AI's Energy Consumption Threatens Climate Change Mitigation Efforts

Artificial intelligence (AI) is consuming vast amounts of energy, straining the US power grid and posing a significant challenge to national efforts to combat climate change, according to expert forecasts.

The demand for energy is surging, fueled in part by the rapid expansion of AI data centers. 

This increase in energy use coincides with a slowdown in renewable energy development and extended operations of coal-fired power plants, causing analysts to rethink their models for reducing greenhouse gas emissions.

AI's energy consumption was a key topic of discussion at New York Climate Week, held alongside the United Nations General Assembly.

Technology companies, rather than traditional fossil fuel companies, took center stage in the discussions, highlighting the disruptive impact of AI on energy demand.

A recent report by Bloomberg NEF warned of slow progress in the US efforts to decarbonize. It predicts that US emissions will decrease by just 34% by 2030 compared to 2005 levels, far short of the country's national goal of a 50-52% reduction by 2030 and net-zero emissions by 2050, as per the Paris Agreement commitments.

Tara Narayan, a lead energy analyst at Bloomberg NEF, expressed concern about AI’s rising energy demands, calling it a "major disruption." 

She noted that the lack of adequate power grid infrastructure is hindering the transition to renewable energy not only in the U.S. but also globally.

China is set to invest $800 billion over the next six years to manage pressures on its energy system as it rapidly transitions from coal to renewable sources.

Meanwhile, in the US, energy demand, which has remained stable for the past two decades, is expected to increase by 9% by 2028 and by 20% by 2033, according to consulting group ICF. This growth is largely attributed to the expansion of AI data centers, reshoring of manufacturing, and increased electrification.

The Electric Power Research Institute predicts that AI data centers will double their share of US electricity consumption over the next decade. Despite this, US Energy Secretary Jennifer Granholm remains optimistic that the US can meet its carbon neutrality goals, thanks to $370 billion in green subsidies introduced under President Biden’s Inflation Reduction Act.

However, renewable energy developers face significant delays in meeting rising energy demands. Building new renewable energy capacity can take up to five years due to permitting issues and grid connection delays.

The proliferation of AI data centers has prompted a race among major tech companies to secure low-emission energy sources. Recently, Constellation Energy and Microsoft signed a 20-year deal to reopen the nuclear power plant at Three Mile Island in Pennsylvania, the site of the country’s worst nuclear accident. Rising electricity demand has also led U.S. operators to delay the closure of coal-fired power plants.

S&P Global Commodity Insights revised its forecast for coal plant closures, predicting a 40% reduction in the number of closures by the end of the decade. Despite increased renewable energy use, experts like Akshat Kasliwal, an energy analyst at PA Consulting, warn that the U.S. is far from achieving a carbon-free power grid by 2035.

Although the US has a substantial renewable energy capacity, with 1.5 terawatts waiting for grid connections, projects built last year took five years to connect to the grid. Additionally, a lack of transmission lines makes it difficult to transport green energy from remote generation sites to demand centers.

A study by Rhodium Group found that if data center demand nearly triples by 2035 and developers struggle to deploy new wind and solar power, energy sector emissions could be 56% higher than expected under moderate projections.

However, AI could also play a role in reducing energy consumption. As data centers become more efficient and AI optimizes daily processes, tech industry executives argue that AI could eventually help mitigate its own environmental impact.