How AI, Data Centres & Carbon-Free Energy Coalesce at Amazon

The sudden ubiquity of AI is driving unprecedented demand for cloud services and digital infrastructure, placing increasing pressure on electricity networks and raising questions over how future power requirements can be met sustainably.

Amazon's 2025 Sustainability Report outlines how the company is combining renewable energy, nuclear power, advanced cooling technologies and AI innovation to support growing computing needs while reducing emissions.

Alongside expanding its own infrastructure, Amazon says these investments are helping customers lower the environmental impact of their digital operations as demand for AI continues to increase.

Powering data centres more efficiently

As AI adoption continues to accelerate, improving the energy efficiency of data centres has become a critical part of meeting rising electricity demand.

During the fourth quarter of 2025, Amazon added more than 1.2GW of new global data centre capacity while continuing to improve the efficiency of its operations.

Research referenced in the report suggests organisations moving workloads from on-premises infrastructure to Amazon can process data around 4.1 times more efficiently.

The company has also developed custom AI chips, including Trainium3, which deliver more than five times more AI output per MW of electricity than previous generations while maintaining comparable performance.

“We've encountered tremendous change in each of the seven years since setting The Climate Pledge in 2019,” says Kara Hurst, Amazon's Chief Sustainability Officer.

“Perhaps none bigger than AI, which is both transforming what's possible, accelerating discovery, optimising systems and unlocking solutions that weren't within reach before, yet also creating new demands for energy, water and infrastructure.

“While the speed and scale of AI adoption is unique, and the change is happening faster and more broadly than anything else we’ve encountered in our lifetimes, the need to stay stubborn on our vision and flexible on the details is familiar territory.

“I remain confident and optimistic in the overarching vision and the long-term progress we continue to make toward it.”

Amazon says advanced liquid-to-chip cooling technology can cut mechanical cooling energy consumption by up to 50% during periods of peak demand without increasing water consumption.

The company has also deployed AI-powered monitoring systems across 820 buildings to optimise HVAC performance, identify faults and improve operational efficiency.

Together, these technologies are designed to reduce the electricity required to operate increasingly powerful AI infrastructure while supporting continued expansion.

A focus on carbon-free electricity

Providing reliable electricity for AI at scale requires substantial investment in clean generation, making carbon-free energy a central part of Amazon's long-term strategy.

By January 2026, the company had enabled more than 712 carbon-free energy projects across 30 countries, including 80 announced during 2025.

Together, the portfolio represents around 42GW of clean electricity capacity, enough to supply approximately 13 million US homes each year while avoiding an estimated 48 million tonnes of CO₂ emissions once all projects are operational.

For the third consecutive year, Amazon matched 100% of the electricity consumed across its operations with renewable energy sources.

The company's energy portfolio now includes 375 utility-scale renewable energy developments comprising 264 solar farms, 109 wind farms and two nuclear energy projects, alongside increasing investment in battery storage and small modular nuclear reactors.

“We were investing heavily in clean energy for many years before it became a dinner table conversation,” says Kara.

“This early commitment has led us to become one of the largest corporate purchasers of renewable energy for six years in a row, one of the most energy efficient data centre operators in the world and an early investor in new technologies like small modular reactors (SMRs).”

Amazon says it is also investing in electricity grid infrastructure to improve local power systems while ensuring AI and cloud operations have access to dependable carbon-free electricity.

The company's approach demonstrates how renewable energy, nuclear generation, battery storage and emerging technologies can work together to strengthen energy security while supporting the continued growth of AI.

AI and the energy transition

While AI is increasing electricity demand, Amazon also sees the technology as an important tool for improving energy efficiency and accelerating wider sustainability goals.

Across its operations, Amazon uses AI to improve efficiency, monitor utilities, reduce waste and optimise the management of energy and other resources.

During 2025, AI-powered systems helped detect water leaks, identify faults in HVAC equipment and improve operational performance across hundreds of buildings.

Beyond data centres, AI supports automated waste sorting capable of identifying more than 110 categories of waste in real time and powers robotic systems that dismantle retired servers for reuse and recycling.

National laboratories are using Amazon AI technology to support research into next-generation nuclear energy, while the company invested US$100m through its Education Equity Initiative to expand access to AI education and technical skills globally.

Amazon also invested US$1m in 43 projects focused on developing AI solutions that contribute to the United Nations Sustainable Development Goals.

Alongside these initiatives, the company continues to strengthen responsible AI governance through comprehensive safety frameworks, transparency measures and policies designed to ensure AI is deployed ethically and sustainably.

Improving water and resource efficiency

As AI infrastructure grows, reducing water use and extending the lifespan of equipment are becoming increasingly important for improving the sustainability of energy-intensive data centres.

Amazon has committed to becoming water positive by 2030 and had achieved 75% of that goal by the end of 2025.

The company reduced its global Water Use Effectiveness (WUE) to 0.12 litres per kWh, a 20% improvement compared with 2024 and a 52% improvement since 2021.

Cooling technologies including in-row heat exchangers and configurable liquid-to-chip systems are helping lower both water and energy consumption while supporting increasingly demanding AI workloads.

Amazon is also applying circular economy principles by extending server lifespans, increasing equipment reuse and using AI-powered robotics to recover valuable materials from retired hardware.

Since 2020, the reuse and resale of data centre equipment has prevented approximately 225,000 tonnes of CO₂ equivalent emissions, while programmes to extend hardware life have avoided the purchase of more than one million new hard drives since 2023.

Collectively, these initiatives demonstrate how cleaner electricity, efficient infrastructure and better resource management are helping support the future growth of AI while reducing the environmental footprint of digital infrastructure.