Sustainable by Design: Advancing the Sustainability of Artificial Intelligence

Written by: Melanie Nakagawa – Sustainability Director.

Over the past year, the pace of adoption of artificial intelligence has accelerated dramatically, leading to breakthroughs, discoveries, and innovative solutions with the potential to help address the biggest problems facing humanity. We see this as a massive shift in platforms, similar to the printing press, which was not just an invention, but a technology that shaped a new economy. In addition to the incredible promise and benefits of AI, we recognize the resource intensity of these applications and the need to address environmental impact from all angles.

In line with our commitment to responsible AI and our ambitious commitments to sustainability, we are determined to meet this challenge so that the world can reap the full benefits of AI. There are three areas in which we are deeply engaged and increasing our focus. he Firstly It is to improve the energy and water efficiency of the data center. he second It is the development of low carbon materials, to create global markets to help promote sustainability across industries. And the third It is to improve the energy efficiency of AI and cloud services, empowering our customers and partners with tools for collective progress.

1. Improving energy and water efficiency in the data center

Over the past decade, our quest to innovate in every part of our cloud infrastructure to deliver more sustainable cloud services has led to many changes in the way we design, build and operate our data centers. As we continue this work, two of the biggest challenges we face are energy management and water density.

Energy management

The energy intensity of cloud services and advanced AI has prompted us to accelerate our efforts to enhance efficiency and reduce energy. In addition, we have expanded our support for increasing the availability of renewable energy, both for our own operations and for the communities in which we operate.

To continue driving improvements in data center energy management, we are working to reduce peak power, securely harvest unused energy, and increase server density in existing data centers through intelligent use and allocation of energy-aware virtual machines, and drive efficiency to the chip and code level.

Recognizing the need to power more renewable energy, we currently have more than 135 renewable energy projects in our global Power Purchase Agreements (PPA) portfolio, a powerful mechanism to support the global energy transition. In the way we design, build and manage our data centres, we are focused on the path to 100% carbon-neutral electricity all the time.

We are also working on solutions that allow data centers to return power to the grid to contribute to local energy supply during times of high demand. For example, in Ireland, we built batteries into wind turbines for a wind power project to capture energy when the turbines are over-performing and deliver that energy to the local grid. In Denmark, excess heat generated by a Microsoft data center will provide heat to the local community, producing enough heat to heat about 6,000 local homes. Both are examples of our work to use our data centers as a source of electricity to relieve pressure on local power grids.

Water density

Today, many data centers rely on water for two reasons: directly for cooling and indirectly for generating electricity. Although on a global scale, data centers’ total water consumption is relatively small, weighing in at about 0.1% of national water use in the United States.1 We recognize the impact of data center operations in water-stressed areas, and we work to minimize that impact and design solutions that advance our progress on the path to positive water.

We take a holistic approach to water reduction across all of our work, from design to efficiency, by seeking immediate opportunities through operational use and, in the longer term, through design innovation to reduce, recycle and reuse water. We have successfully used direct air instead of water to cool data centers, harvest rainwater, and obtain reclaimed water from utilities to reduce our reliance on fresh water. For example, in our data centers in Sweden, we will use a process called free cooling, which is a simple and cost-effective method that reduces energy costs by 30% and uses 90% less water compared to standard systems.

2. Promotion of low carbon materials

For our future data centers and to help drive progress across the industry, another way we can move forward is to help accelerate markets for low carbon building materials. As a sector, building materials, such as steel and cement, are currently the largest contributors to the carbon cost of new construction, together producing around 13.5% of global carbon emissions.2

Innovations in green steel3 Low-carbon cement is emerging rapidly, but these markets are still emerging and require significant investment to expand supply.

Through our $1 billion Climate Innovation Fund, we are investing in accelerating the development and deployment of new climate innovations, especially for underfinanced sectors and markets in short supply, such as low-carbon building materials. For example, we are investing in solutions such as H2 green steel to expand the market offering of near-zero carbon steel.4, which can provide up to 95% less CO2 emissions compared to traditional steel. We are also evaluating the use of near-zero carbon steel in our supply chains for construction materials and equipment.

Likewise, we are expanding the availability of concrete and other low-carbon building materials through commercial projects and collaborations with the world’s largest data center companies. In Washington State, our pilot program uses concrete alternatives such as biolimestone, fly ash and slag with the goal of reducing embodied carbon in concrete by more than 50% compared to traditional concrete mixes. Through these investments, our goal is to facilitate the commercialization of materials that can have a significant impact on carbon reduction, for our own construction and for the industry in general.

3. Improving energy efficiency in artificial intelligence and cloud services

Reducing the energy needed to run AI and cloud services up front is another important element of the solution. We support developers and IT professionals with tools to improve models and code, explore ways to reduce the power requirements of AI, and harness the power of these advanced technologies to drive energy progress.

As a founding member of the Green Software Foundation, we collaborate with other industry-leading organizations to help grow the field of green software engineering, contribute to industry standards, and work together to reduce carbon emissions from software. Across all of our cloud services, we work to ensure IT professionals have the information they need to better understand and reduce the carbon emissions associated with their cloud use.

As AI scenarios become more complex, we enable developers to create and improve AI models that can achieve similar results and require fewer resources. In recent months, we have released a set of small language models (SLMs) called “Phi” that perform impressively on a variety of benchmarks, matching or outperforming larger models by up to 25 times. Now available in the Azure AI Studio Model Catalog, Phi-2 offers a compact model for R&D or fine-tuning experiments on a variety of tasks.

We have learned that the complex sustainability challenges we face today can best be addressed through multidisciplinary and multisectoral collaboration, and progress in energy is no exception. We recently collaborated with the Department of Energy’s Pacific Northwest National Laboratory to use advanced AI models to find new materials that can reduce reliance on traditional battery materials, such as lithium. The team screened more than 32 million materials, discovered 500,000 stable candidates, and assembled a promising candidate into a working prototype, shortening a process that could take years to a few days.

These highlights provide insight into our work to build and operate cloud services more sustainably, to develop solutions that can reduce the future impact of AI. Our ambitious 2030 goals of becoming carbon negative, water positive, zero waste and protecting biodiversity require continuous innovation in all aspects of our operations, and we are committed to sharing what we learn along the way. Stay tuned for more information on this topic in the coming months.

Learn more:

For more information, visit the Microsoft Sustainability website and read the white paper Accelerating Sustainability with AI. To learn how to integrate AI into our sustainability solutions, watch the This is AI digital event. . . For sustainability.

Sources in footnotes:

1Water consumption in the data center | npj clean water (nature.com)

2Cement and steel: nine steps towards net zero (nature.com)

3What is green steel and how can it help us reach net zero? | World Economic Forum

4Iron and Steel – AIE

Check Also

These are the wheels that sink and produce 12 kilowatt hours per day

An invention arrives that will become rivers Power plants. They are sunken wheels and produce …

Leave a Reply

Your email address will not be published. Required fields are marked *