The post 3 ways AI is helping the planet appeared first on The Microsoft Cloud Blog.

Beginning in August 2024, Microsoft launched a new datacenter design that optimizes AI workloads and consumes zero water for cooling. By adopting chip-level cooling solutions, we can deliver precise temperature control without water evaporation. While water is still used for administrative purposes like restrooms and kitchens, this design will avoid the need for more than 125 million liters of water per year per datacenter.*

Zero-water evaporation and the quest for ultra-low Water Usage Effectiveness 

These new liquid cooling technologies recycle water through a closed loop. Once the system is filled during construction, it will continually circulate water between the servers and chillers to dissipate heat without requiring a fresh water supply. 

We measure water efficiency through Water Usage Effectiveness (WUE), which divides total annual water consumption for humidification and cooling by the total energy consumption for IT equipment. We are continually investing in improving the design and operation of our datacenters to minimize water use. In our last fiscal year, our datacenters operated with an average WUE of 0.30 L/kWh. This represents a 39% improvement compared to 2021, when we reported a global average of 0.49 L/kWh.  This WUE reduction is due to our ongoing efforts to actively reduce water wastage, expand our operating temperature range, and audit our data center operations. We also expanded our use of alternative water sources, such as reclaimed and recycled water, in Texas, Washington, California, and Singapore. 

We have been working since the early 2000s to reduce water use and improved our WUE by 80% since our first generation of datacenters. As water challenges grow more extreme, we know we have more work to do. The shift to the next generation datacenters is expected to help reduce our WUE to near zero for each datacenter employing zero-water evaporation. As our fleet expands over time, this shift will help reduce Microsoft’s fleetwide WUE even further.

Mitigating energy impacts 

Traditionally, water has been evaporated on-site to reduce the power demand of the cooling systems. Replacement of evaporative systems with mechanical cooling will increase our power usage effectiveness (PUE). However, our latest chip-level cooling solutions will allow us to utilize warmer temperatures for cooling than previous generations of IT hardware, which enables us to mitigate the power use with high efficiency economizing chillers with elevated water temperatures. 

The result is a nominal increase in our annual energy usage compared to our evaporative datacenter designs across the global fleet. Additional innovations to provide more targeted cooling are in development and are expected to continue to reduce power consumption. 

Pilot projects and implementation 

Although our current fleet will still use a mix of air-cooled and water-cooled systems, new projects in Phoenix, Arizona, and Mt. Pleasant, Wisconsin, will pilot zero-water evaporated designs in 2026. Starting August 2024, all new Microsoft datacenter designs began using this next-generation cooling technology, as we work to make zero-water evaporation the primary cooling method across our owned portfolio. These new sites will begin coming online in late 2027. 

Advancing sustainability: Sustainable by design 

Learn more about how Microsoft is advancing the sustainability of cloud and AI through our blog series:  

• Sustainable by design: Advancing the sustainability of AI
• Sustainable by design: Transforming datacenter water efficiency
• Sustainable by design: Innovating for energy efficiency in AI, part 1
• Sustainable by design: Innovating for energy efficiency in AI, part 2
• Sustainable by design: Advancing low carbon materials

*Based on our FY 2024 global average withdrawal WUE of 0.30 L/kWh.

The post Sustainable by design: Next-generation datacenters consume zero water for cooling appeared first on The Microsoft Cloud Blog.

As we work to advance the sustainability of our business, we are also advancing the sustainability of the datacenter infrastructure needed to deliver cloud and AI innovations. At Microsoft, we are working to decarbonize datacenters by focusing on how we design, build, and operate. To support this work, we are also investing to help scale markets for low-carbon building materials.

As a sector, building materials such as steel and concrete are some of the highest contributors to the embodied carbon of new construction, together producing an estimated 13.5% of global carbon emissions.¹ Embodied carbon is a measure of the carbon emitted during the manufacturing, installation, maintenance, and disposal of a product or material.  

Innovations in lower-carbon steel and concrete are emerging around the globe, however, these markets are still nascent and need significant investment to bring the needed supply online. Through our $1 billion Climate Innovation Fund and the collaboration of pioneering teams across datacenter engineering and procurement, we’re investing to accelerate these markets.  

Innovating for energy efficiency

Explore how we're advancing the power and energy efficiency of AI

Read the blog

Novel construction materials and new methods of creating those materials show promise in sectors that are traditionally described as “hard to abate,” these are sectors we believe are necessary to abate. For example, we’re breaking ground on mass timber datacenters, investing to accelerate market availability of near-zero carbon steel, and expanding options for low-carbon concrete in construction. 

Innovating with mass timber datacenter construction to reduce embodied carbon 

In Virginia, we’re building our first datacenters made with superstrong, ultra-lightweight wood with the goal of reducing the embodied carbon of the buildings by 35% compared to conventional steel construction, and 65% compared to typical precast concrete. 

Although this is a novel approach to datacenter construction, it’s a material we’ve used before. In 2021, when we chose cross-laminated timber (CLT) for a new building on our Silicon Valley campus, the approach brought numerous environmental benefits. With ecological design elements ranging from water reuse to clean energy production to new public pathways and restoration of native ecology, the structure earned recognition for sustainable design excellence from the American Institute of Architects.  

The CLT market is well-established in Europe and rapidly growing in the United States, due to demand in the residential segment and adaptability of CLT to new designs. However, our innovative work to apply this material to building a hyperscale datacenter has required everyone to work differently, from our engineers to our procurement teams to the suppliers involved in construction.  

Because CLT is prefabricated offsite, it brings additional benefits such as a faster and safer onsite installation than traditional corrugated steel. Built commonly out of spruce, pine, or fir, CLT shows remarkable structural integrity and resilience even under high temperatures, developing a char and providing insulation in scenarios where steel is likely to fail. But few datacenter building specialists have experience with the material, reducing the availability of skilled contractors, and the materials come at a premium cost in certain regions.  

Throughout this project, our teams have risen to the challenge by sharing best practices across disciplines, crafting new procurement strategies, ensuring skilling pathways, and working collaboratively to validate new material combinations. Expanding the building material options for datacenter construction opens new paths toward achieving climate goals and contributes to expanding the market for sustainable building materials, including markets for regionally sourced materials and contractors working with these materials. 

Accelerating market availability of near-zero carbon steel 

Last year, Microsoft’s Climate Innovation Fund became an investor in Sweden’s Stegra (formerly H2 Green Steel), which is building the world’s first large-scale green steel plant in northern Sweden, achieving up to a 95% reduction in carbon emissions compared to traditional steelmaking.² Another promising investment within our Climate Innovation Fund is Boston Metal, which uses renewable electricity and a unique process that generates oxygen instead of carbon dioxide when making steel.

In addition, Microsoft is a founding member of the Sustainable Steel Buyers Platform of RMI, a first-of-its-kind buyers’ group accelerating steel decarbonization through collaborative procurement and market action. Our engineering and procurement teams are working to incorporate low-emissions steel and repurposed steel in new construction. 

Expanding options for low-carbon concrete for construction  

The bulk of emissions associated with concrete come from cement production. A key ingredient of cement is limestone, which is typically heated with clay to around 2,650 degrees Fahrenheit in a coal or gas-fired kiln where it undergoes a chemical reaction called calcination that releases carbon dioxide as a byproduct. In Washington, our pilot program utilizes cement alternatives like biogenic limestone (grown in place by algae instead of quarried) and fly ash and slag, testing mixes that can lower the embodied carbon in concrete by more than 50% compared to traditional mixes.  

While transitioning to low-carbon concrete production is not as capital intensive as steel manufacturing, the supply chain is fragmented and manufacturing processes can be complex—causing delays and slowing adoption of new techniques. For this reason, we’re looking to expand options for construction across the low-carbon concrete value chain. 

One of the Climate Innovation Fund’s earliest investments is CarbonCure, a company deploying low carbon concrete technologies that inject captured carbon dioxide into concrete, where the CO2 immediately mineralizes and is permanently embedded as nanosized rocks within the physical product. This not only acts as a carbon sink but also strengthens the material, enabling a reduction in the amount of carbon-intensive cement required. Another investment is Prometheus Materials, a company producing zero-carbon bio-concrete through a unique process that combines naturally occurring microalgae with other components.  

Explore the Sustainable by design series

With these investments, we aim to facilitate the commercialization of materials innovations that can make an outsized impact on carbon reduction for our own buildings and for built environments around the world. 

Learn more about our work to advance the sustainability of AI with the Sustainable by design blog series: 

• Sustainable by design: Advancing the sustainability of AI
• Sustainable by design: Transforming datacenter water efficiency
• Sustainable by design: Innovating for energy efficiency in AI, part 1
• Sustainable by design: Innovating for energy efficiency in AI, part 2

¹ Nature Research, Cement and steel—nine steps to net zero. 

² Stegra, Green platforms—green hydrogen, green iron, and green steel. 

The post Sustainable by design: Advancing low carbon materials appeared first on The Microsoft Cloud Blog.

As we continue to deliver on our customer commitments to cloud and AI innovation, we remain resolute in our commitment to advancing sustainability. A critical part of achieving our company goal of becoming carbon negative by 2030 is reimagining our cloud and AI infrastructure with power and energy efficiency at the forefront.

We’re pursuing our carbon negative goal through three primary pillars: carbon reduction, carbon-free electricity, and carbon removal. Within the pillar of carbon reduction, power efficiency and energy efficiency are fundamental to sustainability progress, for our company and for the industry as a whole.

Explore how we're advancing the sustainability of AI

Explore our three areas of focus

Read more

Although the terms “power” and “energy” are generally used interchangeably, power efficiency has to do with managing peaks in power utilization, whereas energy efficiency has to do with reducing the overall amount of power consumed over time.

This distinction becomes important to the specifics of research and application because of the type of efficiency in play. For an example of energy efficiency, you might choose to explore small language models (SLMs) with fewer parameters that can run locally on your phone, using less overall processing power. To drive power efficiency, you might look for ways to improve the utilization of available power by improving predictions of workload requirements.  

From datacenters to servers to silicon and throughout code, algorithms, and models, driving efficiency across a hyperscale cloud and AI infrastructure system comes down to optimizing the efficiency of every part of the system and how the system works as a whole. Many advances in efficiency have come from our research teams over the years, as we seek to explore bold new ideas and contribute to the global research community. In this blog, I’d like to share a few examples of how we’re bringing promising efficiency research out of the lab and into commercial operations.

Silicon-level power telemetry for accurate, real-time utilization data

We’ve made breakthroughs in delivering power telemetry down to the level of the silicon, providing a new level of precision in power management. Power telemetry on the chip uses firmware to help us understand the power profile of a workload while keeping the customer workload and data confidential. This informs the management software that provides an air traffic control service within the datacenter, allocating workloads to the most appropriate servers, processors, and storage resources to optimize efficiency.

Working collaboratively to advance industry standards for AI data formats

Inside the silicon, algorithms are working to solve problems by taking some input data, processing that data through a series of defined steps, and producing a result. Large language models (LLMs) are trained using machine learning algorithms that process vast amounts of data to learn patterns, relationships, and structures in language.

Simplified example from Microsoft Copilot: Imagine teaching a child to write stories. The training algorithms are like the lessons and exercises you give the child. The model architecture is the child’s brain, structured to understand and create stories. Inference algorithms are the child’s thought process when writing a new story, and evaluation algorithms are the grades or feedback you give to improve their writing.¹

One of the ways to optimize algorithms for efficiency is to narrow the precision of floating-point data formats, which are specialized numerical representations used to handle real numbers efficiently. Working with the Open Compute Project, we’ve collaborated with other industry leaders to form the Microscaling Formats (MX) Alliance with the goal of creating and standardizing next-generation 6- and 4-bit data types for AI training and inferencing. 

Narrower formats allow silicon to execute more efficient AI calculations per clock cycle, which accelerates model training and inference times. These models take up less space, which means they require fewer data fetches from memory, and can run with better performance and efficiency. Additionally, using fewer bits transfers less data over the interconnect, which can enhance application performance or cut network costs. 

Driving efficiency of LLM inferencing through phase-splitting

Research also shows promise for novel approaches to large language model (LLM) inference, essentially separating the two phases of LLM inference onto separate machines, each well suited to that specific phase. Given the differences in the phases’ resource needs, some machines can underclock their AI accelerators or even leverage older generation accelerators. Compared to current designs, this technique can deliver 2.35 times more throughput under the same power and cost budgets.²

Learn more and explore resources for AI efficiency

In addition to reimagining our own operations, we’re working to empower developers and data scientists to build and optimize AI models that can achieve similar outcomes while requiring fewer resources. As mentioned earlier, small language models (SLMs) can provide a more efficient alternative to large language models (LLMs) for many use cases, such as fine-tuning experimentation on a variety of tasks or even grade school math problems.

In April 2024, we announced Phi-3, a family of open, highly capable, and cost-effective SLMs that outperform models of the same and larger sizes across a variety of language, reasoning, coding, and math benchmarks. This release expands the selection of high-quality models for customers, offering practical choices for composing and building generative AI applications. We then introduced new models to the Phi family, including Phi-3.5-MoE, a Mixture of Experts model that combines 16 smaller experts into one, and Phi-35-mini. Both of these models are multi-lingual, supporting more than 20 languages.

Learn more about how we’re advancing sustainability through our Sustainable by design blog series, starting with Sustainable by design: Advancing the sustainability of AI.

¹Excerpt from prompting Copilot with: please explain how algorithms relate to LLMs.

²Splitwise: Efficient generative LLM inference using phase splitting, Microsoft Research.

The post Sustainable by design: Innovating for energy efficiency in AI, part 2 appeared first on The Microsoft Cloud Blog.

Today, we’re announcing a power purchase agreement (PPA) with Constellation that will enable the restart of an 835 megawatt (MW) nuclear facility in Pennsylvania that was retired in 2019. This will bring a significant supply of net-new, reliable, carbon-free electricity to the PJM power grid, the regional transmission organization covering 13 states, recognizing the importance of nuclear energy and complementing our 34 gigawatt (GW) contracted renewable energy portfolio in 24 countries.

As highlighted by the International Energy Agency, complete grid decarbonization will require a multi-technology approach with a broad range of carbon-free technologies such as wind, solar, geothermal, clean hydrogen, sustainable biomass, nuclear, fusion, energy efficiency, and storage, as well as transmission infrastructure to connect these technologies to the grids that need them.¹ Alongside our extensive work on carbon reduction and carbon removal, Microsoft embraces this multi-technology approach as an essential pathway to achieving our goal of becoming carbon negative by 2030.

Carbon reduction

Explore our approach to carbon reduction

Learn more

As we continue to expand our portfolio of solutions to accelerate the energy transition, we collaborate with governments, communities, developers, and energy service providers in many ways. In this blog, I’ll share more about how we approach our work to (1) shape market demand for carbon-free electricity and (2) advance energy policy through advocacy.

Shaping market demand to accelerate the addition of carbon-free electricity

We employ a wide range of contracting mechanisms to meet our goals and secure carbon-free electricity, crafting innovative agreement structures alongside our large portfolio of renewable PPAs. A few examples:

• Our recently announced five-year global agreement with Brookfield Renewable Partners provides a pathway for the development of more than 10.5 gigawatts of new renewable energy capacity in the United States and Europe, almost eight times larger than the largest corporate PPA ever signed. This agreement provides an incentive for Brookfield to build a large portfolio of new renewable energy projects in the coming years, contributing to the decarbonization of the grid, and matched to the locations where Microsoft consumes electricity.
• In Washington state, our agreement with Powerex matches hourly datacenter demand with direct deliveries of carbon-free hydro, solar, and wind power on a 24-hour basis throughout the year. During the day, when our contracted renewable resources produce more power than needed, Powerex takes the surplus renewable power, conserving water from hydropower reservoirs and effectively storing it like a battery. This energy can then be delivered back to the datacenter in later hours, for example at night when wind and solar sources may be offline.
• As a global company committed to decarbonization on a global level, Microsoft also has worked to develop renewable energy in communities and locations that often are not prioritized. An example of this unique approach was our five-year framework agreement with Pivot Energy to develop up to 500 MW of community-scale solar energy projects across the US between 2025 and 2029. The agreement will enable Pivot to develop approximately 150 US solar projects in roughly 100 communities across 20 states, including Colorado, Maryland, Illinois, Delaware, Pennsylvania, and Ohio, with each solar project including significant community benefits.  

Advancing carbon-free electricity through policy advocacy

Our public policy advocacy relating to the electrical grid is focused on accelerating the transition to clean electricity generation, modernizing and improving grid infrastructure, and encouraging an equitable energy future. A grid mix that includes adding and retaining firm carbon-free energy technologies as well as renewables will be pivotal to providing electricity access across the globe and progressing decarbonization.

In December 2023, we published a policy brief on advanced nuclear and fusion energy that highlights the importance of carbon-free electricity and the role advanced nuclear and fusion energy will have in a decarbonized energy future. As advanced carbon-free energy technologies are developed, each comes with its own set of considerations, benefits, risks, regulatory dynamics, and acceptance. Our policy priorities are focused on advancing research, development, and demonstration projects; enabling safe deployment of technologies; and encouraging an efficient and effective regulatory process for new technologies to be deployed.

Explore Sustainable by design

Discover more about how Microsoft is advancing the sustainability of cloud and AI through our blog series on the topic:

• Sustainable by design: Advancing the sustainability of AI
• Sustainable by design: Transforming datacenter water efficiency
• Sustainable by design: Innovating for energy efficiency in AI, part 1
• Sustainable by design: Innovating for energy efficiency in AI, part 2
• Sustainable by design: Advancing low carbon materials
• Sustainable by design: Next-generation datacenters consume zero water for cooling

¹The path to limiting global warming to 1.5 °C has narrowed, but clean energy growth is keeping it open, IEA, 2023.

The post Accelerating the addition of carbon-free energy: An update on progress appeared first on The Microsoft Cloud Blog.

Earlier this summer, my colleague Noelle Walsh published a blog detailing how we’re working to conserve water in our datacenter operations: Sustainable by design: Transforming datacenter water efficiency, as part of our commitment to our sustainability goals of becoming carbon negative, water positive, zero waste, and protecting biodiversity.

At Microsoft, we design, build, and operate cloud computing infrastructure spanning the whole stack, from datacenters to servers to custom silicon. This creates unique opportunities for orchestrating how the elements work together to enhance both performance and efficiency. We consider the work to optimize power and energy efficiency a critical path to meeting our pledge to be carbon negative by 2030, alongside our work to advance carbon-free electricity and carbon removal.

Explore how we're advancing the sustainability of AI

Explore our three areas of focus

Read more

The rapid growth in demand for AI innovation to fuel the next frontiers of discovery has provided us with an opportunity to redesign our infrastructure systems, from datacenters to servers to silicon, with efficiency and sustainability at the forefront. In addition to sourcing carbon-free electricity, we’re innovating at every level of the stack to reduce the energy intensity and power requirements of cloud and AI workloads. Even before the electrons enter our datacenters, our teams are focused on how we can maximize the compute power we can generate from each kilowatt-hour (kWh) of electric power.

In this blog, I’d like to share some examples of how we’re advancing the power and energy efficiency of AI. This includes a whole-systems approach to efficiency and applying AI, specifically machine learning, to the management of cloud and AI workloads. Learn more about how we’re bringing efficiency research from the lab into commercial operations in Sustainable by design: innovating for energy efficiency in AI, part 2.

Driving efficiency from datacenters to servers to silicon

Maximizing hardware utilization through smart workload management

True to our roots as a software company, one of the ways we drive power efficiency within our datacenters is through software that enables workload scheduling in real time, so we can maximize the utilization of existing hardware to meet cloud service demand. For example, we might see greater demand when people are starting their workday in one part of the world, and lower demand across the globe where others are winding down for the evening. In many cases, we can align availability for internal resource needs, such as running AI training workloads during off-peak hours, using existing hardware that would otherwise be idle during that timeframe. This also helps us improve power utilization.

We use the power of software to drive energy efficiency at every level of the infrastructure stack, from datacenters to servers to silicon.

Historically across the industry, executing AI and cloud computing workloads has relied on assigning central processing units (CPUs), graphics processing units (GPUs), and processing power to each team or workload, delivering a CPU and GPU utilization rate of around 50% to 60%. This leaves some CPUs and GPUs with underutilized capacity, potential capacity that could ideally be harnessed for other workloads. To address the utilization challenge and improve workload management, we’ve transitioned Microsoft’s AI training workloads into a single pool managed by a machine learning technology called Project Forge.

Currently in production across Microsoft services, this software uses AI to virtually schedule training and inferencing workloads, along with transparent checkpointing that saves a snapshot of an application or model’s current state so it can be paused and restarted at any time. Whether running on partner silicon or Microsoft’s custom silicon such as Maia 100, Project Forge has consistently increased our efficiency across Azure to 80 to 90% utilization at scale.

Safely harvesting unused power across our datacenter fleet

Another way we improve power efficiency involves placing workloads intelligently across a datacenter to safely harvest any unused power. Power harvesting refers to practices that enable us to maximize the use of our available power. For example, if a workload is not consuming the full amount of power allocated to it, that excess power can be borrowed by or even reassigned to other workloads. Since 2019, this work has recovered approximately 800 megawatts (MW) of electricity from existing datacenters, enough to power approximately 2.8 million miles driven by an electric car.¹  

Over the past year, even as customer AI workloads have increased, our rate of improvement in power savings has doubled. We’re continuing to implement these best practices across our datacenter fleet in order to recover and re-allocate unused power without impacting performance or reliability.

Driving IT hardware efficiency through liquid cooling

In addition to power management of workloads, we’re focused on reducing the energy and water requirements of cooling the chips and the servers that house these chips. With the powerful processing of modern AI workloads comes increased heat generation, and using liquid-cooled servers significantly reduces the electricity required for thermal management versus air-cooled servers. The transition to liquid cooling also enables us to get more performance out of our silicon, as the chips run more efficiently within an optimal temperature range.

A significant engineering challenge we faced in rolling out these solutions was how to retrofit existing datacenters designed for air-cooled servers to accommodate the latest advancements in liquid cooling. With custom solutions such as the “sidekick,” a component that sits adjacent to a rack of servers and circulates fluid like a car radiator, we’re bringing liquid cooling solutions into existing datacenters, reducing the energy required for cooling while increasing rack density. This in turn increases the compute power we can generate from each square foot within our datacenters.

Learn more and explore resources for cloud and AI efficiency

Stay tuned to learn more on this topic, including how we’re working to bring promising efficiency research out of the lab and into commercial operations. You can also read more on how we’re advancing sustainability through our Sustainable by design blog series, starting with Sustainable by design: Advancing the sustainability of AI and Sustainable by design: Transforming datacenter water efficiency. 

For architects, lead developers, and IT decision makers who want to learn more about cloud and AI efficiency, we recommend exploring the sustainability guidance in the Azure Well-Architected Framework. This documentation set aligns to the design principles of the Green Software Foundation and is designed to help customers plan for and meet evolving sustainability requirements and regulations around the development, deployment, and operations of IT capabilities.  

¹Equivalency assumptions based on estimates that an electric car can travel on average about 3.5 miles per kilowatt hour (kWh) x 1 hour x 800.

The post Sustainable by design: Innovating for energy efficiency in AI, part 1 appeared first on The Microsoft Cloud Blog.

Last month, we unveiled our Datacenter Community Pledge, emphasizing that datacenters are not only the backbone of modern technology but also a force for good in the communities they serve. As part of this commitment, at Microsoft we recognize our crucial role in protecting and replenishing freshwater resources both in the regions where we operate and around the world.

That’s why in our datacenter operations, one of the essential engineering questions we ask each day is: how can we continue to conserve water while meeting growing customer demand for cloud and AI innovation?

In datacenters, water is primarily used for cooling and humidification. As demand for high performance cloud and AI applications has grown over the past few years to fuel customer applications and enable a new frontier of discovery and innovation, so have the power requirements for silicon chips—the basic building blocks of cloud and AI computing—that sit within the racks and servers of datacenters. Because advanced chips typically utilize more power, they also generate more heat. To prevent the chips from malfunctioning, more intensive cooling is needed, and this has historically required consuming water.

Water replenishment

Four principles guide our strategy

See our approach

To reduce the water required for operations, a critical path to our company goal of becoming water positive by 2030, we’re innovating everywhere from our datacenter buildings all the way to the chips. Collectively, this work is delivering substantial results. From our first generation of owned datacenters in the early 2000s to our current generation in 2023, we have reduced our water intensity (water consumed per kilowatt-hour) by over 80%. This shows that it’s possible to significantly reduce how much water our datacenters use per kilowatt of power even as our cloud infrastructure expands.

Today, we are sharing more about two focus areas for continuing to drive down water intensity: (1) conserving water at every stage of operations, and (2) innovative technologies that reduce the amount of water needed for cooling.

Conserving water at every stage of operations

At all locations, we work to minimize the amount of water we require for cooling. This includes operating our datacenters at a temperature that allows us to cool with outdoor air the majority of the year, reducing the need for ambient cooling, and conserving water at every stage of day-to-day operations.

We conduct regular audits of our datacenters to identify inefficiencies and areas where design and day-to-day use don’t align. Our 2022 audit resulted in targeted improvements that eliminated 90% of the instances in which excess water was used. In addition, we’re building advanced prediction models that help us anticipate water requirements based on real-time weather and operational data. Comparing anticipated needs to actual consumption patterns enables us to quickly identify inefficiencies, such as water leaks that may otherwise go unnoticed.

To minimize freshwater requirements from municipal water systems, we employ conservation strategies that are tailored to the bioregion of the datacenters. For example, in Texas, Washington, California, and Singapore we’ve expanded our use of reclaimed and recycled water. In the Netherlands, Ireland, and Sweden we’re harvesting rainwater, and we’re also bringing this capability to new datacenters in Canada, the United Kingdom, Finland, Italy, South Africa, and Austria.

Innovative technologies that reduce the water needed for cooling

Innovative cooling technologies are essential to Microsoft’s water strategy, and we are rapidly expanding proven solutions across our datacenter portfolio. This includes solutions that bring cooling directly to the source of heat generation—the chip itself.

Cold plates are a prime example of this: a direct-to-chip cooling technology that provides heat exchange in a closed loop system. Cold plates dissipate heat more effectively than traditional air cooling, directly chilling the silicon and then recirculating the cooling fluid, like a car radiator. This solution significantly improves cooling efficiency and enables more precise temperature control compared to traditional methods.

To harness the increased efficiency cold plates offer, we’re developing a new generation of datacenter designs optimized for direct-to-chip cooling, which requires reinventing the layout of servers and racks to accommodate new methods of thermal management as well as power management. In existing datacenters, we’re also using innovations like the ‘sidekick,’ a liquid cooling system we’re already using adjacent to racks of Microsoft Azure Maia AI Accelerator chips, circulating fluid to draw heat away from the cold plates attached to the surface of the chips.  

We’re also evolving cold plate technologies through our work with microfluidics, a technology that brings cooling inside the silicon by integrating tiny fluid channels into chip designs. Embedding the liquid cooling inside the chip brings the coolant right next to the processors, resulting in even more efficiency and precision.

Our newest datacenter designs are optimized to support AI workloads and consume zero water for cooling. To achieve this, we’re transitioning to chip-level cooling solutions, providing precise temperature cooling only where it’s needed and without requiring evaporation. With these innovations, we can significantly reduce water consumption while supporting higher rack capacity, enabling more compute power per square foot within our datacenters.

Reducing global water use through partnership, investing to replenish water

Our water positive goal guides us to consider not only how we can shift our business practices to reduce our water footprint but also how this work can benefit customers and partners working toward similar goals. The five pillars of water positive: reduction, replenishment, access, innovation, and policy all play important roles in our water positive journey.

Over the past year, we grew our water replenishment program significantly, nearly doubling our water replenishment portfolio to include more than 49 replenishment projects around the world. Together, these have the potential to replenish more than 24,000 Olympic size swimming pools over the lifetime of the projects. We also met our 2030 water access target to provide 1.5 million people with access to clean water and sanitation services.²

In addition, we’re working to reduce global water use by collaborating with customers, partners, local communities and municipalities to advance water infrastructure and policy around the globe. Because corporate approaches to water management generally lag investments in carbon reduction¹, we’re taking an active role in championing effective and innovative water management practices and water policies. Some of our advocacy projects include: (1) serving on a coalition to increase water reuse and recycling across the United States, (2) funding projects that support Tribal Nations and state governments in increasing water security, and (3) supporting research, analysis, and advocacy on water in the European Commission.

Learn more about how Microsoft is advancing sustainability

For more information on our progress towards our sustainability goals, read the Microsoft 2024 Environmental Sustainability Report.

Learn more about how we’re advancing sustainability through our Sustainable by design blog series:

• Sustainable by design: Advancing the sustainability of AI
• Sustainable by design: Innovating for energy efficiency in AI, part 1
• Sustainable by design: Innovating for energy efficiency in AI, part 2
• Sustainable by design: Advancing low carbon materials
• Sustainable by design: Next-generation datacenters consume zero water for cooling

¹Why investment in water is crucial to tackling the climate crisis, World Economic Forum, 2024.

²2024 Environmental Sustainability Report, Microsoft.

The post Sustainable by design: Transforming datacenter water efficiency appeared first on The Microsoft Cloud Blog.

Share Cloud Cultures, Part 6: Accelerating collective growth in Malaysia on FacebookShare Cloud Cultures, Part 6: Accelerating collective growth in Malaysia on XShare Cloud Cultures, Part 6: Accelerating collective growth in Malaysia on LinkedIn

• Tag:Azure Global
• Tag:Cloud Cultures

Innovate, Connect, Cultivate

The Cloud Cultures series is an exploration of the intersection between cloud innovation and culture across the globe. 

Malaysia accelerates growth through digital transformation

Amidst the swiftly changing digital landscape, Malaysia stands out as a dynamic force capturing global attention. This nation—enriched by its diverse population comprised of Malays, Indians, Chinese, and more—is home to people and companies that have adeptly embraced innovative technologies, ensuring the benefits extend to all, not just the tech-savvy elite.

Malaysia has established a culture of digital acceleration through industries like energy, farming, and education by striking a balance between growth and the needs of their people. During my travels, I learned how they’ve embraced cloud innovation in a way that allows them to navigate the modern world with confidence and ensure that everyone is along for the ride.

Before setting out to meet with local companies, I joined General Manager of Energy and Utilities for Microsoft Malaysia, Shereeta (full name: Datin Sharifah Shereeta Syed Sheh), for a traditional Malaysian breakfast at her favorite restaurant. We sat down to talk about our upcoming interviews over nasi lemak—a delicious combination of fried anchovies, fish, hard-boiled egg, cucumber, and sambal on fragrant coconut rice, alongside pancakes, coconut grits, and colorful baked treats. Delighted by the food and excited for the day, we parted ways after breakfast. Shereeta headed out to a local chicken farm while I ventured further into the city.

PETRONAS is building a more sustainable world

I began my visit in the heart of Kuala Lumpur at the Golden Triangle, a hub for shopping, commerce, and entertainment. Standing 88-stories tall with a 17-acre park at its base, the PETRONAS Twin Towers are a wonder to behold. The skyscrapers are complete with malls, museums, a philharmonic orchestra, and a skybridge with views of the vibrant city. This is where I met Phuah Aik-Chong, CEO of Petronas Digital, to learn how PETRONAS utilizes the cloud to accelerate digital transformation.

PETRONAS is a dynamic global energy group with presence in over 100 countries. They produce and deliver energy and solutions that power society’s progress, enriching lives for a sustainable future. PETRONAS’ commitment to sustainability starts at the core of their operations and extends throughout their value chain. People are their strength and partners for growth, driving innovation to deliver a unique spectrum of solutions. PETRONAS’ commitment to Malaysia’s progress doesn’t stop at providing oil and gas—they make a concerted effort to provide development opportunities to underserved populations. One such initiative is the BeDigital Bootcamp, which involves upskilling students from various universities in Malaysia. Partnering with Microsoft, they have collaborated on multiple initiatives that reflect the mutual goal of empowering Malaysians to benefit in tandem with the rapid pace of innovation and digital advancements.

Chop Cheong Bee uses e-farming to feed Malaysia

While I stayed in the city, Shereeta took a break from the bustling metropolis and turned down a quiet dirt road. There, she learned about a local company that helps independent chicken farmers use cloud technology to turn their operations into smart farms—improving food security across Malaysia with affordable, high-quality chicken.

Founded in 1985, Chop Cheong Bee began as a poultry trading company, supplying chicken to local markets and establishments in Malaysia. After a brief period of time, they had to close due to an overwhelming number of manual tasks. However, in the late 2000s, they reopened focusing on technology and e-farming practices.

Cloud technology enables Chop Cheong Bee to create environments where chickens can thrive, utilizing a closed and climate-controlled farming system. The solution they developed collects data to inform how much feed is being consumed and the meat conversion ratios, all in real time. Today, Chop Cheong Bee is a crucial poultry supplier that facilitates a sizable portion of the chicken supply in Malaysia.

General Manager of Chop Cheong Bee, Datuk Jeffrey Ng Choon Ngee shared how e-farming is the future:

“With our solution, we can improve the broiler production index by 20 to 30 points. That’s easily a 10 percent improvement. If more farms can achieve this, then the cost of production will drop. And then hopefully, more Malaysians can afford quality poultry.”

Chop Cheong Bee built a system that can produce about 280 to 340 million chickens annually and supply 80 to 100 customers daily. This new way of farming not only provides millions of people with affordable and nutritious meat, but has also attracted a younger, more technology-focused generation of farmers to this vital industry.

Bersama Malaysia ensures citizens are part of the country’s digital acceleration

My final stop in Malaysia was a basketball court to shoot hoops with a recent graduate, Vaashini Palaniappan, who took part in the Bersama Malaysia (Together with Malaysia) program. Alongside sponsors like the Ministry of Education and Microsoft, the initiative teaches students digital skilling, inspiring young students, and women to dream outside the norm and build careers in tech.

Vaashini Palaniappan, data scientist and recent graduate, shared her future aspirations:

“There are so many women in this data and AI field that want to invent something, that want a brighter future. Because of this, I’m inspired to do something different. I want to be inventive using AI.”

Growing up in a small town, Vaashini didn’t have a lot of exposure to technology. But by participating in university programs, she was able to study sciences, learn technical skills, and understand the impact of advanced technologies on medicine. After seeing a close friend pass from cancer, Vaashini said she was determined to become a doctor and leverage innovative technology for good—specifically, to use AI to detect early signs of cancer and build a hyper-personalized treatment plan for patients.

Bersama Malaysia, along with Microsoft’s Code with Barriers program, were created to ensure citizens of Malaysia are a part of the digital acceleration of the country. These programs are empowering Malaysia’s inclusive digital economy and advancing the nation’s digital transformation across the private and public sectors. Malaysia has consistently been a trailblazer in fostering opportunities for its citizens. Through initiatives like Bersama Malaysia, the nation ensures that no one is left behind in the dynamic landscape of transformation.

If we’d prefer to quote faculty of the program, instead of Vaashini we could also use this one:

“Even before this partnership with Microsoft, we were aware that the Microsoft Learn platform offers a vast selection of professional certifications related to digital skills, but what really stood out was that Microsoft also supports institutions to manage the certification process independently. This way, we can customize the upskilling program according to a timeline and cost that works best for our students”—Dr. Hamizah binti Mohd Safuan, Deputy Dean of FAST

Innovating together makes change happen

Later that evening, Shereeta and I discussed our journey over my first experience with a popular local fruit: the durian. After getting used to the infamous smell, I snacked on the custard-like meat and reflected on Malaysia’s inspiring commitment to extending growth far beyond the gleaming skyscrapers and urban epicenters. This version of cloud culture ensures that as the pace of progress quickens, it doesn’t come at the cost of anyone being sidelined. As is often the case, I saw in Malaysia that the best way to accelerate growth isn’t racing ahead; it’s moving forward together.

In this ever-changing world, there is always more to experience. See you on my next adventure!

Learn more:

• Watch more Cloud Cultures episodes
• Find the Azure geography that meets your needs

Cloud Cultures

Explore the ways people across the globe are using Azure technology

Learn more 

The post Cloud Cultures, Part 6: Accelerating collective growth in Malaysia appeared first on The Microsoft Cloud Blog.

Take environmental sustainability, for example. The stakes couldn’t be higher. Pollution and climate change affect all of us and threaten to have a lasting impact on future generations. People care deeply about protecting our planet, but many do not know what actions they can take to make a difference.

I found an empowering example of using AI and data to support sustainability in a recent episode of the Pivotal podcast. This episode features Loic van Cutsem, Director of international partnerships and development at BeeOdiversity.

This Belgium-based tech company is harnessing a 12 million-strong workforce to monitor biodiversity—the health of an ecosystem.

Microsoft AI

A new era of AI has arrived

Explore solutions

BeeOdiversity uses AI to interpret data on biological health

BeeOdiversity saw an opportunity to use bees as data collectors at scale. As bees search for nectar in flowers and other plants, they collect valuable information from gardens, fields, and other features in the land and air as they browse. By analyzing the pollen and other microscopic materials that the bees bring back to the hive, BeeOdiversity can get a living snapshot of the biological health of any designated area. This data can reveal factors that harm the environment and where they are located, such as invasive species, heavy metals, and pesticides.

BeeOdiversity leverages AI to interpret this data from bees and offer insights to their clients. Their recommendations do more than help companies comply with regulatory requirements. By having this information, communities and businesses can work together to improve the ecosystem that surrounds them, from increasing water quality to protecting soil health, preventing flooding, and mitigating climate change impacts. They have a better understanding of what actions to take to operate more sustainably, and an army of tiny data collectors who can continue to measure their progress.

That’s what I love about this use case—it’s AI-powered innovation that’s incredibly impactful and scalable to communities around the world. And what’s inspiring about BeeOdiversity’s story is that it holds lessons for other organizations that want to use data and AI to solve complex business problems.

Having good data is essential for driving change

Getting clean and accurate data is essential for any organization to drive change. Data provides a common basis to identify issues and discover solutions. This is especially important if you’re using AI to address a societal issue such as sustainability, which requires collaboration across multiple sectors. When there are so many stakeholders at the table, the recommendations from AI need to be powered by data that everyone can trust and accept.

Today there is no standardized data source in the field of biodiversity monitoring. But BeeOdiversity is unique in how they have recruited bees to collect data through their everyday pollinating behavior. By using bees, they are essentially standardizing the data collection, while giving nature a chance to give us feedback. For this reason, BeeOdiversity has not faced much skepticism from clients. You can’t argue with the bees.

The stakeholders are almost in awe of all this information gathered by the bees, and they’re responsive to the findings. They understand that this is what the bees are telling us. It’s the truth.

Loic van Cutsem, Director of international partnerships and development, BeeOdiversity

Democratizing access to data leads to impact at scale

Quality data is a crucial ingredient for success with AI. But at the end of the day, data is only a tool. It’s what you do with the data that matters the most. Data’s real value comes from how it empowers people to make smarter decisions, and you can maximize that impact by expanding the number of people who have access to that data.

BeeOdiversity has been building their proprietary data set for the last decade. Now, they are working to empower as many people as possible to make better sustainability decisions with that data, including businesses, governments, and soon, individuals.

They have customers from a wide range of industries, from agrifood producers to financial institutions, that are all keen on better managing their natural resources to build more resilient businesses. Governments are also working with BeeOdiversity to improve the environment for the communities they serve. For example, Knokke-Heist, a coastal city in Belgium, was able to achieve a 300% drop in pesticide levels and a four-times increase in plant diversity.  

And for maximum impact at scale, BeeOdiversity is developing a mobile app that will allow people to scan the plants in their garden and get tailored recommendations on how to improve their local ecosystem.

By democratizing access to their invaluable data, BeeOdiversity is enabling better sustainability decisions across society. This shows the power of having standardized data that everyone can rally behind.

Accelerate innovation by unlocking new insights

One of the benefits of AI is how quickly it can accelerate innovation for those who embrace it. Just point the AI to new sets of data, and it unlocks more levels of analysis and insight, allowing businesses to easily expand their coverage area or improve their processes.

BeeOdiversity is a great example. They developed a solution called BeeOimpact that incorporates satellite imagery to assess the crops and industries in a given area. It draws from this satellite data, along with data from their existing dataset that was collected by bees, and uses a predictive AI model to provide almost instantaneous estimates of the pesticides that are likely to be there.

BeeOdiversity is also developing a method to allow beekeepers to take photos of pollen, instead of manually collecting the pollen to send in for lab analysis. Their aim is to use computer vision and machine learning to identify the pollen in the photo. This new method would require less human effort and speed up their process of analyzing pollen samples.

As they develop these new capabilities, BeeOdiversity uses Microsoft Azure’s data and AI capabilities to enable their AI use cases, and GitHub as their platform for developers. With these tools, BeeOdiversity can achieve an extraordinary pace of innovation and make a sizable impact for sustainability.

Listen to the full conversation on Pivotal

By standardizing biodiversity data, empowering people to take action, and unlocking new innovation at a fast pace, BeeOdiversity is tackling sustainability head-on with the help of technology.

It’s brilliant how Loic and his team have found a way to leverage the diligence of bees with the analytical power of AI to unlock the kind of insights that many of us have dreamed about for decades. They’re giving a voice to mother nature and giving us humans a chance to take action so that we can live in greater harmony with the planet.

To hear more about BeeOdiversity and their ingenious use of AI, listen to the episode on the Pivotal podcast.

The post How BeeOdiversity leverages 12 million bees and AI to create a more sustainable future appeared first on The Microsoft Cloud Blog.

Gather and analyze all your ESG data in one place with Microsoft Fabric

With sustainability data solutions in Microsoft Fabric (preview), you can analyze your organization’s environmental, social, and governance (ESG) data together with your other enterprise data to inform more holistic decisions and better-targeted outcomes. Gather, harmonize, and transform sustainability data into meaningful, actionable insights, and use advanced analytics and powerful AI to help you prepare data for analysis, regulatory reporting, and AI-driven innovation.  

You can also use sustainability data solutions in Fabric to validate your data and track progress against publicly available data and industry benchmarks. All this comes with a shared governance model across various capabilities and a unified space for your data stewards and sustainability practitioners to interact through Microsoft Fabric.  

Four solution capabilities help you holistically meet your sustainability requirements:  

1. ESG data estate (preview) helps you centralize and standardize ESG data from your disparate data sources to compute, analyze, and disclose ESG metrics for various regulatory reporting and analytics requirements.  
2. Microsoft Azure emissions insights (preview) enable you to report and analyze your Microsoft Azure usage–related emissions data at subscription and resource levels. 
3. Environmental metrics and analytics (preview) help you generate custom reports, metrics, and analytics insights across carbon, water, and waste by connecting to your data in Microsoft Sustainability Manager.  
4. Social and governance metrics and reports (preview) provide insights, dashboards, and metrics to support your needs across various sustainability directives.   

Build your ESG data estate (preview) 

Sustainability disclosures, analytics, and reduction initiatives require rich ESG data sets that originate from disparate sources, and this data needs to be unified and standardized to improve its efficiency and value. sustainability data solutions in Microsoft Fabric (preview) provides pre-built data pipelines and lakehouses to combine social and governance data from different enterprise systems with environmental data from Microsoft Sustainability Manager and other systems. 

With your ESG data estate in place, you can process unified sustainability data to compute ESG metrics for sustainability disclosure requirements such as Corporate Sustainability Reporting Directive (CSRD), Global Reporting Innitiative (GRI), and many others using provided prebuilt data processing artifacts. You can run this process on demand or on a schedule.

Built-in dashboards let you view data and insights from a variety of perspectives—such as by facility or operating unit—and use workflows to help prepare reports. You can mark the metrics required for CSRD and other disclosures and prepare the reports to share with auditors. 

Gain detailed Microsoft Azure emissions insights (preview) 

The typical IT efficiency journey for many organizations starts with migrating and then optimizing workloads in the cloud, which involves factoring potential emissions reduction strategies. Microsoft Fabric facilitates this process by enabling you to unify and analyze your Azure emissions data against your cloud usage.  

With all your Azure emissions data in Microsoft Fabric, you can query and drill down into Azure resource-level emissions for advanced reporting and analysis. Use pre-built data pipelines that ingest and store resource-level Azure emissions data in tabular data. And use Power BI dashboards to drill down and compare emissions data across subscriptions and resources, helping to identify patterns that evolve with time and usage.  

Note that all Azure customers can now easily access data and insights related to their cloud usage-based emissions by simply signing into the Azure portal and navigating to Azure carbon optimization (preview). This feature provides an overview of your subscription’s emissions data for the past 12 months and by service type—such as virtual machines or storage. 

On the Emissions Details page, you’ll see a monthly breakdown of the top Azure resources that contribute to your organization’s total emissions. By comparing this data to the previous month, you can see the percentage change and identify resources to turn off or utilize more efficiently. On the Emissions Reductions page, you’ll find recommendations to improve your organization’s cloud efficiency and sustainability. 

Enrich your understanding of environmental metrics and analytics (preview) 

To better understand your progress on reduction and other initiatives across carbon, water, and waste, you may need to define and compute custom metrics. Microsoft Fabric lets you connect to your relevant data in Microsoft Sustainability Manager and provides tables to query, compute custom metrics, and analyze the data further. You can enrich your sustainability data with the other corporate and business data for advanced machine learning–based analytics and leverage prebuilt Power BI dashboards for detailed insights and visualizations. 

Analyze your social and governance metrics and reports (preview) 

Increasingly, organizations are required to disclose their sustainability performance in social and governance areas—for example, for CSRD. To address this need, Microsoft Fabric enables you to unify and prepare disparate data from corporate systems handling human resources, health and safety, and corporate governance data in a single ESG data estate. From there, you can compute and report social and governance metrics required for disclosures and use Power BI dashboards to visualize and drill down into selected areas. 

Introducing Copilot in Microsoft Sustainability Manager: Quickly turn your sustainability data questions into insights 

When you’re working with disparate ESG data from across your operations and value chains, getting answers to critical questions quickly can be challenging, potentially slowing down progress toward sustainability goals. Copilot in Microsoft Sustainability Manager, now in preview, immediately delivers insights from your data based on natural language queries.  

Ask Copilot a question and it will work across Microsoft Sustainability Manager to quickly understand environmental data and provide an answer. For example, you can ask Copilot the right global warming potential (GWP) value for a given gas and assessment report (AR) version. With Copilot’s assistance, you can more confidently work through tasks like drafting reports on your organization’s emissions or CSRD environmental metrics for a quarterly update. Copilot helps you generate the draft reports, reducing your preparation time. 

Copilot can also help simplify and accelerate complex processes within Microsoft Sustainability Manager, such as creating a calculation model for your mobile combustion data. Within seconds, a calculation model is created using natural language, and within minutes these emissions are calculated, providing a more complete picture of reduction opportunities. 

Copilot in Microsoft Sustainability Manager is trained on Microsoft Cloud for Sustainability data schemas, making it an effective and knowledgeable tool for improving efficiency across various tasks.  

Magnify your visibility into your ESG data with AI-powered insights 

It’s important to trust the accuracy of your organization’s ESG data, which can be large and complex. The faster you can identify errors or missing data, the faster you can resolve or fill in the data. Intelligent insights in Microsoft Sustainability Manager, now available in preview, provides the visibility into your ESG data needed to help identify outliers, trends, and correlations.

See where data needs cleaning or where you need more complete data. This feature scans your organization’s data within Microsoft Sustainability Manager and identifies opportunities—both short-term and long-term—for reductions that align to your sustainability initiatives. Insights are based on historical trends, seasonality, and data anomalies. Looking deeper into the results, you can identify opportunities for reduction in your organization’s Scope 3 emissions.  

Learn more about how intelligent insights can help you make sense of an increasingly complex data landscape. 

Now generally available in Microsoft Sustainability Manager 

Track, manage, and report your water and waste sustainability data 

In addition to expanding AI capabilities, we’re continuing to evolve Microsoft Sustainability Manager to meet broader environmental sustainability objectives. Along with carbon emissions data capabilities, water and waste sustainability data capabilities are now generally available.  

Organizations can track and report their water accounting, water usage efficiency, and compliant water discharges across multiple facilities, safe water discharge regulations, and water usage disclosure standards. Microsoft Sustainability Manager helps you understand the sources and quantities of waste generation at your facilities and how the waste gets disposed of. This can further help you discover avenues to increase waste recycling for specific waste sources and reduce off-site waste disposal through landfills and incineration.

Simplify supplier data collection with ESG value chain solution  

We’re excited to announce the general availability of ESG value chain solution in Microsoft Sustainability Manager, enabling you to gather data more directly and securely from your suppliers. You can use the data to calculate suppliers’ emissions and gather partner-specific emission factors to calculate your Scope 3 emissions based on your consumption with those partners.

Experience the new capabilities in action with This is AI…for Sustainability  

Hear from Melanie Nakagawa, Chief Sustainability Officer, Shelly Blackburn, Global Vice President for Sustainability Go to Market, and Satish Thomas, Corporate Vice President, Microsoft Industry Clouds during this free digital event. Get guidance on your sustainability journey and learn how to drive business transformation with Microsoft data and AI solutions. Watch the webcast on demand.

Learn more about sustainability solutions with Microsoft

• Want to learn more about Microsoft Cloud for Sustainability? Sign up for news and updates and try it for free.
• Discover solutions with Microsoft Sustainability Manager.

The post Drive sustainability transformation faster with new data and AI capabilities appeared first on The Microsoft Cloud Blog.