be 60 data centers in Northern Virginia suddenly separated taken offline resulting from a triggered safety mechanism. This triggered an enormous surge in excess electricity, which, if grid operators had not taken immediate countermeasures, would have resulted in an enormous blackout.
This near miss highlighted the fragility of the network within the face of sudden, widespread data center shutdowns.
Clean energy alone cannot do that
The limitations of Australia's current energy mix are one other source of volatility.
While renewable energy is central to the clean energy transition, it alone cannot meet the baseload and peak demand of knowledge centers. The problem is twofold. First, renewable energy is sporadic. Second, energy storage and backup options have limited scalability.
This means that the majority data centers will proceed to depend on coal or gas in some form.
Most data center operators have committed to using 100% renewable energy by 2030. In practice, nevertheless, this often means purchasing annual renewable energy credits or power purchase agreements.
These mechanisms don’t guarantee clean energy in actual operation – they only help balance annual consumption. Meeting real-time demand with clean energy is a much more complex challenge. It requires greater investment in renewable energy, storage and transmission infrastructure. Better coordination between energy regulators, utilities and data center operators can also be needed.
These challenges were reflected in Australia's latest climate goal – a discount of 62-70% below 2005 levels by 2035. This is below A variety of 65-75% was originally suggested from the Climate Change Authority last yr. Why the reduction? Among the cited “Transition risks“is the numerous growth of knowledge centers.
Mick Tsikas/AAP
Become a worldwide champion
Australia has a chance to develop policies that synchronize data center expansion with more efficient energy and grid management.
First, Australia should promote large-scale computing methods that reduce emissions but don’t compromise performance.
For example, intelligent planning software can handle energy-intensive tasks reminiscent of: B. model training, mechanically shift to off-peak times when renewable energy is commonest. This wouldn’t impact more mundane, less energy-intensive tasks reminiscent of using ChatGPT that require quick responses. Pursue like Google have already adopted this approach to cut back network load without impacting user experience.
Additionally, data centers needs to be required to notify energy providers before large-scale AI training runs that may end up in dramatic energy spikes. Pursue like Hitachi Energy have called on governments to introduce such rules to support grid management, pointing to other energy-intensive industries reminiscent of smelting where advanced warnings are already common practice.
Second, Australia must speed up advanced energy storage innovations, including batteries, pumped hydro and thermal energy storage. Many of those technologies are already being researched in progresssupported by government initiatives and personal investments.
This is, for instance, the information center company AirTrunk explore different paths to implement battery energy storage systems in its latest data centers. However, more targeted financial incentives and support – reminiscent of through Future Made in Australia and the National Reconstruction Fund – may also help close the gap between research and business scalability.
Third, Australia may require data centers to set so-called “Power Usage Effectiveness” or PUE targets to drive energy efficiency.
PUE targets are calculated by dividing the information center's total energy consumption by the energy consumption of its IT equipment. A PUE value closer to 1.0 indicates higher energy efficiency.
PUE limits in China helped to cut back its average PUE from 1.54 to 1.48 in only one yr. Likewise voluntary initiatives reminiscent of those of the European Union Code of Conduct for data center energy efficiency have continually reduced the typical PUE amongst participating facilities.
There is not any escaping the truth that data centers are energy-hungry behemoths. However, with the appropriate planning and policies, Australia could possibly be a worldwide leader in data center growth that supports, quite than derails, the clean energy transition.
