AI-driven data centre growth threatens to lock in high-carbon energy at COP30

The exponential rise in AI infrastructure is accelerating global electricity demand, risking increased reliance on fossil fuels and complicating climate commitments as data centre energy consumption could more than double by 2030, prompting calls for stronger regulation and renewable integration.

The rapid rise of artificial intelligence (AI), especially the development and deployment of generative AI models like ChatGPT, is emerging as a central topic at COP30 climate negotiations due to its profound impact on global electricity demand and environmental sustainability. AI is now recognised as a primary driver behind the exponential growth in power consumption by data centers, which are the backbone of digital infrastructures worldwide. According to Rystad Energy, global electricity demand from data centers is projected to nearly triple by 2030 and surge to an astounding 2,700 terawatt-hours by 2050, up from approximately 400 TWh in the early 2020s, with the United States and China leading this expansion.

This surge poses a critical challenge for the power sector and climate policy. Data centers operate continuously and require near-perfect reliability, often guaranteed through complex systems involving grid connections, uninterruptible power supplies, and backup diesel or gas generators. The increasing reliance on backup gas turbines is exerting pressure on supply chains and inflating costs, with price inflation of over 300% and lead times extending beyond five years for high-efficiency combined-cycle gas turbines. This economic strain threatens the viability of new thermal base-load power plants, particularly in developing economies striving to expand their power capacity, potentially locking in high carbon emissions for decades to come.

While major technology firms like Google, Amazon and Microsoft are leading efforts to offset their electricity usage by procuring renewable energy through power purchase agreements, true climate progress demands achieving 24/7 clean power supply. The current approach of matching annual consumption with renewables does not address the intermittent nature of solar and wind power. When renewable output dips, data centers still draw power from the grid, which often relies on fossil fuels. Addressing this requires substantial investments in diverse renewable generation capacity, grid flexibility enhancements, and advanced energy storage solutions to ensure every kilowatt-hour consumed is carbon-free on an hourly basis.

Beyond electricity, data centers also consume large quantities of water, predominantly for cooling their heat-intensive IT equipment. This has sparked concerns in water-stressed regions where data centers operate. Some companies have responded by adopting “water positive” strategies, which aim to return more freshwater to communities and ecosystems than they withdraw, often through water-efficient cooling technologies or use of reclaimed wastewater.

The United States exemplifies the looming energy challenges posed by AI-driven data center growth. Recent reports estimate that data centers could increase U.S. electricity demand by around 300 TWh by 2030, on top of other major drivers such as electric vehicle charging. Projections suggest that data centers might consume up to 12% of the nation’s electricity by 2028, with AI infrastructure alone driving demand increases of 25% by 2030 and potentially 78% by 2050. This surge threatens grid stability and risks pushing consumer power prices up by 15% to 40%. Current power sources for U.S. data centers remain heavily reliant on natural gas, which supplies around 40% of their electricity, alongside renewables, nuclear, and coal. Although nuclear power may increase its share, the dependence on fossil fuels remains a critical issue.

Globally, the International Energy Agency forecasts that data center electricity demand could more than double by 2030, potentially exceeding the entire electricity consumption of countries like Japan. The environmental burden includes not only direct electricity use but also the construction and ongoing water demands of energy-intensive data centres. The rise of AI infrastructure presents the risk of reinforcing carbon lock-in, where economic and grid pressures extend the lifespan of high-emission power assets and compound costs for consumers.

Looking ahead, the challenge for COP30 will be to transcend voluntary pledges and consider regulatory mandates and carbon pricing that appropriately reflect the true environmental costs of computation and data center operations. While AI holds transformative potential to accelerate climate solutions, its rapidly increasing footprint demands robust global governance framework to ensure that the digital revolution does not come at the expense of climate goals.

📌 Reference Map:

• ^[1] Oilprice.com – Paragraphs 1, 2, 3, 4, 5, 6, 7
• ^[2] Reuters – Paragraph 8, 9
• ^[3] Pew Research Center – Paragraph 9
• ^[4] Le Monde – Paragraph 1, 10
• ^[5] Le Monde Opinion – Paragraph 8, 9
• ^[6] Axios – Paragraph 8, 10
• ^[7] Time – Paragraph 1, 10

Source: Fuse Wire Services