1

Module 1  ·  Track 1: Technical Foundation

Why Digital Sustainability
Matters Now

What IT actually consumes, the scale most people still underestimate, the waste it creates, and where the operational work begins.

Duration32–38 minutes
TrackTechnical Foundation

What you will take from this module

The era of digital abundance is ending.

For two decades, IT strategy was built on an assumption of abundance. Unlimited storage. Elastic compute. Cheap experimentation. Clean up later, optimise later, account for it later.

That abstraction helped organisations move quickly. It also normalised waste. When something feels consequence-free, people stop asking whether it needs to happen at all.

We are now in a different era, and not because of new ambition. Because of physical limits that have caught up with the assumption. Power availability. Water availability. Material and manufacturing pressures. Capacity lead times. Supply chain fragility. Regulatory pressure. These are no longer abstract sustainability issues. They are technology management issues.

Before any of the rest of this course is useful, two things have to land. What IT actually does to the planet. And how big it has become. We take each in turn, then return to what that means operationally.

What IT actually does to the planet.

Most sustainability conversations in IT collapse into a carbon-only frame. Carbon matters. It is also incomplete. The full footprint runs across six categories. A credible programme can name all six and explain how its own estate touches each.

Impact 01

Carbon and energy

Operational electricity used by servers, networks, devices, and cooling. The driver is grid intensity multiplied by demand. Visible on energy bills, but the real footprint depends on where and when the power is drawn.

Impact 02

Water

Evaporative cooling at large data centres. Closed-loop chillers in semiconductor fabrication. A single hyperscale facility can use water comparable to a small town. Often unmeasured, often excluded from disclosure, often material in water-stressed regions.

Impact 03

Materials and minerals

Copper, aluminium, silicon, rare earth elements, lithium, cobalt. Mined, refined, transported, assembled. A modern smartphone contains over seventy elements drawn from multiple continents. The footprint is locked in long before the device powers on.

Impact 04

E-waste

Over 50 million tonnes generated globally each year. The fastest-growing waste stream on the planet. A material share is processed informally, often in conditions that recover little of the value and damage local environments. Refresh culture is the upstream cause.

Impact 05

Social and supply chain

Labour conditions in mining and assembly. Conflict minerals routes. Informal recycling exposure. The social dimension is not optional alongside the . Increasingly it is the dimension regulators, investors and customers ask about first.

Impact 06

Land and heat

Data centre siting concentrates power, cooling and land demand on local grids and communities. Waste heat is rarely reused. Where capacity clusters, the local : on planning, on water, on , becomes the conversation, not the carbon number.

Why this taxonomy matters

If your sustainability story only talks about carbon and electricity, it is incomplete before the first challenge question lands. Each of these six is a different conversation, with different evidence, different owners, and different operational levers. The course returns to all six across later modules.

The scale most people still underestimate.

The most common defensive reflex inside IT is to assume the estate is too small to matter. At organisational scale that argument collapses quickly. Globally, it does not survive the numbers.

The global digital footprint: headline figures

2–4%
of global GHG emissions from ICT, comparable to the aviation sector
~2%
of global electricity consumed by data centres, and rising sharply with AI demand
50m+
tonnes of e-waste generated globally each year. The fastest-growing waste stream
10×
approximate energy for an AI query versus a standard web search

The headlines obscure where the load actually concentrates. National-level numbers reveal the exposure better than global averages.

Concentration: what the averages hide

~20%
of Ireland's national electricity consumption is now drawn by data centres
60–80%
of a laptop's lifetime carbon footprint sits in manufacture, before it reaches the user
70+
elements from multiple continents inside a typical smartphone
1 town
water draw equivalent to a small town for a single hyperscale data centre cooling load

"We're too small to matter" does not survive contact with the baseline

A large bank, retailer, hospital trust or government body running hundreds of thousands of devices, multiple data centres and a significant cloud estate is not too small to matter. The argument only sounds reasonable until somebody builds the baseline. After that, it stops sounding reasonable to anyone in the room.

Digital systems are physical systems.

The single mindset shift that the rest of this course rests on is simple to state and uncomfortable to apply.

Every cloud service runs on physical infrastructure. Every server consumes electricity. Many data centres consume water. Every laptop is mined, manufactured, transported, used, and eventually disposed of. Every dataset sits on physical storage. Every network transaction relies on physical infrastructure somewhere.

Cloud does not remove infrastructure. It moves it. Software does not remove physical consequences. It shapes them. This is not activism. It is basic physics and supply chain reality, applied to the systems we all use every day.

The line you carry into every later module

If digital is physical, then digital consumption is physical consumption. Wasted compute is wasted electricity, wasted water, wasted minerals, and wasted manufacturing capacity. That framing is the through-line for the rest of the course.

Five categories of digital waste.

If reporting is the wrong place to start, and it usually is, waste is the right one. It is where you build credibility fast, because outcomes are tangible. Every category below is simultaneously a cost problem and a sustainability problem.

Idle Compute

Instances left running with no meaningful workload. VMs spun up for testing and never terminated. Dev environments left on overnight. Systems running at 5–10% CPU utilisation with nobody noticing, or nobody accountable.

VMs provisioned for a project that completed, still running
Dev and staging environments with no auto-shutdown policy
Systems at 5–10% utilisation that would righsize to a fraction of the cost
Untagged cloud resources with no owner and no termination date
£2–8k / VM / year in cloud run cost  ·  25–40% of VMs at <5% CPU in a typical estate
25–40%
of VMs in a typical enterprise estate running at less than 5% CPU utilisation
30 days
rightsizing and auto-shutdown typically recoverable within 30 days

Storage Waste

Data retained indefinitely "just in case." Logs held for years with no access. Multiple uncompressed copies. Triple redundancy on non-critical assets. Dark data, stored but never used, is the fastest-growing category in most enterprise estates.

Log data retained for years beyond any compliance requirement
Multiple full copies of the same dataset in different systems
Production databases replicated to dev environments and never cleaned
S3 buckets from deprecated services still paying monthly storage fees
£18–35 / TB / month cloud object storage  ·  Dark data is one of the fastest-remediable waste categories
60–70%
of stored enterprise data never accessed after initial creation
ROT
Redundant, Obsolete, Trivial. The standard classification for storage rationalisation

Application Waste

Duplicate platforms. Zombie applications nobody uses but IT still pays to run. Portfolio sprawl built up over years because rationalisation is politically difficult. Every unused application has a support cost, a licence cost, a security exposure, and an energy draw.

Legacy CRM system running in parallel with the new platform, "just in case"
Applications with zero active users for 6+ months still being patched and maintained
Three different project management tools that nobody chose to consolidate
Custom-built tools whose original developer left three years ago
£40–120k / app / year in run and licence cost  ·  Portfolio rationalisation recovers 10–20% of app spend
15–30%
of the average enterprise application portfolio is marginal or effectively unused
TIME
Tolerate, Invest, Migrate, Eliminate. The portfolio rationalisation model covered in Module 6

End-User Waste

Devices replaced on fixed procurement cycles regardless of condition. Peripheral sprawl: two monitors by default, docking stations issued as standard, cables and chargers treated as consumables. Laptops shipped by air because onboarding was planned at the last minute.

3-year mandatory refresh cycle applied regardless of device condition
Two monitors issued by default, whether the role requires them or not
Docking stations accumulating in storage rooms, never recovered
Laptops air-freighted due to poor onboarding lead-time planning
~80% of a laptop's lifetime carbon footprint is manufacturing. Extending useful life is the primary lever
~80%
of a laptop's lifetime environmental footprint occurs before it reaches your desk
+1 year
extending average device lifespan by one year reduces end-user embodied carbon by ~20%

AI Consumption

Heavyweight model calls for queries that don't need them. No caching strategy. No tiered inference. Prompt inefficiency at scale: verbose prompts sent thousands of times daily by users who never think about the compute behind the response. AI is the fastest-growing ungoverned consumption category in enterprise IT.

Frontier model used as default for writing short emails, 10–50× over-specified
Long context windows sent with entire documents when only a paragraph was needed
No caching for repeated queries that return identical answers
AI consumption : no visibility of tokens per user, cost per call
10× more energy per AI query than a standard web search  ·  Governance, not restriction, is the fix
10×
approximate energy differential between a frontier AI query and a standard web search
Ungoverned
most enterprise AI inference is currently ungoverned: no usage tracking, no model tiering, no demand management

The operational case for GreenOps

"If you're consuming compute, storage, or network that creates no value, you're paying for it twice: once on the invoice, and once in the carbon account."

GreenOps is a discipline that engineers, finance, procurement, and leadership should all care about equally.

The GreenOps action model.

Most organisations that describe themselves as having a sustainability programme are, on closer inspection, running a reporting function. That is a starting point. It is not a destination.

1
Reporting Tells you what happened. The rear-view mirror.

Necessary for disclosure, compliance, and setting baselines. Not sufficient for operational improvement. Most organisations are at this level, and do not always know it.

If this is all you have, you have a communications function, not a sustainability programme. The output is data about the past. The question it cannot answer is: what are we doing about it?

Symptom: metrics exist, dashboards exist, annual reports published, but operational decisions are unchanged
2
Optimisation Doing the same thing with less.

Rightsizing, improving utilisation, reducing facility overhead, extending device lifespans. This is where most operational GreenOps effort sits, and where the most immediate wins are found.

Critical caveat: optimisation does not guarantee absolute reduction. A business growing at 20% per year while improving efficiency at 10% is still growing its footprint. This is where the rebound effect lives (covered in Module 5).

Symptom: efficiency metrics improve, but total emissions stay flat or rise as demand grows alongside
3
Avoidance Designing demand out before it exists.

Consolidating duplicate platforms. Retiring unused applications. Designing services that require less compute. The highest-leverage decisions live here, because you are preventing waste, not managing it after the fact.

This is shift-left in practice. It requires earlier intervention than most governance models currently allow. Portfolio decisions, architecture choices, and data minimisation policies are all avoidance mechanisms.

Most organisations are at Level 1. The goal is to know your position honestly, and identify what is blocking the next step.

Where is your organisation right now?

Honest self-placement is more valuable than optimistic placement. Most organisations sit at Level 1 or between 1 and 2. That is normal. The question is: what is blocking the next step?

Knowledge Check · Module 1 · Q1

Which of the following best describes the environmental impact of enterprise IT?

Select an answer to reveal the explanation.

✓ Correct: Option C

IT's environmental footprint sits across six interconnected categories. Treating carbon as the only impact misses cooling and chip-fabrication water draw, the mineral intensity of devices (70+ elements in a smartphone), 50+ million tonnes of e-waste a year, the social conditions in mineral and electronics supply chains, and the localised land use and waste-heat impact of large sites.

A credible GreenOps programme tracks carbon and the other five. Narrow framing produces narrow action, and usually shifts impact rather than reduces it. Scope-based emissions accounting is one cut through this picture and is treated in depth in Module 3.

Knowledge Check · Module 1 · Q2

An organisation announces it has "moved to cloud and eliminated its data centre carbon footprint." What is the most accurate assessment of this claim?

Select an answer to reveal the explanation.

✓ Correct: Option C

Moving to cloud is one of the most widely held and consequential misconceptions in enterprise IT sustainability. Cloud moves IT emissions from Scope 1 and 2 into Scope 3: it does not eliminate them. Under CSRD and SBTi, Scope 3 is required where material, and for organisations using cloud at scale, it always is.

Cloud providers run on energy with its own carbon intensity, which varies significantly by region. A shared responsibility model applies: providers manage infrastructure efficiency; the IT organisation is responsible for how it uses that infrastructure and for engaging providers on sustainability performance.

Knowledge Check · Module 1 · Q3

In the GreenOps action model, which level involves preventing unnecessary demand from materialising in the first place?

Select an answer to reveal the explanation.

✓ Correct: Option C

Avoidance is the most mature and impactful level. Rather than measuring waste (Reporting) or reducing it after the fact (Optimisation), Avoidance prevents the demand from occurring: through portfolio rationalisation, shift-left architecture decisions, data minimisation policies, and choosing not to run heavyweight AI where it is not needed.

Most organisations are stuck at Reporting. The financial and sustainability case is to progress to Optimisation and then into Avoidance, to shift left in the lifecycle where intervention is cheapest and most effective.

⏸ Pause & Reflect

Take 5–10 minutes. Write answers down. Specificity matters more than completeness.

1Of the six categories: carbon and energy, water, materials and minerals, e-waste, social and supply chain, land and heat, which are most material in your organisation today? Which are not currently visible in any reporting or operational metric?
2Where does your organisation sit on the action model: Reporting, Optimisation, or Avoidance? Be honest. What single change would move it up one level inside the next twelve months?
3Pick one waste category from the five above. Who in your organisation owns it today? If the answer is "no one," that itself is the issue to surface.

Open discussion question: Where is the biggest single source of digital waste in your organisation right now, and who owns it?

Module 1: Key Takeaways

IT has six impact categories, not one.

Carbon and energy, water, materials and minerals, e-waste, social and supply chain, and land and heat. Treating carbon as the whole picture is the first form of narrow action.

The scale is bigger than most people assume.

ICT is 2–4% of global GHG, comparable in order of magnitude to aviation. In Ireland, data centres reached around 20% of national electricity. A single hyperscale site can use as much water as a small town.

Digital systems are physical systems.

Cloud does not remove infrastructure. It moves it. Wasted compute is wasted electricity, water, minerals, and manufacturing capacity. This framing carries through every later module.

Waste is the operational starting point.

Idle compute, storage bloat, zombie apps, fixed-cycle device refresh, and ungoverned AI inference are simultaneously cost problems and sustainability problems. They are where credibility is built fast.

Reporting is necessary, but not sufficient.

Reporting → Optimisation → Avoidance. Most organisations are still at Reporting and do not always know it. Honest self-placement is the precondition for moving up.

Avoidance is the highest-leverage move.

Designing demand out before it exists: portfolio rationalisation, shift-left architecture, data minimisation, model tiering. All of those beat optimising what should never have been built.

We now have the operational reality: what IT consumes, the order of magnitude, the mindset shift that digital is physical, the categories of waste that recur in every estate, and the three-level model that frames where action sits.

In Module 2 we define digital sustainability itself, separate footprint from handprint cleanly, and clear the three confusions (green IT, sustainable IT, and digital sustainability) that still derail otherwise serious conversations. From Module 3 onwards, the course turns to measurement, the estate, and the operational disciplines that follow from this foundation.

← Back to Course Hub