Sustainability

AWS Cloud

Amazon Web Services (AWS) is the world’s most comprehensive and broadly adopted cloud offering, with millions of global users depending on it every day. To build a sustainable business for our customers and for the world we all share, we’re designing data centers that provide the efficient, resilient service our customers expect while minimizing our environmental footprint—and theirs.

Aerial view of a large data center at sunset. The building is surrounded by a paved parking lot. The sun sets casting warm colors in the sky.

Progress

4.1x AWS infrastructure is up to 4.1 times more energy efficient than on-premises and can reduce workloads’ carbon footprint by up to 99%

A person walks down an aisle in a data center.

7B+ liters of water will be returned each year to local communities through AWS’s water replenishment efforts

An irrigation system mists water on rows of leafy green crops.

100% of electricity consumed by Amazon was matched with renewable energy sources in 2023

Our approach

Our sustainability work includes enhancing energy efficiency, transitioning to carbon-free energy, reducing embodied carbon, using water responsibly, driving a circular economy, and enabling sustainability for customers.

Increasing efficiency

We focus on efficiency across all aspects of our infrastructure, from the design of our data centers and hardware, to modeling the performance of our operations for continually enhanced efficiency. By continually improving our efficiency, we can reduce the amount of energy needed to operate our data centers.
An aerial view of an Amazon facility with lights on at night.
A person wearing headphones and an AWS sweatshirt examines items in a data center.
A person examines large pipes in a utility room.
A computer chip sits on a dark table.

Power usage effectiveness

AWS has provided cloud services to customers for nearly 18 years, and continues to innovate to increase efficiency with each generation of our data center designs. As AWS builds new data centers, we seek the optimal balance between energy and water use. The new data center components in AWS data center design are projected to reduce mechanical energy consumption by up to 46%, reduce embodied carbon in the concrete used by 35%, and deliver a power usage effectiveness (PUE) rating to 1.08. PUE is one way we measure the efficiency of our data center operations. A lower PUE indicates a more efficient data center and a PUE score of 1.0 is perfect. AWS calculates PUE using the internationally-recognized principles of the International Organization for Standardization in line with our PUE Methodology.


In 2023, AWS data centers had a global PUE of 1.15, and our best performing site was in Europe with a PUE of 1.04. In the Americas, our best performing site has a PUE of 1.05, and in Asia Pacific it is 1.08. These ratings are below the IDC estimated PUE of 1.84 in 2023 for on-premises data centers.

All regions PUE

The following data includes average PUE ratings between January 1 and December 31 of each year. Data is not available (—) where AWS operated data centers less than 12 months.

Download CSV


AWS Geography PUE and AWS Region PUE

While year-over-year reported PUE has improved or remained stable in the vast majority of our Regions, a few Regions saw slightly higher PUE values in 2023. The primary reasons for this include different weather patterns that require different amounts of cooling, and the launch of new, not-yet fully loaded data centers within a Region. These result in higher average PUEs for the specific Region while data center utilization increases.

AWS Geography PUE

Geography PUE 2022 PUE 2023
Global 1.15 1.15
AMER 1.15 1.14
EMEA 1.12 1.12
APAC 1.27 1.28


AWS Region PUE

AWS Region PUE 2022 PUE 2023
Europe (Stockholm) 1.12 1.12
Middle East (Bahrain) 1.38 1.32
Asia-Pacific (Mumbai) 1.43 1.44
Asia-Pacific (Jakarta) 1.39 1.35
U.S. East (Ohio) 1.12 1.12
Africa (Cape Town) 1.36 1.24
Europe (Ireland) 1.10 1.10
Middle East (UAE) -- 1.36
Europe (Frankfurt) 1.32 1.33
South America (Sao Paulo) -- 1.18
Asia-Pacific (Hyderabad) -- 1.50
U.S. East (Northern Virginia) 1.16 1.15
Asia-Pacific (Melbourne) -- 1.08
Asia-Pacific (Tokyo) 1.30 1.32
U.S. West (Oregon) 1.13 1.13
U.S. West (Northern California) 1.17 1.17
Asia-Pacific (Singapore) 1.33 1.30
Asia-Pacific (Sydney) 1.15 1.15
Canada (Central) 1.26 1.22
Europe (Spain) -- 1.11
China (Ningxia) 1.28 1.26

Efficiency of scale

Our scale allows us to achieve higher resource utilization and energy efficiency than the typical on-premises data center. A study released by Accenture and AWS estimates running optimized workloads on AWS’s infrastructure is up to 4.1 times more efficient than on-premises and can reduce workloads’ carbon footprint by up to 99%.
Download How Moving onto The AWS Cloud Reduces Carbon Emissions

Predicting performance

We use advanced modeling methods, such as computational fluid dynamics tools, to optimize our data center design. This allows us to understand how a data center will perform before it’s built, enabling us to optimize for higher reliability and energy efficiency in our systems. Once our data centers are operational, real-time, physics-based models allow us to further improve and optimize our designs. We build these custom models using AWS services and weather datasets from the Amazon Sustainability Data Initiative to predict system performance for our sites and track their performance against how they should be operating.

Cooling efficiency

Cooling is one of the largest sources of energy usage in our data centers and we’re continually innovating in cooling efficiency. We use different cooling techniques depending on the time of year and use real-time sensor data to adapt to changing weather conditions. We’re also working to optimize the longevity and airflow performance of the cooling equipment used in our data center cooling systems.

Power efficiency

One of the most visible ways we’re using innovation to improve power efficiency is our investment in AWS chips.

Power efficiency

One of the most visible ways we’re using innovation to improve power efficiency is our investment in AWS chips.

Graviton4

This is the latest generation of chips designed by AWS and the most powerful and energy-efficient chip AWS has built as of 2023. Graviton4 provides up to 30% better computing performance, 50% more cores, and 75% more memory bandwidth than Graviton3 processors while using the same amount of energy.

Inferentia2

In 2023, AWS launched Inferentia2, the second generation of its Inferentia chip, developed to deliver the highest performance at the lowest cost per watt. Inferentia2 is up to 50% more energy-efficient and can reduce costs by up to 40% against comparable Amazon Elastic Compute Cloud instances.

See how Amazon Web Services is enabling sustainability solutions.

Visit AWS
A close-up view from the side of a wind turbine.
A person wearing safety equipment stands in front of solar panels.

Carbon-free energy

We're proud that in 2023, we've achieved our goal to match 100% of the electricity consumed across our operations with renewable energy—a commitment we've met seven years ahead of our original 2030 goal.
Explore carbon-free energy Explore carbon-free energy
Rows of solar panels in a desert climate.

We procure renewable power from utility-scale wind and solar projects that add new sources of carbon-free energy to the grid. These new renewable energy projects support hundreds of jobs while providing hundreds of millions of dollars of investment in local communities. We may support these grids through the purchase of environmental attributes, like Renewable Energy Certificates and Guarantees of Origin, in line with our Renewable Energy Methodology.

 

Amazon’s energy supply from utilities, combined with the renewable energy we procure across the U.S., means that 100% of the electricity consumed by 22 AWS data center regions is matched by renewable energy sources.

Reducing embodied carbon

We’re focused on reducing the indirect emissions associated with building AWS data centers and the manufacturing of our hardware.

Lower-carbon concrete and steel

We’re working to reduce the embodied carbon of materials like concrete and steel, which are used to build our data centers. Embodied carbon is emitted during the extraction, manufacturing, and transportation of materials to construction sites. In 2023, AWS built 36 data centers with lower-carbon concrete and 31 data centers with lower-carbon steel.
Concrete and steel beam infrastructure set with blue sky in the background.

Prioritizing a circular economy

AWS embraces circular economy principles for its server racks by designing reusable and lower-carbon rack systems from the outset. In addition, AWS works to keep equipment operating efficiently and to recover value from securely decommissioned equipment through reuse, repair, and recycling. By working to maximize resource value for as long as possible, AWS reduces waste generation from its global operations, decreases the use of raw materials, and reduces carbon emissions across its supply chain. As a result, in 2023, 14.6 million hardware components were diverted from landfills by being recycled or sold into the secondary market for reuse.

Design better

We concentrate on avoiding excess material such as steel or plastic, increasing recycled and bio-based content, and planning for repair, reuse, and recycling from the beginning.        

Operate longer

When AWS uses equipment for as long as operationally efficient, it reduces the carbon footprint associated with manufacturing and using new hardware and avoids unnecessary waste.

Recover more

AWS sends retired infrastructure hardware from around the world to its regional reverse logistics hubs. These hubs help AWS consolidate, assess, repair, and recirculate functional equipment back into its inventory or to third parties to be sold for reuse.

Committed to being water positive

At Amazon, we know that water is a precious resource. AWS is committed to being water positive by 2030 and making more water available to the communities where we operate.
Explore water stewardship

Popular downloads

Discover the latest in AWS

1 3

% buffered 00:00
00:00
Loading
Article author image

Transcript