The two great technology revolutions of the modern era -- artificial intelligence and renewable energy -- are colliding.
In September, more electricity was generated by solar and wind in Australia than by coal for the first time. Globally, it happened a few months ago.
In its 2025 Electricity Statement of Opportunities, published in August, the Australian Energy Market Operator (AEMO) predicted that electricity demand from AI data centres would increase at 25 per cent a year for 10 years which, with transport and household electrification, will more than triple the rate of growth in electricity demand generally.
The good news is that nearly 40 per cent of households now have rooftop solar, contributing 13 per cent of the national electricity market in the year to June, up from zero 10 years ago, but the growth in that is slowing as saturation is approached.
The insatiable hunger for power of AI data centres will soon start to squeeze out other industrial users, leading to job losses in addition to those caused by AI directly, and push wholesale electricity prices higher for everybody.
It's challenging for renewables to supply data centres with power because they need it consistently, 24 hours a day, which solar and wind obviously can't do.
And according to AEMO, 16 existing Australian base-load coal-fired power stations are due to be retired over the next 10 years.
With the only other consistent power source -- nuclear -- banned by the federal government and every state, and very unlikely to be unbanned, this is going to put enormous pressure on the storing of power from solar and wind in grid batteries and pumped hydro.
Even in countries where nuclear isn't banned, it's expensive and dangerous, so not preferred. China, racing to electrify its whole country ahead of the rest of the world, is currently building 26 new nuclear power stations, but also about 100 coal-fired ones, as well as about 2,000 wind farms and more than 3,000 solar farms.
Nevertheless, Bloomberg New Energy Finance estimates that 83 per cent of incremental demand from global data centres will come from fossil fuels by 2030, and this falls to only 63 per cent by 2035.
Which is why the world is now urgently stepping up the effort to make fusion energy work.
The pull of limitless power
Fusion is where atoms of hydrogen isotopes are joined (fused) at 100 million degrees Celsius to form helium. It's the reaction inside stars and is safer than fission nuclear power because the fuel is water, not uranium, and the waste is helium not plutonium. More importantly, it is limitless.
Today in Chengdu in China, the International Atomic Energy Agency will hold its 30th global fusion energy conference -- yes, there have been as many of these meetings as UN climate conferences, starting 35 years earlier in 1961, but with much less attention.
That's because fusion has always been the physicists' fantasy, the joke being that it's 30 years away and will always be 30 years away.
Except this year's conference will be held against a background of frantic activity: 150 experimental fusion devices and testing facilities are either operating, under construction or planned, and more than 20 fusion plant designs are being developed.
And there are two Australian outfits in the thick of it: super fund Hostplus and a small private company based in Sydney called HB11.
Hostplus has invested $330 million for a 4 per cent stake in the Boston company that is generally seen as leading the race to commercial fusion in the United States: Commonwealth Fusion Systems (CFS).
China is also racing to develop fusion energy, but not much is known about what it's up to, although last week an American body called the Commission on the Scaling of Fusion Energy published a paper containing this sentence: "The central dynamic of the fusion race is clear: the United States laid the scientific groundwork; China is positioning to win the industry."
Hostplus's chief investment officer Sam Sicilia (who has a PhD in theoretical physics as well as a master of applied finance), told me last week that "This is the biggest potential change in the world's energy mix that we have seen in our lifetime."
He reckons CFS could be a US$1 trillion company, which would make it easily Hostplus's biggest investment.
"We're looking at fusion as clean energy and ... infinite power forever. That's what's at stake here.
He went on: "[The] timeframe is now down to five years and we can see the progress that's been made in the fusion field, we can see the high temperature super conducting magnets that have been produced to hold a very, very hot plasma, the plasma is required to fuse the two hydrogen isotopes together."
I asked him whether fusion power would be cheap enough and he said it would be "much, much cheaper than coal-fired power stations because you don't need to dig up coal and transport it, et cetera."
"It'll also be much cleaner. There's a cost in having the by-products of carbon and ... once you get competition in the fusion space, then that should drive the price down further. It doesn't necessarily need to be cheaper than solar; it just needs to be a technology that's added to the energy mix."
One can imagine every data centre having a fusion electricity generator attached to it, not to mention every aluminium smelter and shopping centre, and township, or even suburb.
Lasers or magnets
The technical challenge with fusion is to contain the 100-million-degree plasma so it doesn't leak out and incinerate everything around it.
There are two ways, at present, to do that: magnets and lasers. Commonwealth Fusion Systems uses magnets: Sam Sicilia says they stack 16 powerful magnets on top of each other, and they could lift an aircraft carrier.
An Australian company, HB11, says it is leading the world in using lasers, and is currently building a prototype in Adelaide, which is apparently one of the best places in the world for lasers.
The company is still mainly owned by the two scientists who started it, CEO Warren McKenzie and 94-year-old Professor Heinrich Hora, as well as some early investors who have funded their work.
Chief operating officer Greg Ainsworth told me they were building a business around high-powered lasers that could shoot down a military drone from about 3 kilometres away, to help fund the work on fusion energy.
He says HB11 expects to have a pilot plant generating fusion electricity in the 2030s.
The other bit of good news is that fusion nuclear energy doesn't seem to be caught by Australia's moratorium on nuclear energy. It certainly shouldn't be -- that would be very stupid.
The word "fusion" didn't get a mention in this year's "Statement of Opportunities" from AEMO, but it won't be long before it does figure in the official 10-year outlook for electricity supply.