Power demands of AI present ‘huge’ investment opportunity, BlackRock’s Fink says

AI is 'capital intensive, compute intensive and energy intensive,' General Atlantic chief says at WEF meeting in Saudi Arabia

Larry Fink, chairman and chief executive of BlackRock, said 'the amount of power that's needed to use AI has a huge impact on society'. Reuters

The development of artificial intelligence (AI), with its significant power consumption, is raising questions about the impact on energy prices and the grid, the chief executive of BlackRock, the world's largest asset manager, said on Monday.

Despite the potential challenges, the need for increased infrastructure presents a “huge” investment opportunity, Larry Fink said at the special meeting of the World Economic Forum in Riyadh.

“The amount of power that's needed to use AI has a huge impact on society. Where's that power going to come from? Are we going to take it off the grid? What does it mean for elevated energy prices?” Mr Fink said.

“Forget about the use of it, but just the generation of it [requires] massive power … but that is a huge investment opportunity.”

The International Energy Agency has forecast that data centres' total electricity consumption could reach more than 1,000 terawatt hours (TWh) in 2026, which is roughly equivalent to the current power demand in Japan.

“Updated regulations and technological improvements, including on efficiency, will be crucial to moderate the surge in energy consumption from data centres,” the agency said in a report this year.

Meanwhile, parallels have been drawn between the AI boom and the tech bubble of the late 1990s, a period marked by speculative investment in many internet-based companies.

While the tech bubble was largely centred on the internet and related technologies, such as e-commerce and online services, the AI revolution encompasses a broader range of technologies and applications, including natural language processing, robotics, and autonomous systems.

“The difference between this technology cycle around AI and the other ones is that it's very capital intensive, compute intensive, and energy intensive,” William Ford, the chief executive of General Atlantic, said during the same panel session.

“The internet cycle was not capital intensive at all. It could be widely available and actually connected up the world,” Mr Ford said.

Some of the large tech companies are spending $20 billion to $50 billion on infrastructure related to AI and compute capacity every year, he added.

Energy transition

As many developed economies switch to renewable energy at a faster pace, concerns are rising regarding whether clean energy sources can adequately meet the growing power demands of data centres.

A surge in power usage from AI data centres could significantly boost natural gas demand in the second half of the decade, according to Tudor Pickering Holt & Co.

An additional 8.5 billion cubic feet per day of natural gas might be needed to keep pace with the increasing demand, the investment bank said in a report last week.

“We talk about AI [and] we talk about all of these huge requirements for energy and I ask how are we going to deliver the hundreds and hundreds of gigawatts that would be required for these technologies,” Saudi Arabia’s Energy Minister Prince Abdulaziz bin Salman said on Sunday.

The rise of AI technologies is happening amid debates about the continued use of fossil fuels, which contribute to over 75 per cent of global greenhouse gas emissions.

Producers have argued that a rapid transition away from fossil fuels could lead to energy insecurity and economic disruptions. Coal, crude oil and natural gas are expected to be around for longer in developing economies.

“We need all sources of energy,” said Amin Nasser, the chief executive of Saudi Aramco.

“We need to understand very clearly that policy without technological readiness and without economic competitiveness is not going to drive a solution. Incentives will drive a solution but it's not sustainable,” Mr Nasser said in a separate panel session.