Here's a grim question. Which poses the bigger threat to humanity: climate change or artificial intelligence?
Andrew Forrest, who is staking billions on both challenges, has an equally grim answer: "Let's just say that AI will be a lot faster. Maybe a lot less painless."
The Fortescue boss was in New York this week, a place heaving with world leaders and a foreboding sense the world is suddenly at an urgent and critical juncture on AI.
"AI may decide the best way to get rid of climate change is to get rid of humanity," Forrest told the Insiders on Background podcast during a reflective conversation from an office overlooking Manhattan.
If humans don't maintain control an off button for AI, he said, "it may well decide to have the off button for us".
This bleak warning is not coming from someone opposed to the technology. Forrest was an early and enthusiastic champion of artificial intelligence and now has ambitious plans to invest heavily in the industry's future.
The scale and success of Fortescue's green transition at its Pilbara iron ore operations, it seems, has caught the eye of the world's biggest AI players.
"We've got data centre companies, frontier AI companies, hyper-scalers super interested because our product is green and our product gives some independence from the grid," Forrest said.
At an estimated cost of more than $10 billion, Fortescue is set to deliver more than 2 gigawatts of solar and wind power along with enough battery storage to power fully electric trucks and mining equipment across its operations.
The goal now is to repeat this exercise "many times" over to build renewable-powered, off-grid data centre hubs across Australia.
"If Nvidia wants us to consume their chips and they want to do joint ventures with us, we're going to look at it," he said. "If Anthropic, Open AI, or any of the other great players — and there's hundreds out there, we're only mentioning a few."
Forrest, it seems, is preparing to dive into the data centre business.
Loading...
The two sides of AI
Forrest's capacity to dream big about jumping on board the AI revolution while at the same time acknowledging the technology could end us all is a perfect example of the dilemma being confronted in New York this week.
World leaders, including Anthony Albanese, are looking to AI as a saviour with the capacity to kickstart productivity, drive growth and solve all manner of health and other problems.
There's just that small issue of its potential to also wipe us out.
The joint plea at the UN this week from various middle power leaders, including Albanese, for AI to "remain under human direction, oversight and control" was entirely uncontroversial. Their call for common global safety standards was consistent with how the world might once have dealt with rapidly emerging threats.
But once again the limits of middle-power influence were realised.
AI can be added to the list of things, including ongoing wars and trade battles, the UN has been unable to control.
The two big powers who really matter when it comes to AI are ignoring the rest.
China's President Xi Jinping didn't even bother attending the UN General Assembly, skipping it for a direct meeting with Donald Trump at the White House instead.
Trump did attend, using his lengthy speech to forcefully reject the calls for international controls over what he prefers to call "super intelligence".
"I'm not going to stifle growth of something that will be bigger than the industrial revolution," he told his nervous fellow leaders, who sat mute through these comments.
Trump and Xi are set to discuss a new "AI dialogue" between the two big powers, with a mechanism to alert each other to incidents that affect national security. It's something, but nowhere near the clear, preventative safeguards middle powers would prefer.
Loading
Al industry left to police itself
On one level, this week's unsuccessful push to put some international controls in place is simply another example of the world we're now in, where the big powers set their own rules and do what they like.
On another level, this a far more serious fracture, given the stakes facing humanity.
The inability of the UN this week to achieve any sort of unified position, given the alarm being expressed by the tech industry itself, felt like a pivotal moment.
Albanese, for his part, remains reluctant to directly criticise Trump on this. But clearly, the US president's refusal to accept the risks of rapidly advancing AI and to do something about it is not the outcome the Australian prime minister was hoping for.
As for that grim question on the greatest threat, Trump argues fears about both climate change and AI are simply a "hoax".
It's now left to the tech industry to police itself.
