Peak eccentricity

Bizarre renewable energy schemes won't meet the demand for reliable electricity in the AI era

Renewable energy superpower status is supposedly in Australia's grasp now that the federal government has given Mike Cannon-Brookes the green light to export solar power to Singapore.

Tanya Plibersek announced environmental approval for the tech billionaire's eccentric proposal last week, taking a swipe at Opposition Leader Peter Dutton’s “expensive nuclear fantasy that may never happen.”

By contrast, the Environment Minister would have us believe that Cannon-Brooke’s plan to siliconise the NT Outback is a done deal. All that's left to do is raise $35 billion ($US 24 billion) in capital, install 120 square kilometres of solar panels, build a modest 788 km transmission line to Darwin, and lay a 4200 km high-voltage cable on the seabed, and we're good to go.

The Sun Cable AAPowerLink project feels like it was stolen from a Heath Robinson cartoon: a convoluted, unnecessarily elaborate, and impractical contraption designed to accomplish a mundane task. It may mark the beginning of the end of the renewable romance in Australia, the point at which the transition to wind, solar and hydropower collapses under the weight of its own absurdity.

There is increasing evidence that the US has reached the point of peak renewables as the pool of private investors shrinks and winning community approval becomes harder. Research by the Lawrence Berkley National Laboratory showed roughly one-third of utility-scale wind and solar applications submitted over the last five years were cancelled, while about half of wind and solar projects experienced significant delays. The US Department of Energy says the national electricity network needs to grow by 57 per cent by 2035, the equivalent of approximately 21,000 km a year. Last year's total was around 200 km, down from just over 1000 in 2022.

Meanwhile, the challenges of grid synchronisation and storage at scale remain unresolved, and the technical problems for offshore wind turbines, in particular, are mounting.

Last week, turbine manufacturer GE Vernova announced an investigation into a blade failure in the 3.6 GW Dogger Bank project in the North Sea off the coast of the UK. It is the third blade failure this year. In July, a newly installed blade crumbled at the Vineyard Wind offshore plant, creating debris that washed up on Nantucket Island, Massachusetts. At 107 metres long and weighing 55 tonnes, they are the most enormous blades deployed commercially. The failure of three in quick succession suggests the quest to increase output by installing ever-larger blades has reached its natural limits.

Yet the imperative of expanding generating capacity is hardening. The principle driving force is not electric vehicles but the rapid growth of artificial intelligence. AI requires at least ten times the power of conventional computing programs. Exabytes of accumulated data must be stored in the cloud, which is actually a series of vast air-conditioned warehouses housing storage housing tens of thousands of servers dissipating tens of millions of joules of excess energy every second in the form of heat.

In the US, Data centres account for about 2.5% of power and demand could rise to 7.5% by 2030 according to Boston Consulting Group. In Ireland, data processing and storage use 12 percent of electricity produced, forcing the authorities to limit the number of connections to the grid.

Silicon Valley has long abandoned the notion that it can be powered by silicon photovoltaic panels while burying stray emissions in the Amazon forests. In April, the tech giant Amazon paid the best part of $1 billion ($US650 million) for a sizeable block of land next to Pennsylvania's Susquehanna nuclear power station. It will be the site for a hyperscale data centre powered by up to 480MW of carbon-free electricity delivered reliably around the clock using energy from a nuclear reactor.

Shares in US nuclear power companies like Consolidated Energy, Talem, and Vistra have soared by between 80 percent and 180 percent in the last year. So-called green energy stocks, on the other hand, are static or falling, while coal is making an unexpected comeback.

In May, the Financial Times reported that the retirement dates for coal-fired power stations are being pushed back as operators become concerned about grid security. Allianz Energy has delayed the conversion of its Wisconsin plant from coal to gas for three years to 2028. Ohio-based FirstEnergy announced in February that it was scrapping its 2030 target to phase out coal, citing “resource adequacy concerns”.

The effect of AI on electricity demand was largely unanticipated at the beginning of the decade. AI chips will undoubtedly become more efficient, but there is no telling how much further the demand for AI will grow since the technology is in its infancy. Nor can we begin to guess what other power-hungry forms of technology might be developed by 2050.

What we do know, however, is that if Australia's demand for electricity exceeds 313 TWH a year in a 2050, we're in trouble. That's the target the Australian Energy Market Operator (AEMO) has set in its updated Integrated System Plan (ISP), the official roadmap for the great energy transition.

As Energy Minister Chris Bowen points out, that’s going to take a lot solar panels and wind turbines. Bowen has set a target of 22,000 new solar panels a day and a new 7MW wind turbine every 18 hour to meet AEMO’s interim 2030 target of a mere 202 TWH. For the record, the speed of the rollout in the first two years of Labor government is less than a tenth of that.

One of the hallmarks of the anointed is an unwavering conviction in the integrity of their analysis and the effectiveness of their proposed solutions. They feel no need to hedge their bets by factoring in contingency arrangements should their predictions turn out to be wrong.

Nothing in AEMO's ISP indicates that its experts have given any thought to scaling up electricity production in line with actual demand, which may well be considerably higher than they have anticipated.

If they had, they would have to acknowledge that there are limits to the renewable energy frontier determined by energy density, the demand for land, and the requirement for firming. The silicification of northern Australia cannot continue forever, nor can we expect to rely on China for most of the hardware and pretend there are no geopolitical consequences.

AEMO would have recognised that the only proven low-carbon technology capable of filling the gap is nuclear, and that the moratorium on nuclear power in Australia simply cannot be sustained.   

As for our nuclear-phobic Prime Minister's dream of turning Australia into the Saudi Arabia of green hydrogen while simultaneously sitting at the cutting edge of quantum computing, forget it. Both technologies need massive amounts of power that won’t be available at anything resembling a viable cost if it can be found at all.

In 2006, as the shadow minister for the environment, Anthony Albanese gave a speech at the Swansea RSL on avoiding dangerous climate change. “Why on earth would we want to take the big health and economic risk of nuclear energy when we have a ready-made power source hovering peacefully in the sky every day?” he asked.

If he doesn’t know the answer to that question 18 years later, he probably never will. 

Nick Cater is a senior fellow at Menzies Research Centre and a columnist with The Australian newspaper. His articles are available via Substack at nickcater.com