Why the climate promises of AI sound a lot like carbon offsets

illuminem summarises for you the essential news of the day. Read the full piece on MIT Technology Review or enjoy below:

🗞️ Driving the news: A new report from the International Energy Agency (IEA) finds that artificial intelligence (AI) could substantially reduce global greenhouse gas emissions by 2035—potentially offsetting the energy use of AI data centers
• However, the IEA cautions that these future benefits remain speculative, while emissions from today’s booming data infrastructure—largely powered by fossil fuels—are already rising rapidly

🔭 The context: Major data centers supporting AI tools such as ChatGPT are driving massive demand for electricity, much of it supplied by natural gas
• While tech leaders have touted AI's potential for climate breakthroughs—from optimizing energy systems to improving battery materials—there are no binding regulations to ensure these tools are deployed for climate benefit
• The narrative closely mirrors that of carbon offsets, which often promise future benefits to justify current emissions but lack robust enforcement or accountability

🌍 Why it matters for the planet: The “AI will fix the climate later” argument risks delaying necessary decarbonisation today
• Future AI applications may reduce emissions, but today’s data centers could lock in decades of fossil fuel use if powered unsustainably
• Without strict standards and investment in clean energy infrastructure, the sector may accelerate climate risks instead of mitigating them
• Delayed action now could outweigh any future emission savings, given the cumulative impact of carbon dioxide in the atmosphere

⏭️ What's next: For AI to contribute meaningfully to climate action, the industry must urgently prioritise clean power sourcing
• This includes investing in renewables, nuclear, geothermal, and grid decarbonisation at the pace and scale that matches AI’s growth
• Policymakers may need to step in with incentives or requirements to align AI’s development with climate goals
• Otherwise, the technology’s climate promise could remain an overhyped offset with little near-term impact

💬 One quote: “There is currently no momentum that could ensure the widespread adoption of these AI applications. Therefore, their aggregate impact, even in 2035, could be marginal if the necessary enabling conditions are not created.” – IEA Report

📈 One stat: The IEA estimates AI could reduce emissions by 1.4 billion tons by 2035, but that outcome depends on highly optimistic assumptions about adoption, deployment, and regulation

See here detailed sustainability performance of companies like OpenAI and Deepseek

Click for more news covering the latest on carbon market