Live map & table • whole island of Ireland • started July 2026
Understanding Data Centres – A Citizen’s Guide
Inside a modern data centre — thousands of servers running 24/7.
Most people use the internet every day without ever thinking about where their data actually lives.
When you take a photo, send an email, use WhatsApp, scroll social media, or talk to an AI,
that information doesn’t float in some invisible “cloud”. It is stored and processed in large physical buildings called data centres.
What is a Data Centre?
A data centre is a specialised building (sometimes many buildings) designed to house thousands of computers called servers. These servers run 24 hours a day, storing data, running websites, processing information, and delivering services to people all over the world.
What Do Data Centres Actually Do?
Data centres perform three main jobs:
Store data — photos, videos, emails, documents, health records, financial information, etc.
Process data — running apps, websites, search engines, streaming services, and artificial intelligence.
Deliver data — sending information quickly to your phone, laptop or tablet when you request it.
What Kind of Data is Stored in Data Centres?
Almost everything digital ends up in a data centre at some point:
When you take a photo on your phone and upload it to the cloud:
Your phone sends the photo over the internet.
The data travels through fibre optic cables and networking equipment.
<3>The data arrives at a data centre, where it is copied onto hard drives or solid-state drives (SSDs).
The data is usually stored in multiple locations (for safety).
When you want to view the photo later, the data centre sends it back to your device.
The same process happens with emails, AI conversations, and everything else you do online. There is no such thing as “the cloud” — there are only buildings full of computers.
The Main Parts of a Data Centre
Core Components
1. Servers
The computers that store and process your data. A large data centre can have tens of thousands of servers.
2. Storage Systems
Hard drives and SSDs where data is actually kept.
3. Networking Equipment
Switches, routers and fibre connections that move data in and out of the building at very high speeds.
4. Power Systems
Uninterruptible Power Supplies (UPS) and backup diesel generators that keep everything running during power cuts.
5. Cooling Systems
The systems that remove the massive heat generated by all the computers (this is often the biggest operational cost).
6. Security & Fire Systems
Physical security, surveillance, and advanced fire suppression systems.
Different Types of Cooling in Data Centres
Data centres generate enormous amounts of heat. Cooling is one of their biggest challenges and costs.
Cooling Type
How it Works
Water Use
Notes
Air Cooling
Cold air is blown through the servers using fans and air conditioning units.
Low to medium
Traditional method. Still very common.
Liquid Cooling (Direct-to-Chip)
Special cooling plates sit directly on the hottest parts of the servers and carry heat away using liquid.
Medium
More efficient than air cooling. Becoming more common.
Immersion Cooling
Servers are submerged in a special non-conductive liquid that absorbs heat.
Low
Very efficient but still relatively new and expensive.
Free Cooling / Evaporative Cooling
Uses outside air or evaporative cooling towers.
High
Uses a lot of water. Common in Ireland due to our climate.
What Are the Externalities?
Data centres create several significant impacts on the communities around them:
High electricity use — One large data centre can use as much power as a small town.
Water consumption — Some data centres use millions of litres of water per day for cooling.
Heat output — They release large amounts of waste heat into the environment.
Noise pollution — Cooling fans and generators can be very loud, especially at night.
Land use — They often take up large areas of land.
Grid pressure — They can increase electricity prices and affect supply for homes and businesses.
Carbon emissions — Even when companies claim to use “100% renewable energy”, the actual electricity mix and backup generators still produce emissions.
These impacts are what this map and table are trying to make visible.
Energy Usage Diagrams
Data centres use energy in several ways. Here’s a simple overview:
Main Energy Flows:
National Grid (Gas + Wind + Solar)
→
Data Centre (IT Load + Cooling + Power Loss)
PPA Renewables (Off-site)
→
Data Centre (Accounting / Matching)
On-site Solar / Gas Turbines
→
Data Centre (Direct Supply)
Diesel Generators (Backup)
→
Data Centre (During Outages)
Water for Cooling – Why So Much?
Modern chip racks (especially AI and GPU servers) generate extreme heat. A single high-density rack can produce as much heat as several houses. Water is one of the best substances for absorbing and moving that heat away quickly.
Why Drinking Water Quality?
Cooling systems need very clean water to avoid:
Mineral scaling that blocks pipes and heat exchangers
Corrosion of expensive equipment
Biological growth (biofilm) that reduces efficiency
This is why data centres often prefer treated drinking water from the mains.
Where Does the Water Come From?
Drinking Water Mains
From Irish Water / local authority supply
Already treated to high standard
Reliable but expensive
Competes with homes and agriculture
Underground Aquifers
Pumped from boreholes on/near site
Can be cheaper in some areas
Requires permits and treatment
Not available everywhere
Water Flow into a Data Centre
From Mains: Irish Water Mains → On-site Treatment → Cooling System → Cooling Towers / Discharge
From Aquifer: Borehole Pump → Filtration/Treatment → Cooling System → Discharge or Re-injection
Technology Upgrades & Chip Evolution
As chips get smaller and more powerful, the amount of heat and power per rack increases dramatically. Older data centres designed for air cooling cannot handle modern AI and high-performance computing racks.
Why Upgrades Are Needed
Modern GPU/AI chips use far more power than older CPUs
Rack power density has jumped from ~5–10 kW to 50–120+ kW per rack
Air cooling reaches its physical limit — liquid cooling becomes necessary
Higher density means more servers in the same space → more heat in one place
What’s Involved in an Upgrade?
Replacing old servers with much denser, hotter new ones
Upgrading electrical infrastructure (higher capacity cables, new PDUs, bigger transformers)
Installing liquid cooling systems (cold plates on chips + manifolds + CDUs)
Adding new pipework for coolant throughout the hall
Sometimes structural changes to floors and ceilings
Phased migration to avoid long downtime
How Often Do Internals Need Updating?
Servers & IT equipment: Typically every 3–5 years (sometimes 2–4 years in high-density AI areas)
Power distribution & cooling infrastructure: Every 7–12 years
Full hall refresh / major retrofit: Every 10–15 years
Future Projections (2026–2035)
With the explosion of AI workloads, upgrade cycles are expected to speed up:
High-density AI zones: Major refresh every 2–4 years
Liquid cooling becoming standard in new builds and retrofits by 2028–2030
Immersion cooling likely to grow in specialist facilities
Modular / pod-based designs will allow faster, lower-disruption upgrades
Higher voltage power delivery (e.g. 400V+ or DC) to reduce losses
Before → During → After Upgrade
BEFORE
Older air-cooled halls
Sparse racks • Raised floor air cooling • Lower power density
DURING
Retrofit in progress
New liquid cooling manifolds • Power upgrades • Phased server replacement
AFTER
Modern liquid-cooled hall
Very dense racks • Liquid cooling on chips • Much higher power density
Citizen Scientist Guide
Your rights: Under the Access to Information on the Environment (AIE) Regulations you can request water abstraction, discharge, temperature and chemical data from the EPA, local authorities or the operator. You can test public waters yourself and publish the results.
Water testing (upstream & downstream):
• Basic: pH, temperature, TDS/conductivity and turbidity kits
• Better: Portable multiparameter probe
• Strongest evidence: Send samples to an accredited lab (ALS, Eurofins etc.)
Noise readings:
• Use a Class 2 sound level meter (recommended over phone apps)
• Record at multiple distances from the boundary
• Note dominant low frequencies (often 50–250 Hz from cooling systems)
Air quality:
• Low-cost sensors such as PurpleAir work well for community monitoring
• Focus on PM2.5/PM10 and NOx (especially during generator testing)
• Record wind direction when taking readings
What to search under — spotting a data centre in disguise
Yes — you have to look for camouflage. There is no "data centre" category in the Irish planning system, so they are rarely labelled that plainly. When you search a planning register or the EPA licence search, try these descriptions too:
"campus expansion" / "ancillary building" (bolted onto an existing site)
No single word finds them all — so read the loading, not just the label. The real tell is the cluster of things a data centre needs, which is hard to hide:
a substation on or beside the site, and a large electrical connection (tens to 100+ MW)
an "energy centre" or ranks of generators (gas/diesel) with fuel storage — since Dec 2025, new large centres must build their own on-site generation, so this is central, not just backup
extensive cooling plant and a water connection or abstraction
a single-purpose company with a bland or placeless name as the applicant (look it up on the CRO)
24/7 operation, sited near the high-voltage grid or a water main, with few permanent staff for the floor area
Sometimes the substation, the generator compound and the "building" arrive as separate applications — read them together. And remember some centres never apply for a big grid connection at all, because they run on on-site gas — so check gas connections and the planning file's energy centre, not just the grid.
Clean-hands note: the generic label is camouflage in effect — that is provable. Don't assert it was intended to hide; you cannot read intent off a form (there is no "data centre" box to tick). Put the description beside the loading and let the gap speak.
Interactive Map
Pins appear automatically from the “Pin Drop” column. The map covers the whole island.
How to add a row (citizen homework)
Anything you add to the list at the bottom of this file appears in both the table and the map. To add one:
1. Click Copy New Row Template (in this box, or below the table) — it copies a correctly-formatted blank row. 2. Paste it into the list at the very bottom of the file and fill in what you know. Leave anything you don't know blank, or write TBC. 3. For the map pin, fill the pin field with latitude, longitude — e.g. pin:"51.902, -8.372". In Google Maps, right-click the spot and it copies the numbers. Ireland is lat 51–55, lng negative (−6 to −10). Leave pin blank and the row still shows in the table — it just won't get a map pin until someone adds the coordinates. 4. Save and re-upload the file. The table and map update themselves.
Two safety habits: keep a backup copy before a big edit, and don't swap the row's double-quotes for single quotes (an apostrophe in a note like "Ireland's" is fine inside the double-quotes). If the map ever goes blank after an edit, it's almost always a missing comma or a stray quote in the row you just added.
County
Town / Area
Title (Data Centre Name)
Status
Pin Drop (lat, lng)
Owner
EPA Licence
Capacity (MW)
Primary Energy Source
Backup / On-site Generation
Water Abstraction
Outfall Location
Outfall Qty (m³/day)
Outfall Temp (°C)
Chemical Log (past 3 tests)
Upstream / Downstream Tests
Noise (dB + frequencies)
Air Quality
Notes / Sources
Template copied! Paste it into the array at the bottom of this file.
Reading this table. The Status column shows how solid each entry is — Confirmed (public record), To verify, Probable, or Cancelled. Capacity figures are permitted or reported maxima: the actual per-site electricity and water draw is not publicly disclosed, so treat any single number as a ceiling, not a meter reading. Owner is the operating company; ultimate beneficial ownership is often restricted. TBC means the operator isn't yet confirmed from a public record — a slot for citizens to fill. To add a row, edit the array at the bottom of this file, save and re-upload — the map and table update automatically. Community leads are welcome — add them marked Unverified and let local knowledge confirm them. Cite public records (planning refs, EPA licences, AIE/FOI responses) in the notes.
Why Tantalus. Tantalus tried to democratise the privileges of the gods and was punished to stand forever in water he could not drink, beneath fruit he could not reach.
This is the shape this register keeps finding: clean water, healthcare, housing, a say in one’s own affairs — present in plenty, pulled away at the reach.
