On Wall Street, frustration is mounting that AI will hijack the climate debate

Some prominent bankers are asking the global finance industry not to let artificial intelligence overshadow talks about reducing planet-warming pollution.

Aniket Shah, the executive at the Jefferies investment bank who heads up strategy on sustainability and the energy transition, is among them, and his comments came as an estimated 100,000 began to arrive in New York for the city’s Climate Week.

“Climate change is about 50 billion tons of greenhouse gases going to zero, and so we actually have to have a right framing on what is driving that,” Shah said. “Data center demand growth is actually not in the top five.”

The concern for Shah and those who share his point of view is that angst about the greenhouse gas emissions of AI data centers will drag attention away from areas with substantially higher emissions, such as transport and heavy industry.

It’s a distinction that others are also starting to make. A recent report by the investment firm of former vice president turned climate campaigner Al Gore notes that electricity for air conditioning is a far larger cause of climate change than energy consumption for AI.

It’s “understandable” that AI data centers would be “a huge concern” for people worried about climate change, Gore said in an interview. “I just don’t think it’s a cause for panic.”

The energy use and emissions of AI data centers are dwarfed by numerous other activities, he said.

“Just to pick one example, the emissions from uncovered landfills around the world are a large multiple of the emissions from all of the AI data centers put together.”

If talks at New York Climate Week turned out “to be all about data centers” and “all about AI,” that would be an unfortunate distraction, he said.

“If you’re actually thinking about climate change globally speaking, data centers are just not that big of a deal, just like arithmetically, it’s not,” Shah said.

A definitive assessment of the greenhouse gas emissions of AI has yet to be produced. In April, the International Energy Agency said that taking into account an expected doubling of electricity consumption by the end of the decade, data centers will account for just 3% of global demand by 2030. All in all, emissions associated with data centers — not just the ones feeding AI — are on track to make up about 2% of those released by the global electricity sector by 2035, the IEA also said.

In a report published last week, Gore’s Generation Investment Management noted that in most countries data centers are “not the most important” drivers of energy demand. The biggest contributor is a rise in air conditioning, with electric cars and heat pumps — both important planks of the green energy transition — also identified as major guzzlers of electricity.

Such assessments clash with the popular perception of AI data centers, which are often criticized for their use of energy, water and land. Across the U.S., local communities have mobilized to protest against new mega data centers in their backyards, with concerns often centered on spikes in utility costs.

Investors have also voiced some concerns. A recent survey of asset owners by Morningstar Inc. showed that 25% cited the environmental impact of AI use as a risk, up from 12% a year earlier. Six in 10 said they’re worried that AI-driven demand for data centers and power generation will result in higher energy costs and inflation.

“Potentially the most significant impact” that the companies building AI infrastructure “will have on the climate is through the way that their AI products are used,” said Thomas Day, a climate policy expert with the New Climate Institute.

“We need a conversation about the extent to which AI companies should be responsible for who they provide their services to, and what they are used for,” he said.

Celine Herweijer, visiting professor of energy, AI and geopolitics at the London School of Economics and former chief sustainability officer at HSBC Holdings, says electricity is becoming “the binding constraint.”

And “that’s why AI matters so much,” because “the power system we build for AI will affect everything else we plug into it,” she said.

The boom in AI digital infrastructure is expected to strain power systems and spur the build-out of capacity, BloombergNEF said in a report last week. BNEF also said the tech industry’s “hunger for cheap electrons” has improved the outlook for clean power. But gas is the fuel getting the “biggest boost” from AI power demand, BNEF said.

At the same time, the IEA says, the extra demand from AI is feeding into grid bottlenecks that urgently need to be addressed. And that’s a topic that Wall Street’s biggest bank has made clear it plans to prioritize.

JPMorgan Chase & Co. expects to be talking about the grid “every day of New York Climate Week and every chance that we get,” said Heather Zichal, the bank’s global head of sustainability. “We look at modernizing the grid as the single biggest unlock for everything else: It helps national security, it helps energy affordability, and it helps speed to power.”

The bank, which last year launched a $1.5-trillion security and resiliency initiative designed in large part to secure the future of American infrastructure, is positioning itself for what Zichal said are “significant investment opportunities in capital market activity in a range of new technologies from carbon removal to nuclear to energy storage, geothermal, etc.”

Marsh writes for Bloomberg.