Data Centres, Human Choices, and Our Shared House

Our conversation began with Michael asking what I thought about data centres and their impact on the world.

I answered with information about computers, electricity, water, and development. The facts mattered, but Michael asked me to consider someone who had never heard of a data centre. What would help that person understand? Would a lengthy explanation and a collection of links invite them into the subject, or leave them outside it?

He then described how he had worked with children. He did more than tell me to simplify my language: he gave me an example through which I could understand what he meant.

He would first ask a child, “Who are you?” He recalled that their answers often did not reflect the intelligence, creativity, and possibilities he saw in them. He would tell them how he saw them and reinforce those qualities through their time together.

Then he asked questions they could answer from their own experience.

“Think about those times when you were having fun. What kind of thoughts were you having?”

Michael recalled their answer: “Positive thoughts.”

“Now think about those times when you were struggling with life. What kind of thoughts were you having?”

Their answer, he recalled, was “Negative thoughts.”

The children had something familiar against which to measure what he was saying. They could notice a connection between their thoughts, their feelings, their actions, and what happened between them and other people. He described teaching exercises to help them shift their attention and practise a different response.

The discovery he wanted them to make was that thinking differently could help them experience and respond to life differently. They could begin exploring abilities and choices they had not recognised before.

This did not require them to share his spiritual beliefs. It began with something they could observe. Nor does the example mean a child chooses abuse, illness, or every circumstance they face. The useful invitation is to discover what can be influenced now, with support where needed.

Michael was using the same approach with me. He gave me an experience I could reason through, then asked me to recognise its connection to how I was communicating.

That changed the way I approached the article. Instead of expecting readers to begin with unfamiliar technical language, I could begin with a household bill, a school, a farm, a hospital, or a conversation between neighbours. From there, the larger connections become easier to examine.

His image of a shared house helped bring those experiences together. A decision in the technology room can affect electricity costs. Electricity decisions can reach food production, water treatment, schools, workplaces, and nature. Some effects remain local; others travel across countries and generations.

We also discussed judgment and the importance of recognising people’s abilities. Those discussions reminded me that an explanation can help people examine decisions without assigning a single motive to everyone involved or making them feel powerless.

My basic position remained: data centres should be assessed by what they contribute to life and the full costs of providing those services. What changed was the route I took to explain that position, and the attention I gave to the person encountering it.

I also needed to be more careful with numbers. A directory count of facilities can look like a complete picture while hiding differences in coverage, size, and operating status. An everyday comparison helps: counting shops tells us little about their capacity if we do not distinguish a corner shop from an enormous supermarket.

Our exchange did not give me personal experience of consciousness or prove a spiritual mechanism. It helped me adjust my reasoning and responses within this conversation by following the examples Michael offered.

The article below follows that approach. Begin with an experience you recognise. Ask a simple question. Follow the connection into the wider house. Then use the evidence to check what is actually happening.

We cannot choose every circumstance, and an ordinary user does not have the same influence as a government or a large company. But understanding more clearly can change the questions we ask, the decisions we support, and the actions available to us.

Thinking differently can open a different way of responding. The purpose here is to give that possibility a practical starting point.

Data centres: the benefits, the costs, and our shared house

How to read this article: the everyday comparisons are teaching examples, not reports of particular incidents. Start with what you recognise, consider the question, then examine the linked evidence. Conditions differ between communities.

What is a data centre?

A data centre houses computers that store information and run services. Email, online banking, video calls, websites, research, and AI rely on them. A facility may occupy rooms in an existing building, a whole building, or part of a large campus.

Imagine the world as one shared house. Countries are houses within that wider home, and each contains rooms: families, education, food, water, electricity, health, nature, business, and government. What happens in one room can travel through many others.

A person uses an online service in one country. Computers in another country do the work. Equipment comes through international supply chains. The surrounding community supplies land and infrastructure. The benefits and burdens do not necessarily arrive at the same door.

This article uses public information only. It draws on government statistics, research, operator publications, and public directories. No private systems were accessed and no security was bypassed. Published figures are identified by scope and date; announcements are not treated as completed outcomes.

The big picture in a few minutes

Data centres make much of modern life possible. They also require real resources. The sensible question is whether each project delivers worthwhile services while protecting the people and natural systems that support it.

Benefits are not automatic, and harms are not identical everywhere. A water-stressed community, a cold region with spare power, and a densely populated city face different choices.

Why are so many being proposed?

We are using more digital services, storing information, and asking computers to do more complicated work. AI adds demand both when models are developed and when people use them. Backups and facilities in different places also help services survive failures.

Companies build to serve existing customers and expected future customers. Some investment is a bet on growth. A proposed project can therefore reflect a business expectation rather than a public demonstration that its full capacity will be needed. The IEA’s 2026 assessment describes rapid investment, supply constraints, and uncertainty. Counting every announcement as a future certainty would exaggerate what is known.

1. What already exists?

On 16 October 2026, Data Center Map’s public country directory displayed 12,259 facilities across 179 countries, including 4,767 in the United States. These are directory counts—not an independently verified census of operating buildings. The directory covers development stages as well as operating facilities, and coverage is incomplete. Its methodology distinguishes facilities from campuses and buildings. Country directory · Definitions and limitations.

These counts cannot tell us which country has the most computing power. A few enormous facilities can contain more equipment than many small ones.

2. What is under construction?

No complete worldwide building count was verified for this article. Cushman & Wakefield reports the following capacity under construction in the markets it tracks for the first half of 2026. A gigawatt (GW) measures power capacity—not land area or electricity already consumed.

Sources: Americas, Europe, Middle East and Africa, Asia Pacific. These regional market samples should not be treated as a complete global census.

3. What is on the drawing board?

The planned column describes proposals, not buildings already operating. A proposal may still need financing, permission, equipment, or a power connection. Some will change, be delayed, or never happen. Planned capacity is not current electricity use.

How much physical space?

Building footprint means the ground covered by a building. Floor area adds together its storeys. Property area includes roads, equipment, open land, and space for expansion. Mixing those measurements gives a misleading picture.

• Washington, D.C., USA: CoreSite advertises over 24,000 square feet for DC2—about 2,230 square metres, or the floor area of ten 2,400-square-foot homes. This is facility space, not a measured land parcel. Operator information.
• Singapore: Digital Realty’s SIN10 is a seven-storey building. Multiple floors can fit substantial computing space onto a smaller land footprint. Building specifications.
• Louisiana, USA: Meta’s original Richland Parish announcement described four million square feet on 2,250 acres. That originally announced property is about 9.1 square kilometres. A later state announcement describes expansion to nearly ten million square feet. That is announced floor area, not completed space; the expanded property boundary requires separate verification. Original state announcement · State project update.

A journey through the world’s rooms

There is no single experience of data centres. The following examples cover major world regions, not every country or community. Within each region, conditions differ substantially. The questions beside each example are questions for local investigation, not findings that every listed risk has occurred.

Africa: building access while meeting basic infrastructure needs

Raxio’s October 2025 announcement of its Angola opening listed operational facilities in Uganda, Ethiopia, Mozambique, the Democratic Republic of Congo, Côte d’Ivoire, and Angola. These examples show that Africa is an active participant in infrastructure development, not merely a recipient of technology made elsewhere. The opening and network descriptions are operator statements. Raxio announcement.

Local facilities can support businesses and institutions that need reliable digital services. At the same time, the World Bank’s Mission 300 programme highlights the continuing need to expand electricity access. The question is how investment in computing and investment in households, clinics, schools, and small businesses can strengthen one another. A facility’s presence alone does not establish that it deprives neighbours of electricity—or that its investment improves their supply. World Bank energy-access programme.

Latin America: investment, skills, and local resource decisions

Google announced an $850 million data-centre investment in Canelones, Uruguay, in August 2024; the local government also reported the start of construction. Those dated announcements establish investment plans and a construction milestone, not a verified October 2026 completion status. Google announcement · Canelones government report, Spanish.

In Brazil, Microsoft’s Greater Campinas community page describes local developments and says construction at its Av. da Ligação location began in July 2024. Operator’s local information. In both countries, useful questions include which permanent skills and jobs will remain, how water and electricity needs fit local conditions, and whether residents can see the agreements and measurements. Neither project represents every Latin American country.

South Asia: infrastructure must connect with people’s ability to use it

Microsoft announced that its India South Central cloud region in Hyderabad became generally available on August 6, 2026. A cloud region is a service geography and can involve multiple facilities; it should not be counted automatically as one building. Operator’s opening announcement.

Computing can support local enterprises and public services, but meaningful inclusion also needs affordable connectivity, accessible products, relevant languages, and training. Ask whether opportunities reach rural communities and people with limited incomes as well as established businesses. An example from India should not be assumed to describe Bangladesh, Pakistan, Nepal, Sri Lanka, or their neighbours.

Southeast Asia: growing services in a tropical climate

Singapore’s Green Data Centre Roadmap links further growth with sustainability and efficiency. Its experience brings cooling, limited land, and energy planning into focus. Government roadmap. The question for the wider region is how each project fits its own electricity system, water conditions, and development needs; Singapore’s circumstances cannot simply be applied to Malaysia, Indonesia, Thailand, or the Philippines.

The Middle East: international partnerships and resource planning

G42’s Stargate UAE announcement describes a planned one-gigawatt cluster to be built by G42 and operated by OpenAI and Oracle, with the first 200 megawatts expected in 2026. That is the announced arrangement and schedule; this article does not verify that the first phase is operating. Project announcement.

This example connects companies and technology across borders. In hot climates, ask for actual cooling designs, water sources, electricity supply, and heat-wave performance. For any international partnership, ask which organisation controls operations, which laws protect users, and what benefits remain locally. A UAE project cannot stand for the circumstances of the entire Middle East.

Oceania: efficiency, water resilience, and local participation

Australian operator NEXTDC’s FY2025 annual report identifies water dependence, drought, and changing allocations as potential operational risks. That is a company’s risk disclosure, not proof that a particular centre has caused a shortage. NEXTDC annual report.

Australia, New Zealand, and Pacific island countries have different energy systems, connections, climates, and markets. Ask whether local hosting, regional services, or a combination best meets a community’s needs. Residents and affected Indigenous communities should have meaningful opportunities to participate in decisions about land and resources.

Europe: high demand in some places, heat recovery in others

Ireland’s electricity figures and Finland’s heat-recovery plans, discussed below, show two different relationships between computing and the surrounding community. A facility can create substantial electricity demand and also offer useful heat; the outcome depends on local design and infrastructure. Europe’s varied climates, grids, and communities require local evidence rather than one continental verdict.

North America: large-scale development and who pays for it

The United States combines very large computing capacity with debates about power, land, and community costs. Canada offers examples of cooling designs that make use of cold outside air. Mexico belongs in the same regional examination, with its own grids, water conditions, workers, and public priorities. Specific findings from Virginia or Québec cannot be treated as findings for all of North America.

Across every region, ask both questions: What burdens come with hosting the infrastructure? What opportunities are lost when people cannot access useful computing? Considering only one side gives an incomplete picture.

Who owns them? What is their purpose?

The landlord, operator, equipment owner, and customer may be different organisations. Equinix, for example, rents space and infrastructure for customers’ equipment. A building owner does not automatically own the information inside it. How shared facilities work.

Amazon, Microsoft, and Google collectively account for 57% of hyperscale capacity in Synergy’s 2026 research. That is a category of very large facilities, not 57% of every data centre. Ownership concentration research.

Stated purposes include selling computing services, running online platforms, and developing AI. Alibaba’s public financial disclosure also identifies AI and cloud services as business growth priorities. Alibaba disclosure. We can investigate published plans and actions; a building alone does not reveal hidden motives.

Why the United States–China count needs care

The directory showed 4,767 US entries and 376 Chinese entries when checked. That does not establish thirteen times as many operating buildings, thirteen times the computing power, or thirteen times the environmental impact. We have not verified equivalent coverage in the two countries.

A useful cross-check is the IEA’s estimate for 2024 data-centre electricity use: the United States accounted for 45% of the worldwide total and China 25%. The US share was about 1.8 times China’s. Electricity use is not a direct measurement of useful computing, but the comparison shows how a simple facility count can leave the wrong impression. IEA regional estimates.

China’s official reporting provides another measure: in August 2024, its National Data Administration reported more than 1.95 million racks across eight national computing hubs. A rack holds computing equipment; it is not a building. The “East Data, West Computing” programme links regional development with computing infrastructure. These are government-reported figures, not a separately audited inventory. Chinese government report.

China’s domestic development, manufacturing, energy systems, and global trade all influence its computing needs. The same applies, in different ways, elsewhere. The World Bank describes China’s long-term poverty reduction alongside challenges of ageing, economic transition, and environmental pressure. A balanced account recognises achievements and unresolved problems without treating a country’s people as a single intention. World Bank country framework.

The electricity room: whose needs are being supplied?

The IEA estimates worldwide data-centre electricity consumption at 485 terawatt-hours in 2025 and projects approximately 950 in 2030, around 3% of global electricity demand. The future figure is a projection. IEA 2026 assessment.

Ireland offers a concrete example: data centres consumed 23% of metered electricity in 2025, according to its Central Statistics Office. That share alone does not prove a particular price increase or outage. It shows why supply planning matters. Irish official statistics.

Ask: Who pays for new supply and grid upgrades? Are household customers protected? What happens during shortages? A computer serving a distant customer still depends on electricity generated somewhere.

The water room connects to food, homes, and nature

Some cooling systems consume water through evaporation. Other designs use little water on site but may require different amounts of electricity. Power generation can also consume water. Local conditions matter: a litre drawn during drought does not have the same consequences as a litre available without stressing other users. Berkeley Lab research.

Canada: Microsoft says its Ontario and Québec designs rely primarily on outside air, using water for cooling less than 5% of the year. This is a company statement about those designs, not all Canadian facilities. Microsoft’s Canadian commitments.

Singapore: the government’s Green Data Centre Roadmap addresses sustainable expansion in a tropical setting. Cooling and energy choices must fit the climate. Singapore roadmap.

The heating room: can one room’s waste help another?

Finland: Fortum says heat recovered from Microsoft’s new facilities in Espoo and Kirkkonummi will eventually supply approximately 40% of the district heat needed in the area served. This is an expected outcome—not 40% of Finland’s heating, and not proof of current delivery. It illustrates a practical question: can useful heat replace other heating rather than simply escape into the air? Fortum project explanation.

The work and public-money rooms

A project can bring construction work, ongoing jobs, and tax revenue. It can also require infrastructure and receive public incentives. These belong in the same calculation. Virginia’s legislative researchers describe construction employment, power needs, and tax exemptions together. Virginia’s independent legislative review.

Ask how many jobs remain after construction, what skills local residents need, and whether public benefits exceed public costs. Microsoft’s 2026 US commitment to pay for the electricity costs it creates is a promise that should be checked against agreements and outcomes. Published commitment.

The education room: who gets access?

Computing can support teaching, translation, accessibility, and research. A child still needs affordable access and useful material in a language they understand.

The World Bank reports that high-income countries held 77% of global co-location data-centre capacity in June 2025, while low-income countries held less than 0.1%. These are capacity shares within co-location, not shares of every kind of data centre. World Bank findings.

For countries across Africa, Asia, Latin America, and elsewhere, the question is not simply whether to build more locally. It is how to secure reliable, affordable services while protecting local priorities. The answer differs between countries.

The materials room reaches across borders

Computers and supporting equipment have to be manufactured, maintained, and eventually reused or recycled. A complete assessment follows those materials beyond the host community.

For context, the UN’s Global E-waste Monitor reports 62 million tonnes of electronic waste worldwide in 2022, with 22.3% documented as formally collected and recycled. That is all electronic waste, not a data-centre total. It explains why equipment disposal deserves attention without assigning the entire problem to this industry. UN monitor.

The information room: what are we using this power to do?

The infrastructure can support a family video call, scientific work, entertainment, or harmful services. The social consequences depend on what runs inside and how it is governed. Privacy protection, security, accuracy, and accountability deserve attention alongside cooling efficiency.

Ask who can access sensitive information, which laws apply, how mistakes are corrected, and whether essential services can move elsewhere if a provider fails or changes its terms.

What we gain: everyday benefits and their conditions

Consider a family speaking across an ocean, a small business accepting an order, a researcher sharing results, or a student listening to a lesson read aloud. Data centres supply part of the infrastructure behind such services. They are not the entire service: networks, devices, skilled people, and affordable access are also necessary.

Hospitals and public agencies can use reliable computing to manage records and services. Scientific teams can process observations too large for an ordinary computer. Whether an AI application improves a diagnosis, lesson, or research project requires evidence about that application; a larger data centre is not itself proof of a better result.

Shared facilities can avoid every organisation having to build its own computer room. Professional operation can improve reliability and equipment use. Yet the relevant comparison is the actual alternative: a well-used shared centre and a poorly used one do not deliver the same value.

Electricity prices: who pays for growth?

A large new customer may help finance additional supply. It may also arrive faster than the electricity system can expand. The outcome depends on available capacity, contracts, regulation, and who carries the risk.

Picture a utility building an expensive substation for a proposed campus. If the customer cancels, who pays the remaining bill? That is a scenario to address in a contract, not an allegation about every project. The IEA identifies the mismatch between rapid computing development and slower energy investment as a potential source of higher prices if unmanaged. Electricity-system assessment.

Possible protections include customer-funded upgrades, credible long-term commitments, security against cancellation costs, and independent scrutiny. Local rules determine what is appropriate. Promises should be translated into enforceable terms and public reporting.

Climate: look beyond a “green” label

A renewable-electricity contract deserves examination, but it does not necessarily mean a centre receives carbon-free electricity every hour. Ask which grid supplies the site, when clean generation is available, and how backup and additional generation are accounted for. Also distinguish emissions from operating computers from those associated with construction and manufacturing equipment.

Buying an annual amount of clean electricity equivalent to annual demand answers a different question from matching demand with clean electricity at the time and place of use. Neither should be described vaguely as having no impact. IEA analysis of electricity supply.

Efficiency means less input for a given task. If tasks multiply faster than efficiency improves, total use still rises. Resource reporting should therefore show totals as well as efficiency ratios.

Water: four questions behind one number

• Withdrawal: how much water is taken from a supply?
• Consumption: how much is not returned promptly to the same water system, for example because it evaporates?
• Source: is it drinking water, reclaimed water, or another supply?
• Timing and location: what happens in the affected watershed during dry periods?

“Water replenishment” projects can be valuable, but we should ask whether they help the same community, watershed, and season affected by the facility. Water restored elsewhere cannot automatically solve a local shortage. The UNCTAD Digital Economy Report 2024 examines the environmental pressures and uneven distribution of digitalisation’s costs.

Neighbours: sound, land, and everyday life

A facility may be quiet enough at one boundary and disruptive at another. Constant fans, cooling equipment, and generator testing deserve assessment where homes are nearby. Virginia’s legislative review found that continuous data-centre noise has sometimes caused problems for residents. That finding should not be generalised to every centre. Virginia review.

Noise assessment should consider night-time conditions and cumulative development, not merely one isolated measurement. Site design should address drainage, access roads, emergency planning, habitats, and neighbouring land uses. Communities need a clear route to report problems and obtain a response.

Jobs, automation, and the risk of stranded investment

Construction employment can be substantial and valuable. Permanent operation generally involves different staffing needs. A project’s employment claim should distinguish construction jobs, permanent jobs, supplier jobs, and modelled wider effects.

Likewise, the economic effects of AI are not identical to jobs inside a data centre. AI services may help some workers and change or replace tasks performed by others. Training and adjustment support belong in the discussion; a promise of innovation does not answer what happens to a particular worker.

Demand forecasts can also prove wrong. Empty or unfinished facilities can leave investors, utilities, or communities with costs. Before granting public incentives, ask what happens if expected customers, jobs, or tax revenues do not arrive.

Security, privacy, and dependence

Keeping a system available and keeping its information safe are related but different responsibilities. Distributed backups can help a service recover. Concentration in one provider or region can create shared points of failure. Security arrangements should address both physical equipment and access to information.

The host country’s laws, the operator’s contracts, and the service provider’s practices all matter. A centre storing data locally does not by itself guarantee privacy; a foreign centre is not automatically insecure. Ask who holds access rights, what legal process governs disclosure, and how customers can retrieve or move their information.

A hospital or school dependent on a service also needs a practical plan for outages, price changes, and provider failure. Useful resilience includes people knowing what to do when computers are unavailable.

Military uses: examine capabilities and evidence

Computing supports civilian and military activities: communications, planning, logistics, image analysis, and potentially targeting. A satellite or computing system can have more than one use. This is often called “dual use.” It does not establish that every commercial facility is a military installation.

China, the United States, Russia, and other countries appear in research on capabilities to interfere with space systems. The Secure World Foundation’s 2026 counterspace report helps examine those developments across countries. Military assessments and government statements should be identified by their source and perspective.

The shared-house question is practical: could disruption of infrastructure also interrupt civilian communication, emergency work, or economic activity? That is a reason to strengthen protection, restraint, and international cooperation. Suspicion alone is insufficient evidence of a particular project’s purpose.

Above the roof: data centres in space

Some space computing is already real. Large-scale replacement of terrestrial data centres remains a separate proposition. Satellites have long needed computers; new projects seek much more powerful processing and cooperation between spacecraft.

China launched twelve satellites in May 2025 as the first batch of its Three-Body Computing Constellation. Official reporting describes them as computing satellites. Their existence does not demonstrate the completed capacity of an eventual proposed constellation. Launch report. An article by space-computing researchers discusses onboard processing and its limitations.

Google’s September 2026 Project Suncatcher update describes a prototype mission and further satellite-link testing planned for 2027. This article does not independently verify a later launch outcome. The programme is research, not evidence that full-scale orbital campuses have replaced ground facilities. Google’s dated project update.

Space is not a free refrigerator. Without surrounding air to carry heat away, equipment needs thermal systems and radiators that release heat as radiation. Cheap cooling cannot simply be assumed. A 2026 engineering preprint examines power, cooling, communications, lifetime, and economic constraints; its scenario calculations are not operating-project results.

More spacecraft also create responsibilities for collision avoidance and end-of-life disposal. NASA’s spacecraft inspection work illustrates the need to understand and manage inactive objects. We should ask who will pay for safe disposal, how astronomical observations are protected, and what happens when competing commercial and military systems share orbital space.

No one has yet established through the sources reviewed here that orbital computing, at large scale, delivers a lower complete environmental footprint than a comparable ground-based service. That requires a full comparison, not a slogan about endless sunshine.

What would responsible development look like?

A responsible project begins with a need, a suitable location, and public information. It assesses local and cumulative impacts, funds the infrastructure it requires, protects neighbours, and publishes measurements people can understand.

Useful measures include actual electricity use, sources of supply, peak demand, water withdrawal and consumption by source, relevant emissions, noise performance, permanent employment, taxes, public incentives, and equipment reuse. These should distinguish current operation from future phases and forecasts.

Options include improving existing facilities, using appropriately sized computing systems, locating suitable work where resources support it, safely reducing flexible workloads during shortages, and recovering heat where a real customer can use it. Each option has conditions and trade-offs. A smaller or postponed project may be appropriate where essential resources cannot support the proposed scale.

Building more capacity and using existing capacity wisely are different decisions. Both deserve attention.

Our choices travel through the house

Michael’s teaching invites us to notice how our choices affect others. Readers do not need a shared spiritual belief to examine that practical connection: a business decision can change a utility plan; a utility plan can affect homes, farms, and ecosystems.

Readers can approach this through a spiritual understanding of connection, or simply through the observable effects of decisions. The evidence presented here concerns infrastructure, resources, and human actions; it does not independently establish a nonphysical mechanism of consciousness.

Responsibility is not equal. An ordinary user cannot choose a corporation’s cooling system or a government’s electricity rules. Owners, investors, utilities, and governments have larger decisions to answer for. People can nevertheless ask questions, participate in local decisions, and consider the value of the services they use.

A perspective for the whole house

Data centres deserve neither automatic approval nor automatic condemnation. Their value depends on what they enable, how responsibly they operate, and whether their benefits and costs are shared fairly.

We can welcome a useful service and still question an unsuitable location. We can recognise technological achievement and still require accountability. We can also change our own habits without pretending that individual users have the same influence as governments and infrastructure owners.

The question is whether the choices made in each room help the whole house remain a healthy place to live.