Cathay Financial CIO Sophia Cheng and Ceres CEO Mindy Lubber argue that investors and companies should take greater notice of AI’s growing strain on water resources.
The rapid growth in artificial intelligence (AI) has defined the economy of the 2020s – not only because AI itself is reshaping the ways we work, but because the infrastructure to support it is impacting the places we live.
As investors and companies continue to focus on supporting AI and the digital future, they must not lose sight of the financial risks from the real-world buildout of data centres. Their impacts on energy demand have become a point of major focus as utility bills have increased across the US and in other regions saturated with data centres. And, amid growing local backlash to their construction, there has also been increased focus on the data centres’ water consumption.
The direct, on-premises water usage of each data centre is certainly significant, primarily used for cooling purposes so the facilities’ computers can keep operating. But cooling tells only part of the story. The indirect and cumulative impacts beyond their property lines are substantial and have received little attention.
To understand these broader pressures, consider the cumulative impacts of data centres on water supply. Data centres are often built in clusters, guided by local resources, as well as local economic and political landscapes. While any one individual data centre may have limited impact on supplies, it is a much greater risk when many are drawing water from the same basins.
This is why researchers expect that by 2030, 52% of global data centre hubs will face high or very high water risk – increasing to 58% by 2050 – due to increased demand for data centres and climate-related water scarcity.
Carbon connection
And because data centres require significant energy to operate, they also impact the amount of water used by power plants to generate the electricity. Taken together, these interactions show how tightly water use is interlinked with energy demand and the emissions associated with power generation. This water, energy, and carbon connection means that companies and investors need to evaluate these risks as an integrated system rather than separate issues.
Ceres research in Phoenix, Ariz., for example, found that direct water use in data centres there account for about 1.5 million cubic metres of water per year, which is projected to grow to about an eye-popping 14 million cubic metres in the next few years. But indirect usage already accounts for 11 million cubic metres, which is expected to grow to 55 million cubic metres in the same timeframe.
The indirect water demand for electricity compounds the cumulative impact of many data centers and strain on already delicate basins. Our research found that the Phoenix area’s projected growth in data centres will exacerbate water stress in parts of the region by as much as 32%, with even greater impacts in the summer months when cooling demand is greatest.
Across the supply chain
And yet, fully understanding data centres’ water risks requires looking to the supply chains that enable them. Data centres, for example, require millions or even billions of semiconductors, electronic chips that power virtually all modern technology. And manufacturing them – a practice that is rapidly growing to meet data center demand – takes up a lot of water, too.
The average semiconductor factory today uses about as much water as 33,000 U.S. homes, or 10 million gallons per day, a figure that has quintupled in the last decade and is expected to double again over the next 10 years. And, like data centres themselves, an outsized number of semiconductor factories are found in areas that already have high water stress. By 2040, two out of every five will be drawing from already strained basins.
Recognising this risk, some chipmakers are working to curb water use with water recycling systems and advanced monitoring tools that flag leaks and maintenance needs early. These are just a slice of water management solutions other companies can adopt.
Further up the supply chain, mining for the raw materials that ultimately shape the digital world adds its own significant water use.
All told, data centres are driving a lot of water consumption, most of which investors and companies would not account for if only looking at cooling and other on-site uses.
And with so much of this consumption happening in already strained water basins, there are significant financial risks to companies and their investors – and thus to the burgeoning AI industry that they are trying to shape. These include water scarcity disrupting operations and intensifying pressure on water resources, triggering reputational, legal, and political risk.
Indeed, we’re already seeing local communities pushing back against existing and planned data centres, in some cases shutting down proposed projects. In Chile, following community protests over water usage, a local court temporarily revoked authorisation for a US$200 million data centre. The company halted plans for the facility in order to revise the data centre design to address water use and comply with more stringent environmental requirements.
Closing gaps through dialogue
These risks make it especially important for data centre operators and their investors to account for and address their water usage. That is why Cathay Financial Holdings uses water-related indicators and defines water as one of its key engagement issues to help determine which companies it should engage with, ensuring that firms with significant water dependence or meaningful gaps in their current practices are prioritised for dialogue, including companies from a broad range of sectors such as semiconductors and apparel.
The Valuing Water Finance Initiative supports investors engaging with high-tech companies on how they assess and disclose their water challenges and impacts and develop strategies and best practices to tackle them. The recently released Valuing Water Finance Initiative Benchmark report helps inform these engagements.
Tech companies and data centre operators, meanwhile, should take action to better understand and mitigate their water risks. This includes considering water risk in business planning and siting decisions and working with peers across the industry to continue to develop best practices and accelerate technological solutions to the industry’s water challenges.
As AI increasingly impacts our economy, it’s important that all stakeholders have a thorough, holistic view of data centre water consumption, one that looks across the supply chain and at their cumulative impacts on local and regional water stress. Otherwise, they will miss some of the most sizable risks facing their investments – with ripple effects that extend into the balance sheet and the markets.
Cathay Financial CIO Sophia Cheng and Ceres CEO Mindy Lubber argue that investors and companies should take greater notice of AI’s growing strain on water resources.
The rapid growth in artificial intelligence (AI) has defined the economy of the 2020s – not only because AI itself is reshaping the ways we work, but because the infrastructure to support it is impacting the places we live.
As investors and companies continue to focus on supporting AI and the digital future, they must not lose sight of the financial risks from the real-world buildout of data centres. Their impacts on energy demand have become a point of major focus as utility bills have increased across the US and in other regions saturated with data centres. And, amid growing local backlash to their construction, there has also been increased focus on the data centres’ water consumption.
The direct, on-premises water usage of each data centre is certainly significant, primarily used for cooling purposes so the facilities’ computers can keep operating. But cooling tells only part of the story. The indirect and cumulative impacts beyond their property lines are substantial and have received little attention.
To understand these broader pressures, consider the cumulative impacts of data centres on water supply. Data centres are often built in clusters, guided by local resources, as well as local economic and political landscapes. While any one individual data centre may have limited impact on supplies, it is a much greater risk when many are drawing water from the same basins.
This is why researchers expect that by 2030, 52% of global data centre hubs will face high or very high water risk – increasing to 58% by 2050 – due to increased demand for data centres and climate-related water scarcity.
Carbon connection
And because data centres require significant energy to operate, they also impact the amount of water used by power plants to generate the electricity. Taken together, these interactions show how tightly water use is interlinked with energy demand and the emissions associated with power generation. This water, energy, and carbon connection means that companies and investors need to evaluate these risks as an integrated system rather than separate issues.
Ceres research in Phoenix, Ariz., for example, found that direct water use in data centres there account for about 1.5 million cubic metres of water per year, which is projected to grow to about an eye-popping 14 million cubic metres in the next few years. But indirect usage already accounts for 11 million cubic metres, which is expected to grow to 55 million cubic metres in the same timeframe.
The indirect water demand for electricity compounds the cumulative impact of many data centers and strain on already delicate basins. Our research found that the Phoenix area’s projected growth in data centres will exacerbate water stress in parts of the region by as much as 32%, with even greater impacts in the summer months when cooling demand is greatest.
Across the supply chain
And yet, fully understanding data centres’ water risks requires looking to the supply chains that enable them. Data centres, for example, require millions or even billions of semiconductors, electronic chips that power virtually all modern technology. And manufacturing them – a practice that is rapidly growing to meet data center demand – takes up a lot of water, too.
The average semiconductor factory today uses about as much water as 33,000 U.S. homes, or 10 million gallons per day, a figure that has quintupled in the last decade and is expected to double again over the next 10 years. And, like data centres themselves, an outsized number of semiconductor factories are found in areas that already have high water stress. By 2040, two out of every five will be drawing from already strained basins.
Recognising this risk, some chipmakers are working to curb water use with water recycling systems and advanced monitoring tools that flag leaks and maintenance needs early. These are just a slice of water management solutions other companies can adopt.
Further up the supply chain, mining for the raw materials that ultimately shape the digital world adds its own significant water use.
All told, data centres are driving a lot of water consumption, most of which investors and companies would not account for if only looking at cooling and other on-site uses.
And with so much of this consumption happening in already strained water basins, there are significant financial risks to companies and their investors – and thus to the burgeoning AI industry that they are trying to shape. These include water scarcity disrupting operations and intensifying pressure on water resources, triggering reputational, legal, and political risk.
Indeed, we’re already seeing local communities pushing back against existing and planned data centres, in some cases shutting down proposed projects. In Chile, following community protests over water usage, a local court temporarily revoked authorisation for a US$200 million data centre. The company halted plans for the facility in order to revise the data centre design to address water use and comply with more stringent environmental requirements.
Closing gaps through dialogue
These risks make it especially important for data centre operators and their investors to account for and address their water usage. That is why Cathay Financial Holdings uses water-related indicators and defines water as one of its key engagement issues to help determine which companies it should engage with, ensuring that firms with significant water dependence or meaningful gaps in their current practices are prioritised for dialogue, including companies from a broad range of sectors such as semiconductors and apparel.
The Valuing Water Finance Initiative supports investors engaging with high-tech companies on how they assess and disclose their water challenges and impacts and develop strategies and best practices to tackle them. The recently released Valuing Water Finance Initiative Benchmark report helps inform these engagements.
Tech companies and data centre operators, meanwhile, should take action to better understand and mitigate their water risks. This includes considering water risk in business planning and siting decisions and working with peers across the industry to continue to develop best practices and accelerate technological solutions to the industry’s water challenges.
As AI increasingly impacts our economy, it’s important that all stakeholders have a thorough, holistic view of data centre water consumption, one that looks across the supply chain and at their cumulative impacts on local and regional water stress. Otherwise, they will miss some of the most sizable risks facing their investments – with ripple effects that extend into the balance sheet and the markets.
Share via:
Recommended for you