Power, water, and community: Key pieces in the data center jigsaw of sustainability

Less than a decade ago, data centers were an almost invisible infrastructure, quietly enabling a more digital world. Today, they have been thrust into the limelight. According to Morgan Stanley, global capex spend for data centers will total $3 trillion between 2025 and 2029 with unprecedented visibility. Furthermore, data centers are of significant geopolitical significance, as the US, China, and Europe seek the commanding heights in the AI race.

Yet, even as data centers captivate investors and motivate global leaders, their resource intensive nature has sparked critical conversations around sustainability. From energy to water consumption, the use of materials and carbon emissions to local community engagement, the footprint of a data center extends far beyond its computational capabilities.

In some regions, opposition from local communities is rising, as the perceived challenges of data centers outweigh the benefits this critical digital infrastructure brings.

Addressing sustainability in the data center market is a little like piecing together a jigsaw. Each theme, such as energy, water, and community engagement is like an individual part of the puzzle. However, data center owners and developers must be able to see the big picture if they are to resolve all aspects of sustainability.

In the future, successful data centers will depend not just on their core functional capabilities, they will also need to address social and environmental concerns. The challenge now is a holistic one: how can we make data centers sustainable for our society?

Data centers are critical digital infrastructure

Attend any data center conference these days and a common theme emerging is the recognition that the industry has been firmly thrust into the public eye after many years of playing behind the scenes. For some players in this sector this will be a novel feeling and perhaps a little disconcerting.

Moreover, this sector is of strategic importance to society and continues to deliver important benefits to consumers. During the worst of the pandemic, data centers ensured essential services were maintained – from banking to streaming content.

The US has designated data centers as critical national infrastructure. The Trump administration in 2025 issued an executive order to accelerate permitting and the development of data centers, citing their role in national security and emphasizing the importance of AI leadership.

In the UK as well, data centers are recognized as critical national infrastructure, meaning they are on par with energy, water, and the emergency services. In the UK and Germany, developers are exploring recycling waste heat from data centers to feed urban district heating systems.

Given the substantial waste heat generated, these facilities can repurpose this surplus energy, through integration with heat networks, to drive whole system efficiency and reduce their environmental impact.

Satisfying a voracious energy appetite

Data centers, particularly those focused on AI learning and inference, consume significant quantities of power. According to the International Energy Agency (IEA), one large data center consumes as much energy as 100,000 households, or the equivalent to a small city!

In addition, over the short term, the power demand of data centers is disproportionately high in some geographies. While space cooling and heating, plus transport electrification dwarf data centers in absolute volumes as a share of final power demand over the medium term, data centers already represent 30 percent of all power demand in Virginia and over 20 percent in Ireland.

As for sourcing that energy, the reality is if a data center cannot secure a power grid connection, and the grid is already stretched and in urgent need of upgrades in many regions, natural gas is the likely end solution.

Renewables, nuclear power (including SMRs), and geothermal energy are all contenders for power sources over time. In the UK for example, a consortium including EdF, recently announced the deployment of a 300MW SMR at the former Cottam coal-fired power station in Nottinghamshire to power data centers.

However, while these low-carbon technologies have been adopted or are being explored – there are challenges. In the case of renewables, their intermittency precludes a reliable baseload of power. Nuclear will take some years to deliver with SMRs unlikely to be deployed and standardized until the early 2030s. Hence the reason why in the US we see many data centers opt for ‘island power’, whereby combined cycle gas turbines (CCGTs) generate onsite power and are fed by natural gas pipelines.

The IEA sees power generation investment for nuclear and renewables dominate by 2030, with natural gas being the incumbent fuel in the next few years for those data centers unable to secure power grid connections.

It is worth noting that even in the case of natural gas, connections will also need to be secured, which will have a time, cost, and consenting impact. Moreover, the use of gas will increase methane emissions and challenge the technology sector’s net zero ambitions.

Water sustainability and lifecycle carbon are the elephant in the room

For all the discussion around energy demand, water consumption is likely to become as important in the data center world. The World Economic Forum highlighted that a 1MW data center can use up to 25.5 million litres of water annually for cooling, which is equivalent to ten Olympic-sized swimming pools. Bear in mind that data centers are getting larger with some exceeding 1GW, therefore the impact on water demand could be considerable.

Earlier this year, Bloomberg highlighted that data centers are increasingly being built in water stressed regions – from West Coast US to the Middle East. In many cases, major data centers source most of their water from potable sources – putting pressure on local community water systems.

In response, data center owners and developers are implementing innovative water management practices. This can range from onsite reuse to co-location with nearby water treatment and industrial facilities. Moreover, new AI data centers will use liquid cooling solutions that can reduce water consumption by up to approximately 50 percent.

As for lifecycle carbon (the combination of embodied carbon and operational carbon), this will play a larger role in a data center’s carbon footprint as grids decarbonize. As in the case of electric vehicles (EVs). The viability of this technology was rapidly questioned by consumers once they became aware of sustainability issues associated with batteries – for example, the use of child labour in the Democratic Republic of Congo to mine cobalt, a critical element used in electric vehicle batteries.

Strategies to reduce lifecycle carbon range from use of sustainable materials and the reuse of existing buildings, to accelerating low carbon solutions across the supply chain.

Local community engagement will become increasingly more important for project delivery

Local communities around the world are being consulted about new data center builds and there is opposition to construction from some residents. Greenfield sites are inevitably more sensitive than brownfield. Residents may be nervous about their electricity bills going up, as new power infrastructure needs to be built to address AI power demand. They may also be concerned about water consumption levels by the center if they live in or near a water stressed area.

In the US, there are examples of residents opposed to data centers where there is an adverse impact on air quality (a case in point being the xAI data center in Memphis, where portable methane gas turbines were used to power the facility without air permits). As for looking ahead into the medium term, how will residents react if the data center owner advises a SMR will power the facility? Will residents warm to the idea of this being installed in their neighbourhood?

As we are already witnessing with the Great Grid Upgrade in the UK, some residents, while in favour of the energy transition, may oppose the erection of electricity pylons in the countryside. Similarly, the economic benefits of opening a data center (jobs, district heating, and the funding of local projects) may be outweighed by the concerns highlighted above, pitting local communities against developers. Google, for example, halted a $200m data center project in Chile following environmental concerns.

Looking at the big picture of sustainability and resilience

It may seem rather daunting for data center owners and developers to tackle this jigsaw of themes collectively. The temptation may be to focus on the more immediate and pressing issue of power, for example, while addressing the additional aspects of sustainability at a later date. However, like a jigsaw, understanding the broad outline of the picture will allow the individual pieces to be slotted in correctly and more quickly.

Here are a few thoughts to steer thinking and provide a more coherent road map to address sustainability:

1) Siting of data centers is key and optimal sites are dwindling in number

The siting of data centers at strategic locations, where there is already grid access for electricity and gas, such as decommissioned power stations or industrial sites, will be important. However, even these industrial sites may not be ideal for new build AI data centers given the size of the campuses required and the reinforced load bearing floors needed for the equipment weight.

As data centers proliferate, operators will need to think more creatively about securing their water supply while minimizing the environmental impact. This could involve co-locating the data center near sources of industrial or municipal wastewater that could be reused for cooling.

Siting data centers near renewable energy sources could allow more air cooling – a process that requires energy to power fans. Siting in colder regions could also reduce the energy needs for cooling, hence the popularity of the Nordics for data centers.

2) Energy procurement strategies need to be diversified and progressive

Access to power and energy is by far the primordial concern for those players looking to establish AI data centers. While access to a decarbonized grid is the preferred option, there are challenges. Aside from regions plagued with connection queues, where access could be several years out, the grid more broadly needs to be upgraded and added to.

Therefore, the energy portfolio powering a data center needs to be diversified and teed up for newer technologies when ready to be adopted. So, natural gas feedstock and onsite CCGTs might be the key power source over the next few years, as we see in the US for example. There are also bridging technologies, such as fuel cells and potentially biogas, as well as longer term energy plays regarding nuclear energy and the deployment of SMRs. Developers and owners will need to consider immediate energy solutions while keeping one eye on the future.

As for redundancy solutions, data center players might consider replacing diesel back up generators with natural gas, as well as adopting complementary technologies, in the shape of battery energy storage solutions (BESS).

However, it is worth bearing in mind the route to market for adopting these technologies, as well as the complexity and financial implications of these decisions.

On the former, partnerships are a key strategic option to adopt new technologies. Companies like Equinix, which is keen to partner with energy technology innovators, has collaborated with Bloom Energy to pioneer the deployment of fuel cells and has struck multiple nuclear partnerships for microreactors, SMRs, and molten salt breed and burn reactors.

As for the latter point around complexity, some new technologies may require a change in the site configuration to adopt a new technology. A good example is the use of hydrogen at an industrial site where there will be conversion costs for using a gas with a different risk profile to what the site was originally designed for.

Partnerships are a key strategic option to adopt new technologies. These partnerships will need to combine financial investors, technology companies, and energy providers. By doing so, market players will successfully syndicate financial and technical risk.

3) Engagement with all key stakeholders early on will determine long term success

As data centers proliferate and consume natural resources, it is essential owners and developers engage key stakeholders well before the construction process. For example, municipal planning authorities and water utilities will need to co-ordinate with one another to understand the implications of data center deployments in potentially water stressed geographies and address the implications for water management.

Developing an engaging market narrative, to align local communities early on regarding the benefits of data center deployment, will be necessary to ensure projects are delivered in a timely manner. Some governments are grappling with the narrative around AI and the impact on employment, especially for the younger generation. Similarly, data center owners and developers will need to articulate the importance and benefits of their facilities to local residents

The data center sustainability jigsaw is complex. While energy demand dominates current headlines, water demand, lifecycle carbon, and community engagement are rapidly becoming the next focal points for discussion. Understanding the different elements individually and holistically will be key to long term success. There is already some market speculation of an AI bubble. By ensuring all aspects of sustainability are addressed in data centers, this will not only be good for the environment, but it also makes for more resilient business models.