AI’s Hidden Climate Cost: Data Centers Fuel Fossil Fuel Dependence at COP30
The escalating environmental impact of artificial intelligence data centers is emerging as a key point of contention at the U.N. climate summit, COP30, currently underway in Belém, Brazil. While proponents tout AI’s potential to address the climate crisis, climate groups are raising alarms about its growing reliance on fossil fuels and the lack of discussion surrounding its carbon footprint in international negotiations.
The debate centers on a fundamental paradox: can a technology marketed as a climate solution simultaneously exacerbate the problem? As one advocate explained, AI is currently being “touted as some type of solution to the climate emergency,” with numerous panels at COP30 highlighting its potential for grid balancing. However, a “deep concern” remains regarding its “severe climate consequences,” a topic largely absent from official discussions.
The United States is at the epicenter of this issue, responsible for 45% of global carbon emissions from AI. This stems from the nation’s position as “ground zero for data center development,” hosting approximately 90% of all computing power. The core problem isn’t the technology itself, but how it’s powered. Currently, AI’s immense computing demands are being met primarily by fracked gas and, increasingly, coal.
This reliance is not accidental. A senior official noted that the Trump administration actively encouraged data center expansion through executive orders streamlining permitting processes, effectively “rubber stamping” connections to readily available, but polluting, energy sources. This has led to a surge in demand for power, with a single data center potentially consuming as much electricity as 100,000 homes.
The impact is being felt acutely at the local level. Communities across the U.S. are mobilizing against the construction of new data centers, citing concerns over energy consumption and water usage. Representatives from the Ponca Nation in Oklahoma, for example, are actively fighting a Google data center in Stillwater, joining similar efforts in Memphis, Tennessee. These communities are “completely caught flat-footed” by the scale of these projects and the associated environmental costs.
While renewables can power data centers, and are being advocated for within the UNFCCC process, this is not the current reality. Utilities, driven by profit margins, are incentivized to build out fossil fuel infrastructure, as it offers a more lucrative return on investment.
However, the picture isn’t entirely bleak. A recent study by the UNFCCC Technology Mechanism acknowledges AI’s potential benefits, particularly for developing nations vulnerable to climate change, through improved early warning systems. One analyst noted that AI has also significantly advanced the field of medicine. But, as one expert cautioned, these positive applications are often overshadowed by AI’s growing use in “defense and militarization,” areas that do not necessarily serve the public interest.
Currently, the COP30 text does not directly address AI’s environmental impact. A coalition of global civil society organizations is working to change that, aiming to present a clear “ask” to countries at next year’s summit: mandatory disclosure of data center emissions and a commitment to powering them with sustainable, renewable energy sources, rather than relying on unproven technologies like carbon capture and storage (CCS).
The absence of a U.S. delegation, led by President Trump, at COP30 is particularly significant. According to one source, Trump views the summit as an opportunity to “peddle his wares” – namely, natural gas, fracked gas, and coal – prioritizing profit over climate action. However, this absence also presents an opportunity to advance critical policies, such as a fossil fuel phaseout roadmap, that the U.S. could be bound by upon its eventual return.
Ultimately, the future of AI and its role in addressing the climate crisis hinges on a fundamental shift in priorities. As one advocate stated, we must empower communities and countries, especially in the Global South, to carefully weigh the potential benefits of AI against its significant environmental costs. The challenge lies in harnessing AI’s potential for good while mitigating its hidden climate footprint.
