Artificial intelligence is shaping up as the hottest topic of the year. Analysts are predicting how it would benefit virtually every industry, and developers are eager to try it everywhere they can. But there is one potentially huge problem with AI: its energy and raw material requirements.
The energy problem surfaced last year, with one scientist doing the maths on how much electricity data centers hosting AI hardware would require. Warnings began coming in saying that the utilization of artificial intelligence will reverse the transition from oil and gas to wind and solar because data centers require a reliable, uninterrupted supply of electricity. For now, that only comes from hydrocarbons and nuclear.
But energy consumption is just one aspect of the AI problem that the energy transition seems to be facing. The other may be much more serious: raw materials.
"Just to meet demand growth before AI we need a lot of copper," White House energy adviser Amos Hochstein said at a recent event, discussing the prospects of critical mineral demand. "With the advent of growth in AI, we're going to need a significant amount of power generation and transmission, which is going to require even more," Hochstein added, as quoted by Politico.
The problem is that wind and solar installations also need a lot of copper because they require a lot of cables. When you add electric cars, the future looks extremely bright for copper miners and just as bleak for everyone who needs copper.
Investment research firm the Oregon Group last month released a report predicting a 10-year supercycle in critical minerals. This supercycle was not going to be driven by the transition, however. It was going to be driven by artificial intelligence.
"The [data center] market is moving a lot faster than supply chains that have historically supported a very physical business have been set up to do... It's a sprint. It's a sprint that requires all the capital in the world," the co-founder of CoreWave was cited as saying in the report, providing a good summary of the situation,
The information technology industry is planning to spend hundreds of billions of dollars on artificial intelligence, meaning new data centers and new electricity networks. With that, it has automatically become a competitor for transition technology developers—because they all need the same materials.
Copper is a good example. It is not rare, it is not expensive but both AI evolution and the transition need large amounts of it—and there are no new mines being opened at the rate they need to be open to support the coming surge in demand.
Warnings about a copper shortage have become part of the news landscape over the past couple of years and yet the mining industry has been extremely slow in doing anything about this, possibly because the bullish demand forecasts have not always panned out, notably in EVs. So miners are biding their time until the strong demand does begin to show signs of emerging.
"You can cynically say that the mining execs are just kind of looking for the next story to underpin the typical narrative, which is: 'Demand is going to increase faster than supply, and we're staring structural deficits in the face,'" Chris Berry, battery supply chain consultant, told Politico.
Yet this is not leading to new copper mines—but it is leading to big plans about sourcing the necessary critical minerals. In his speech at this week's Milken Institute's Global Conference, Hochstein urged miners to go into risky jurisdictions and ignore the risk to secure the minerals and challenge China's dominance in the area.
"We can all live in the capitals and cities around the world and say, 'I don't want to do business there.' But what you are really saying is we're not going to have an energy transition," Hochstein said, as quoted by Reuters.
However, It is doubtful whether miners will respond to what is effectively a challenge to reprioritize, putting the transition above things like security and predictability. Besides, miners appear to be conscious of the fact that China's dominance in critical minerals is the result of decades of work—and that kind of work cannot be replicated in 24 months.
The situation does not look good—AI and the transition are vying for one pool of resources. But there is a silver lining, although it is not the silver lining transition champions like to speak about. The massive demand growth in demand for various metals and minerals coming from transition projects has basically failed to materialize so far. Lithium prices are down. Copper is down. So is EV demand. Miners are justified in their wait—and—see approach.
Yet, unlike the transition, artificial intelligence looks ready to take off. There is quite a bit of momentum building up behind it, and all it really needs is metals, minerals, plastic, and electricity. Then AI will become a real problem for the transition.
By Irina Slav for Oilprice.com