Bill Gates has expressed hope that artificial intelligence (AI) will significantly add in achieving climate goals, despite concerns about the environmental impact of new data centers. Speaking to journalists, Gates emphasized that AI could help reduce energy consumption by improving the efficiency of technology and electricity grids, even as the need for data centers increases. He downplayed fears that AI would create climate issues, noting that data centers might only add 2% to 2.5% to energy demand. Gates believes that AI will play an important role in reduction energy use that will outplay the additional demand from data centers.
Concerns about the energy demands of AI are not new. A Goldman Sachs estimate suggests that AI queries, such as those processed by ChatGPT, require nearly ten times the electricity of a Google search, potentially doubling data center carbon emissions between 2022 and 2030. Some experts predict that the rise of AI data centers could increase electricity demand by up to 10% in developed countries. However, Gates countered these concerns at a London conference hosted by Breakthrough Energy, his venture fund. He argued that the additional energy demand from AI data centers would likely be offset by new investments in green electricity, driven by tech companies' willingness to pay a premium for clean energy.
Breakthrough Energy has invested in over 100 companies focused on the energy transition, and Gates himself is a significant investor in AI through the Gates Foundation Trust, which allocates about a third of its $77 billion wealth to Microsoft. Microsoft, in turn, is the largest external investor in OpenAI, the creator of ChatGPT, and has integrated AI tools into its Windows operating system under the Copilot brand. Gates's confidence in AI's potential to reduce carbon emissions is shared by others; a Nature Scientific Reports paper found that generative AI uses significantly less CO2 for simple tasks compared to humans.
Despite AI's potential to reduce emissions, the overall carbon impact of AI remains complex. Google, for instance, utilized AI from DeepMind to reduce its data center cooling costs by 40%, resulting in a 15% reduction in electricity use for non-IT tasks. However, Microsoft’s emissions report indicates that indirect emissions from building new data centers are increasing, and these facilities cannot yet be constructed using solely renewable energy. Furthermore, the rise of "on-device" AI, such as Microsoft's Copilot+ PCs and Apple's Siri enhancements, complicates the picture as these new devices consume more power.
Gates said that achieving net-zero emissions goals by 2050 may be unrealistic due to the slow pace of transitioning to greener sources of energy. He projected that the world may miss its climate targets by 15 years as things are not moving quickly enough. This delay could hinder the decarbonization of high-emission sectors like heavy industry, making it challenging to meet the 2050 net-zero target.
This warning follows a global report revealing that, despite a record increase in renewable energy in 2023, fossil fuel consumption also reached a new high due to rising demand. Gates's remarks highlight the critical balance needed between advancing AI technologies and accelerating the transition to green energy to meet climate goals effectively.