US golf courses use 30 times more water than US data centres. About one in seven Americans played golf at least once in the past year, according to the PGA. Yet virtually everyone uses data-centre-backed services daily. One is critical to our society’s standard of living and to most organisations’ operations. Yet only one would you want to live next to – would boost the value of your home, rather than detract from it. And one tends to pay a lot of local taxes, where the other often gets lighter treatment.
This is the quandary: everyone uses data centres, few use golf courses. Yet public opinion has turned aggressively – and arguably unfairly – against data centres. This isn’t to rat on golfing! It just shows how many people’s assumptions and expectations of modern digital technology have got significantly out of balance. And it’s up to the hyperscalers to bring the public back on side.
Cloud computing needs to be grounded on terra firma
The ‘cloud’ sounds weightless, as if our photos, emails and videos are floating somewhere in the sky. Yet in reality, the cloud is very physical: steel, concrete, fibre optic cables and computer chips are the building blocks for vast data centres that are very much grounded on terra firma.
These buildings were already proliferating fast before the recent remarkable acceleration as increased AI adoption drives a surge in demand for computing power. For cost and operational reasons, data centres tend to cluster, rather than being evenly spread. This is true both globally and locally. To reduce ‘latency’ – the lag between when one computer interacts with another – it helps if the data centres are close to one another. And to reduce the latency that you, as a user, experience, the data centres need to be close to the people that are using them. This means that it’s not as easy as simply building a bunch of data centres in the middle of nowhere – they often need to be close to population centres.
AI exacerbates this problem because latency becomes an even greater concern and the sheer computing power the latest models require mean much larger data centres. So what was once relatively hidden from view is starting to encroach on our communities.
Data centres don’t just need land; they can also place real strain on electricity and water systems. Their power needs can require new substations and transmission upgrades that take years and cost a lot of money for regional networks, leaving residents worried about higher bills. They can also consume significant amounts of water through evaporative cooling, which is required to offset heat created from servers and AI chips. Both these resource demands can make securing local community support challenging.
It’s therefore no surprise that resistance for new data centres is on the rise. Data Center Watch (a research company that tracks grassroots opposition to data centres across the US) estimates that $162 billion of US data-centre projects have been blocked or delayed since 2023, underscoring how ‘not in my backyard’ is becoming a material constraint on future growth.
Hyperscalers’ response: making ‘yes’ easier to say
Global technology giants Microsoft and Alphabet (held across many of our funds) are near the front of the queue when it comes to new data centre capacity. Their future growth in cloud services and AI is heavily dependent on the ability to secure new data centre sites.
Microsoft’s Community‑First AI Infrastructure plan is probably the best example of the broad industry response. It’s designed to reduce local backlash by making data centre growth feel less like a burden that communities must absorb and more like a local benefit. Their approach is to tackle the common sources of pushback directly. Microsoft itself has made five commitments that sum up the industry’s response well: they’re aimed at electricity affordability, water resilience, jobs, taxes and community investment.
1) Electricity: ‘We’ll pay our way’
Microsoft’s core objective is that households should not subsidise data centre expansion. It says it will work with utilities and regulators on rate structures designed to cover the full costs of serving its data centres, with the aim of preventing those costs being passed onto residential customers. It also states it will pay for transmission and substation improvements required by its expansion. For example, in the US Midwest it has contracted to add 7.9 GW of new generation – more than double its current consumption there.
2) Water: ‘Use less – and replenish more than we use’
Microsoft commits to a 40% improvement in data-centre water-use intensity by 2030 across its owned fleet. Its ‘closed-loop’ cooling design deployed in places including Wisconsin and Georgia recirculates cooling liquid and avoids using drinking water. Beyond its own sites, Microsoft says it will fund upgrades to local water infrastructure which will support more resilient systems and help utilities maintain stable rates and pressure. It also commits to replenishing more water than it withdraws within the same water districts where its facilities operate.
3) Jobs and skills: ‘Create jobs for local residents’
Data centres typically create thousands of construction jobs and once completed hundreds of operations roles. Also, through workforce partnerships and its Datacenter Academy the company will help train local residents to fill ongoing job vacancies.
4) Taxes: ‘We won’t ask for special breaks’
In an era where some communities worry they’re giving too much away, Microsoft says it won’t ask municipalities to reduce local property tax rates to attract data centres – by paying their fair share this should help strengthen local funding for public services.
5) Community investment: ‘Bring AI benefits locally’
Finally, Microsoft says the communities powering AI should be among the first to benefit, including investments in local AI training (for schools, colleges and libraries) and support for nonprofits.
Your friendly neighbourhood AI
While most hyperscalers are aware of the need to get communities onside, we think Microsoft’s plan should give it a genuine edge. It makes the trade-offs clear and explains how it plans to mitigate or pay for the effects of its development.
If a community can see that a company intends to pay for the grid impacts it creates, invest in local water resilience, expand skills and employment pathways, and contribute fully to local tax bases, the overall proposition becomes easier to support. This plan should give Microsoft three distinct advantages.
Firstly speed: as fewer delays can mean capacity comes online sooner which is valuable in an AI cycle where companies are racing against the clock to innovate. Secondly cost and certainty: by reducing the risk of last-minute opposition this can improve project predictability and reduce costs as data-centre buildouts involve long lead times and complex supply chains. Thirdly reputation and repeatability: if Microsoft can build a track record of being a ‘good neighbour’ across multiple regions it can point to these as useful case studies for subsequent approvals going forward.
Being a good corporate citizen to local communities could be the key to success in the AI race, and this understanding may also prove increasingly valuable as competition for land, power and water intensifies.
