AI's Energy Consumption: Is the Tech Boom Undoing Climate Progress?

Artificial intelligence is booming, but it's sparking a debate about its environmental cost. The immense energy needed to power AI data centers might reverse recent climate gains.

The Climate Change Conference Focus

The annual UN climate conferences aim to cut hydrocarbon use for electricity. Power plants using coal or natural gas account for about 45% of carbon emissions, making them a key target.

Shift to Renewables

The push to reduce these emissions has seen some success, especially in the Western world. In the European Union, coal emissions have dropped significantly as wind and solar energy have become more popular.

AI's Growing Energy Needs

Still, the rise of AI is changing the game. Tech companies like Microsoft, Amazon, and Google need massive amounts of power to scale up their AI operations, leading to huge investments and a surge in energy demand.

Quick Fixes and Old Habits

To meet these demands quickly, many are turning to gas-fired power plants and extending the life of older coal plants. Amazon's Jeff Bezos even suggested building power plants in space to harness solar energy.

The Original Green Promise of AI

Initially, AI was promoted as an environmentally friendly technology. It promised smarter electricity grids, better climate modeling, streamlined factories, and reduced traffic. All of which, theoretically, holds true.

The Hidden Costs

Here’s the catch: powering this AI "revolution" requires enormous amounts of electricity and water for cooling. The Financial Times reports that AI investments are driving a significant portion of U.S. economic growth and stock market gains this year.

Reversing Climate Gains?

Given AI's current and projected energy needs, there are concerns it could undo climate progress. A report by the Lawrence Berkeley National Laboratory found that U.S. data centers consumed 4.4% of the nation's electricity in 2023, up from 1.9% in 2018.

Future Consumption

By 2028, this could rise to between 6.7% and 12%. That's potentially equivalent to the average annual electricity production of Russia or Japan.

The Hunt for Power Sources

The race is on to find new power sources for data centers. For example, the decommissioned Homer City power plant in Pennsylvania is being revived as a massive gas-fired plant.

Gas Demand Concerns

Global Energy Monitor warns that the U.S. has the second-largest number of gas-fired plants in development, driven by AI's energy demands. A gas-demand bubble could leave the U.S. with billions in stranded power assets.

Nuclear Partnerships

AI's energy needs are so great that some companies are partnering with nuclear power generators. Data centers will use renewable energy when possible, but it’s not enough to meet their constant demands.

The Problem with Renewables

That said, renewable sources are intermittent, while data centers need power around the clock. This makes gas plants the preferred option.

Greenhouse Gas Emissions

The International Energy Agency expects greenhouse gas emissions from data centers to double in the next five years as a proportion of emissions from burning fossil fuels.

The True Cost of AI?

Maybe the motto of the tech industry should be, “burn, baby, burn.”