Can we power both humanity and machines with cheap renewables?

As the AI revolution charges ahead, reshaping manufacturing, construction, and the wider industry, a deeper question lurks beneath the shiny narratives of productivity and efficiency: who gets priority in the energy economy of the future: humans or machines?

In traditional factories, energy was largely spent on lighting, HVAC systems, and the welfare of human workers. These were predictable, manageable demands. Robots and AI systems, however, flip the equation. They run 24/7. They require constant electricity for movement, computation, cooling, and communication.

Autonomous mobile robots (AMRs), the sensors that draw data from outside to feed into digital twins, and the machine learning models processing external variables that ensure the optimum manufacturing conveyor belt all consume energy in vast quantities. To remain competitive, the Australian manufacturing industry is increasingly turning towards autonomous robots which increases energy demands within the manufacturing environment.

This is in addition to the energy demands of AI systems from sensors that continuously ingest data from the environment, process and analyse data in real-time to schedule automation workflows. With each step in this closed-loop system, another layer of energy demand is added to the mix. 

As these AI systems scale, they require the kind of hyperscale data centres that are increasingly rivalling aluminium smelters in electricity consumption.

According to Morgan Stanley, Australian data centres could use 8 per cent of Australia’s total power generation. These centres demand high-voltage connections, liquid cooling infrastructure, and back-up generators. 

We can argue that renewable energy such as wind and solar PV and Australia’s investment in large-scale batteries could easily fill the increasing demand for energy consumption. However, we have yet to test if the rise of AI and growth of Australian data centres could stress the demand for energy.

A March 2024 report by SemiAnalysis predicts a 900 terawatt-hour (TWh) increase in global electricity demand from AI and data centres by 2030 – equivalent to almost 4.8 times of Australia’s National Electricity Market (NEM) consumption in the 2023 financial year.

Do we need a tiered energy system?

Humans, AI and robots will compete for the same energy resources and now is the time to rethink energy allocation itself. We need a tiered system, one that distinguishes between Energy for Humans and Energy for Robots.

Energy for Humans covers residential, commercial, and essential services – things like housing, hospitals, schools, and the human-facing parts of industry.

Energy for Robots includes the high-intensity power demands of data centres, autonomous machines, and AI-driven production facilities.

Without deliberate governance, AI-powered industries could leverage their size and influence to negotiate bulk rates, preferential access, and firm renewable energy contracts, potentially pricing out human consumers. The emerging industrial oligopoly of hyperscale AI providers could result in energy market distortions where robots get cheaper, greener power, while households pay more for the same electrons.

The question remains around who maintains control over the allocation of Australia’s renewable energy assets. Will it be up to government agencies, planners and regulators to ensure there is equitable access, or multinational corporations who are securing clean energy and Power Purchase Agreements for their data centres and factories?

Without policy safeguards, we risk repeating past mistakes—where industrial priorities supersede human needs. The aluminium industry’s long-term, low-cost power deals with state governments left many wondering whether the public got a fair return. Now, history threatens to repeat itself on a much larger scale.

Powering progress, protecting people

We need forward-thinking regulation that recognises the divergent energy demands of humans, AI and robots. Here are some of our policy ideas:

– Set energy allocation targets: Mandate a minimum percentage of renewable energy dedicated to essential human needs—housing, health, education—before AI and robots use.

– Energy pricing transparency: Require public disclosure of bulk energy deals made with factories that use AI and data centres. In the same way that transparency in AI development leads to greater trustworthy outcomes, ensuring that these deals are visible will allow the public to interrogate the fairness of these energy deals.

– Human-first firming: Prioritise energy storage (large-scale solar, batteries, pumped hydro) for residential and critical services, ensuring the demand for energy isn’t compromised by demand spikes from data centres and robots.

The growth of data centres, AI and automation are transforming industries, but we cannot afford to prioritise Energy for Robots over Energy for Humans. As we automate more of our economy, we need to be just as deliberate about who gets the power, how it’s priced, and who benefits.

A study in Renewable and Sustainable Energy Reviews (Basilico et al., 2023) highlighted that models that promote the adoption of renewable energy must ensure equitable treatment of smaller community members and balance the interactions between energy consumers and the electricity grid.

Australia has an opportunity to lead here. We can build an energy system that is smart, equitable, and sustainable – where AI-driven innovation doesn’t come at the cost of human well-being.

But to do so, we’ll need courageous policy, transparent markets, and a new social contract for the energy age of AI.

Dr Brydon Wang is associate director, commercial advisory and transactions, deal advisory and infrastructure at KPMG Australia

Gabriel Wong is co-founder and director of Positive Good

Want the latest clean energy news delivered straight to your inbox? Subscribe to our free daily newsletter.