You might think your online life is weightless. But your clicks, streams, and uploads run on real machines and real power. Servers, cables, and data centers consume vast amounts of electricity and water right now.

Look at the numbers: the sector now produces emissions comparable to aviation and used about 460 TWh of electricity in 2022. Growth in devices and servers is steep, and energy needs rose roughly 70% between 2013 and 2020.

This section grounds the topic in facts. You’ll see how your everyday actions tie to manufacturing, transport, operation, and end-of-life systems. We’ll map devices, networks, and data centers so you can judge scale without hype.

Expect a clear, data-led guide that points to where reliable reports live and where more research is needed. By the end, you’ll understand why change matters for climate goals and practical steps toward sustainability.

Why your clicks aren’t carbon-free: framing the digital environmental impact today

Every click you make leans on machines, power plants, and cooling systems you never see. The sector now accounts for roughly 3.7–4% of global greenhouse gas emissions, and that share is rising as more devices and servers come online.

Data centers used about 460 TWh of electricity in 2022. Some researchers warn consumption could exceed ~1,050 TWh by 2026 if growth continues. North America alone nearly doubled data center capacity from 2,688 MW at end-2022 to 5,341 MW at end-2023.

That scale matters. Billions of user devices and tens of millions of servers turn your simple actions into real energy and water demand. Cooling, power quality, and local grid mix shape the carbon and water footprint you can’t see on your bill.

Key takeaways:

• Personal use scales up: small behaviors map to global systems and measurable emissions.
• Electricity and water are central inputs: choices in cooling and power affect communities and carbon loads.
• Data gaps exist: better information and independent research are vital to plan realistic reductions.

From footprint to trendline: what the data says about the digital environmental impact

The numbers and trends map where pressure is rising. Global shares show the sector accounts for roughly 3.7–4% of greenhouse gas emissions today. Trusted reports also flag rapid growth in electricity use and resource needs.

How big and where it comes from:

• Devices and manufacturing drive a large share of the footprint; for example, smartphone production can be about 75% of a phone’s total footprint.
• Networks move almost everything — 99% of intercontinental traffic uses subsea cables — while video streaming makes up roughly 54% of traffic and nearly 80% of bandwidth.
• Data centers consumed about 460 TWh in 2022 and may reach ~1,050 TWh by 2026 in some scenarios.

Present-day pressures: Electricity grids face rising load as new data centers and facilities come online. Water is a local constraint too; cooling needs strain supplies in some regions. Upstream mining and manufacturing shape pollution and resource footprints across supply chains.

What’s next: Rapid uptake of AI and larger training models pushes power density and processing demands higher. You’ll see infrastructure expand fast, and systems planners will need to balance efficiency gains with ever-growing consumption.

Devices first: manufacturing, materials, and e-waste you don’t see

The biggest environmental burden for phones and laptops happens before you ever turn them on. Manufacturing drives roughly 75% of a smartphone’s carbon footprint. Smartphones contain 70+ materials and around 50 metals, and many phones are replaced while still working—about 62% of them.

Laptops tell a similar story. They can emit 160–480 kg CO2e in manufacture and transport and require nearly 600 kg of materials. Terminals — your devices — account for a large share of sector emissions.

Embodied carbon and raw materials in phones and laptops

Extraction, component production, and shipping concentrate carbon, water, and pollution long before daily energy use kicks in. Recycling recovers only a fraction of rare metals, so first-life choices matter.

Why extending device lifespans and refurbishing matter

Repairing a battery, replacing a screen, or choosing a refurbished model reduces waste and cuts the need for new raw materials. Manufacturers that publish materials lists and repair policies make it easier for you to choose low-footprint options.

• Buy with support and parts availability in mind.
• Prefer repairs and trade-ins over early replacement.
• Push brands for transparency in materials and resource use.

The physical internet: undersea cables, networks, and the hidden miles your data travels

Subsea fiber links are the hidden highways that move almost all intercontinental data. Nearly 99% of cross‑border communications ride under the sea, not by satellite. That long haul adds real energy and consumption to every transfer.

Subsea cable infrastructure and its share of emissions

Cable systems carry tremendous volume, and their lifecycle matters. ADEME credits roughly 28% of network-related emissions to cable infrastructure when you count manufacture, laying ships, and landing stations.

Data-in-transit energy and the logistics of a “virtual” world

Data can travel thousands of kilometers; an email may average ~15,000 km to reach a hosted mailbox. Ships that lay and maintain cables use large amounts of fuel, and shore facilities draw power and water for operations.

• Route length and redundancy: longer paths and backup routes raise consumption.
• Traffic and latency: bandwidth demands and peering choices shape where packets run.
• Practical levers: smarter caching, fewer unnecessary transfers, and efficient content delivery cut transit footprint.

By tuning sync settings, upload quality, and network design, you can reduce the power and energy your use requires and help align networks with climate and environment goals.

Data centers and AI models: electricity, water, and hardware behind intelligence

Modern AI workloads push data centers into new regimes of power density, water use, and hardware churn. You’ll see how training and inference scale electricity and what that means for grids, water supplies, and supply chains.

Electricity demand and grid effects

Training big models pulls huge amounts of power. For example, training a GPT-3–scale model used an estimated 1,287 MWh and produced ~552 tons CO2. A ChatGPT query can use about five times the electricity of a simple search.

Water and cooling risks

Cooling can use roughly 2 liters of water per kWh. In water‑stressed areas, that converts to meaningful pressure on municipal supplies and local ecosystems.

Hardware, manufacturing, and scaling

GPUs and specialized racks add footprint before they compute. Shipments of GPUs hit ~3.85 million units in 2023, driving materials, logistics, and embodied emissions.

• Power density: AI training clusters can be 7–8× denser than average workloads.
• Grid stability: Operators still rely on diesel backups during spikes, raising short‑term emissions.
• Efficiency levers: better scheduling, heat reuse, and cleaner procurement cut overall energy consumption.

Streaming, search, and everyday use: how behavior translates to emissions

Small habits add up. Streaming in HD or 4K, leaving autoplay on, or sending many large attachments pushes demand along networks and into data centers. That extra load shows up as higher electricity use and higher emissions across the chain.

Video’s bandwidth dominance and its carbon consequences

Video accounts for roughly 54% of global internet traffic and nearly 80% of bandwidth use. Streaming is estimated to produce about 300 Mt CO2 per year worldwide.

Warum das wichtig ist: higher bitrates and resolution need more processing and more power from networks and racks. Large data centers can use as much electricity as a city of 50,000 residents.

Small habits, big numbers: queries, emails, and content efficiency

A single ChatGPT query can use about five times the electricity of a simple web search. In the UK, sending one fewer email per adult per day could avoid over 16,433 tons of CO2 a year.

• Choose lower resolution when you can.
• Turn off autoplay and trim automatic syncing.
• Design content with lighter images and fewer redirects to cut load.

Wegbringen: tweak settings and habits to shrink your personal carbon footprint. When companies make these defaults, small changes scale into meaningful climate benefits.

From problem to progress: data for the SDGs, digital governance, and practical sustainability

When you can measure resource use, you can manage it. Closing global data gaps — the UN notes 68% of environment-related SDG indicators lack enough data — is the first step to smarter policy and better development outcomes.

Closing gaps to guide action

Better global data and local measurement let you target reductions in energy and water use, track greenhouse gas emissions, and steer investments where they cut the environmental footprint fastest.

Governance and cooperation

Forums like IGF, ITU, UNEP, UNCTAD, and WMO align standards, reporting, and procurement. Initiatives such as the UN Roadmap for Digital Cooperation and ITU’s Green Digital Action push coordinated rules to scale change.

Business moves that work

Teams can adopt digital sobriety: streamline assets, reduce duplicate content, and right-size infrastructure. Use lifecycle metrics, climate-aligned SLAs, and water stewardship to make sustainability operational.

• Prioritize high-impact reductions.
• Invest in measurement and research to guide choices.
• Join transparency frameworks and data-sharing compacts to scale solutions.

For practical guidance and shared tools on data and the environment, see the data and the environment resource.

Abschluss

This report ends with clear, practical steps you can take to cut waste and emissions tied to your tech use.

You now know the biggest levers: manufacturing waste, streaming consumption, and fast model cycles. Focus on extending device life, reducing unnecessary uploads or high‑bitrate streams, and asking providers for transparent training and deployment data.

Small changes add up. Trim consumption, choose refurbished gear, and favor efficient content to lower your carbon footprint and curb gas emissions. Push teams to measure resource use and make it part of procurement and design.

Use this report as a checklist: prioritize high‑yield actions, demand better data, and align incentives so your choices reduce waste and strengthen climate change resilience.